What has Longley got against common sense?
David Longley
From the preface -"Irrationality: The Enemy Within"
S Sutherland 1992
"Pace Aristotle, it can be argued that irrational
behaviour is the norm not the exception. In order to
demonstrate this, I have provided many startling
examples of irrationality in everyday life and in the
activities of the pro- fessions. It turns out that the
decisions of doctors, generals, engineers, judges,
business men and others are no more rational than those
made by you or me though their effects are often more
calamitous.
However, the real proof of the prevalence of
irrationality comes from the massive amount of research
on the topic undertaken over the last thirty years by
psychologists. Their discoveries- unlike those of
cosmologists are as yet scarcely known to the general
public. Although I have not myself worked directly on
the topic, I became fascinated by the ingenuity of their
experiments and by the light they throw on the workings
of the mind. This book integrates the many factors that
have been shown to cause irrational behaviour, including
social and emotional biases as well as the many quirks
of thought pro- duced by such failings as not taking
account of negative cases or being too swayed by what
first comes to mind. Many of the experimental findings
are so surprising that the reader's credulity may well
be strained: almost all of them have, however, been
replicated many times. To stave off the sceptical
reader, there is a rather daunting list of sources,
which needs to be consulted only by those lacking faith
in my veracity or desiring to pursue specific issues in
more detail.
I have tried to make clear to the layman work that is
often hard to follow in the technical journals; for the
most part I have avoided mathematical and statistical
concepts, but of necessity a few elementary ones are
introduced and explained towards the end of the book."
To return to the theme of this thread - if indeed, empirical
psychology has lead to a revolution in the way professionals look
at natural human decision making processes, ie that psychologists
no longer expected normative, information theoretic, effective
computer models based on the predicate calculus (the norm up to
the mid 70s at least in memory and attention research) to work,
why should those working within the GOFAI tradition endeavour to
incorporate "common sense" knowledge into their systems. The
target here is the sort of programme which Hayes on the one hand,
and Lenat on the other had explicitly embarked upon - but the
implications run much wider. The reason why they might, I suggest
is simple - dogmatic preconceptions & ignorance of the literature.
The problem is that there is a tacit assumption or dogma in AI
research generally. That assumption is that there is something to
emulate, and if only this can be emulated, the objectives of AI
could be realised or at least put onto a realisable footing. It's
also an orthogonal dimension to the controversy between what some
see as existing between behaviourists vs. phenomenologists.
This - "something" - is, I suggest, chimerical. What we regard as
intelligent behaviour(s) are, I have suggested, a range of
behavioural fragments which are in themselves effective
procedures which have been learned. The actual processes which
support that learning also account for the acquisition of a much
larger set of quite superstitious and generally irrational
behaviours.
Whilst these heuristics may be adequate to account for the
diversity of human behaviour (far more diverse than most non-
specialist may appreciate), they are not much use as a basis for
fabrication of intelligent systems, since, apart from having to
be fabricated form the extensional stance in the first place,
the biases they are subject to are precisely the sorts of biases
which humans invest considerable efforts setting up
contingencies to select out (through education and formal rule
systems with sanctions).
I have illustrated the dynamic relationship between folk
psychology and the extensional stance in a concrete area of
public administration - that of "corrections".
http://www.longley.demon.co.uk/Frag.htm
And I have elaborated on points made in that paper (and others at
the same site), in a number of threads within this newsgroup over
the past three years.
My applied work is orientated towards bringing the extensional
stance to bear on the problems faced by those required to manage
effectively, & accountably. Those interested in this applied work
should contact me by e-mail or via one of the alternatives listed
on the website below.
--
David Longley
Longley Consulting London, UK
Behaviour Assessment & Profiling Technology,
Research, Data Analysis and Training Services,
Small IT Systems http://www.longley.demon.co.uk
David Longley
"In the 1970's, Cognitive Science was formulated in
terms of Putnam's doctrine of functionalism, in which
mental states are the functional states of an abstract
digital computer, thinking is abstract symbol
manipulation as in the operation of a computer program,
and the mind's symbols get their meaning by denoting
things in the world. Functionalism has since become the
mainstream doctrine within the philosophy of mind. But
in the mid-1970's, overwhelming empirical evidence
against functionalism began to pile up, and in the late
1970's, Putnam himself found functionalism to be
logically incoherent. In this volume, Putnam brilliantly
reveals the philosophical fallacies in the doctrine he
founded, showing why functionalism must fail as a
philosophy of mind. The fall of functionalism has major
consequences for generative linguistics, artificial
intelligence, and cognitive and developmental
psychology".
Lakoff (1988).
Which is a vindication of what Skinner long argued.
'Cognitive psychology is frequently presented as a
revolt against behaviorism, but it is not a revolt, it
is a retreat. Everyday English is full of terms
derived from ancient explanations of human behavior.
We spoke that language when we were young. When we
went out into the world and became psychologists, we
learned to speak in other ways but made mistakes for
which we were punished. But now we can relax.
Cognitive psychology is Old Home Week. We are back
among friends speaking the language we spoke when we
were growing up. We can talk about love and will and
ideas and memories and feelings and states of mind, and
no one will ask us what we mean; no one will raise an
eyebrow.'
('The Shame of American Education')
B.F. Skinner 1987
'Regardless of how much we stand to gain from supposing
that human behavior is the proper subject matter of a
science, no one who is a product of Western civilization
can do so without a struggle. We simply do not want such
a science.'
B F Skinner (1953)
Can Science Help? - The Threat to Freedom
(in Science and Human Behavior p.7)
http://www.longley.demon.co.uk/Frag.htm provides a sample of some
of the data and philosophical context which is representative of
the "overwhelming empirical evidence against functionalism" which
has accumulated over the past 30 years, and what this means
practically for both AI research and applied behavioural science.
--
David Longley (check end reply line #)
Neil Rickert
David Longley <Da...@longley.demon.co.uk> writes:
> From the preface -"Irrationality: The Enemy Within"
> S Sutherland 1992
> "Pace Aristotle, it can be argued that irrational
> behaviour is the norm not the exception.
Right. This can be argued. And Sutherland does argue it, aided by
considerable self delusion on his part.
> In order to
> demonstrate this, I have provided many startling
> examples of irrationality in everyday life and in the
> activities of the pro- fessions.
What Sutherland actually does, is take many examples of quite
rational behaviors, seriously misunderstand these behaviors in a way
that seems too common among psychlogists, and thus misconstrue them
as being irrational.
> It turns out that the
> decisions of doctors, generals, engineers, judges,
> business men and others are no more rational than those
> made by you or me though their effects are often more
> calamitous.
And anybody with just a little bit of common sense (or an iota of
familiarity with the base rates of performance by doctors, general,
engineers, judges) would recognize this as an absurd generalization
inconsistent with the evidence.
Of course we all know that Longley has no common sense, so it is no
surprise that he fails to see the absurdity.
>...
>why should those working within the GOFAI tradition endeavour to
>incorporate "common sense" knowledge into their systems.
Perhaps so that their systems will not make the same sort of
irrational judgments as has Sutherland in what was quoted above.
David Longley
Use the following examples to illustrate the point to yourself,
write down your answers as you read the questions.
A. Which do you consider to be more frequent, words having the
form 1) -----n- or words with the form 2) ----ing ?
B. Consider a group of 10 people. Make an intuitive estimate of
how many committees of 1) 2 people can be made from these people,
and 2) have many committees of 8 people can be made from these
people.
What we are seeing here as responses within this thread (and
probably off-line too) are quite classic responses where folk
respond with their natural beliefs, or pre-conceptions. But the
whole point of the research and the papers I have written is to
point out something which folk unfamiliar with these findings
generally do not appreciate.
There's no point bringing these preconceptions into the
discussion as counterarguments, as they are instances of the very
problem which prompted the research and my posts to this
newsgroup.
--
David Longley (check end reply line #)
Longley Consulting London, UK
Neil Rickert
Da...@longley.demon.co.uk (David Longley) writes:
>Use the following examples to illustrate the point to yourself,
>write down your answers as you read the questions.
>A. Which do you consider to be more frequent, words having the
>form 1) -----n- or words with the form 2) ----ing ?
Obviously (1).
>B. Consider a group of 10 people. Make an intuitive estimate of
>how many committees of 1) 2 people can be made from these people,
>and 2) have many committees of 8 people can be made from these
>people.
Obviously, both questions have the same answer (45).
But these are not real questions. They are academic questions with
little relevance to everyday life. That many people get these wrong
is of little consequence.
In everyday life, the number of committees you can form with 8 people
is considerably smaller than the number of committees you can form
with 2 people. For example, with 8 people it is very hard to find a
meeting time that fits all of their schedules, and it is very likely
that there will be disruptive personality clashes between some of the
people.
When this question is asked to somebody in a psychology experiment,
there may be no way that the experimenter can judge whether the
respondent is giving a rational response to an everyday life
interpretation of the question or a wrong answer to the ivory tower
academic interpretation of the question.
It is understandable that a mathematician might only consider the
ivory tower academic interpretation of a question of this type. But
a psychologist ought to understand that such questions are likely to
be seen in the context of real life experience. Thus it can be very
difficult to determine what problem the experimental subject was
actually solving. In my opinion, it is inept psychology to judge the
rationality of responses to questions of this type.
>What we are seeing here as responses within this thread (and
>probably off-line too) are quite classic responses where folk
>respond with their natural beliefs, or pre-conceptions.
We are certainly seeing a lot of Longley responding with his
unnatural beliefs and his preconceptions.
David Longley
Hopeless - you've missed the point ...yet again.
The empirical FACTS are that most people spontaneously respond
(2) for the first and think that more groups can be made of 2
individuals.
The bias has been replicated many times, and is an example of the
availability heuristic - the ease with which stereotypes are
recalled. The correct answers are arrived at through the
application of normative reasoning - the extensional stance -
something we teach....
But you it seems would like to think these skills are natural.
I can assure you, as should a lot of objective evidence besides -
including international cultural practices and investment - it is
*not* natural.
Once you understand this, you may begin to understand something
of what has been said - your above response makes it quite clear
that you have not (which to me is no surprise).
Neil Rickert
Da...@longley.demon.co.uk (David Longley) writes:
>Hopeless - you've missed the point ...yet again.
>The empirical FACTS are that most people spontaneously respond
>(2) for the first and think that more groups can be made of 2
>individuals.
And, as I explained, that answer for (2) is a better answer than the
one obtained by an application of combinatorics.
>The bias has been replicated many times, and is an example of the
>availability heuristic -
Yes, absolutely. But where is the bias?
I am saying that the bias is in the psychologist who designed the
experiment, and the heuristics are the method used by the confused
experimenter to try to explain the responses by the subjects.
The experimenter has no evidence of irrationality by the subjects.
To determine that the subjects responded irrationally would require
knowedge of:
What were the goals of the experimental subjects as they answered
the questions?
What was the complete set of input data used by the experimental
subjects?
Unless the experiment is able to read minds, he cannot answer either
of these questions.
Thus we see that the experimenter imputes to the subject a goal,
without any conclusive evidence that this is the subjects goal. And
the experimenter jumps to the conclusion that the subject is using as
input data what the experimenter thinks the subject should be using
as input data.
The only irrationality I see here is that of the experimenter in the
jumping to conclusions not warranted by the evidence.
> The correct answers are arrived at through the
>application of normative reasoning - the extensional stance -
>something we teach....
The correct answers to what problem? And based on what
information?
There really isn't any mystery to this.
Experimental design:
Design some ambiguous questions which are likely to be interpreted
differently by technical people than by non-technical people.
Get a group of experimental subjects who have little experience
with these kinds of questions, and who have very little reason for
a committment to a particular answer.
Judge the answers on the basis of how technical people would
interpret the question.
Publish in a psychology research journal.
The real question is why does Longley give so much importance to
this?
David Longley
There is nothing in what you say worth taking seriously. If you
bothered to read any of the material you would have realised that
these alternatives have all been controlled for.
The fact that you haven't given the researchers and those who
have reviewed the work over the past twenty years the credit for
having considered these possibilities can be taken as
illustrative of your arrogance and ignorance.
Rather than dashing of these ill-founded responses why don't you
spend some time studying the literature.....
Neil Rickert
Da...@longley.demon.co.uk (David Longley) writes:
>In article <6m458g$7...@ux.cs.niu.edu> ric...@cs.niu.edu "Neil Rickert" writes:
>> The experimenter has no evidence of irrationality by the subjects.
>> To determine that the subjects responded irrationally would require
>> knowedge of:
>> What were the goals of the experimental subjects as they answered
>> the questions?
>> What was the complete set of input data used by the experimental
>> subjects?
>> Unless the experiment is able to read minds, he cannot answer either
>> of these questions.
>> ...
>There is nothing in what you say worth taking seriously. If you
>bothered to read any of the material you would have realised that
>these alternatives have all been controlled for.
I guess we can take this as a clear and direct assertion by Longley
that psychologists can successfully and correctly read minds.
So much for Longley's supposed extensionalism.
David Longley
You clearly have no idea of the extent of this research - if you
did, you wouldn't have the audacity the post the material that
you do. But ignorance of what you're talking about has never
prevented you for sharing your views in the past, and I suspect
in your case, such exhibitionism is just one of your
idiosyncrasies. In this instance, past behaviour *is* alas,
likely to be the best predictor of future behaviour.
Still - as far as I'm concerned your posts serve some purpose -
if only as instantiations of the irrationality of human
"cognition" when it strays outside its limited domains of
expertise.
lsa...@aol.com
In article <898012...@longley.demon.co.uk>,
Da...@longley.demon.co.uk# wrote:
>[...]
> What counts in the final analysis is whether the case I make, and
> the practical technology which is derived from it, solves the
> problems it purports to and advances the illustrative field as I
> propose.
What really counts in the final analysis is whether the case
you make explains how to build a general driving robot. If it
does not, it has no relevance to AI or the philosophy of AI.
We have existence proof that human beings can learn to drive cars
reliably and our robots will mimic that learning capability
whether you like it or not. After all is said and done, your
"contribution" to AI and robot making will simply vanish into
oblivion.
Louis Savain
-----== Posted via Deja News, The Leader in Internet Discussion ==-----
http://www.dejanews.com/ Now offering spam-free web-based newsreading
David Longley
In article <3586FFBC...@sandpiper.net>
j...@sandpiper.net "Jim Balter" writes:
> So the correct inference is that, while T&K's views may not
> represent a cut and dried consensus of the views of mainstream
> psychology, they are the *correct* views, and any failure to consent
> is the result of error. The reasoning to support that inference
> has been given by Longley. To disagree with Longley, given the
> VAST evidence he has presented, is to commit sophistry:
>
You clearly don't know much if anything about the standard
research methodology within psychology. As I have said elsewhere,
there is a problem with null hypothesis statistical testing as a
method of testing. It is not a point prediction testing
methodology. One has to look at the overall picture - the weight
of evidence - and in this case you either have to take my word
for what I am asserting or do the same analysis of the literature
(not only form psychology but the other field I have covered too)
to see whether what I have said is, on balance, likely to be the
best appraisal or not.
Petty quibbling about issues which I have already considered and
discounted does not amount to falsification or serious evidence
against the point I am pressing. If you don't know what I mean by
this go and look again at the Quine-Duhem thesis.
> > > > But the material I present is more extensive than just the
> > > > examples cited - and I have spent as much time on the others. The
> > > > alternative hypotheses presented by others just aren't credible -
> > > > claiming they are is just sophistry.
>
> An idiosyncratic use of the word "sophistry" if ever there was one.
>
See above.
The case presented has clear practical implications and
applications. If you spend more time looking at the overall
picture you might appreciate what is being said somewhat more
than you currently do. But that would require some *real* work on
your part, and you clearly aren't prepared to put in the required
time and effort. So why bother responding at all?
David Longley
> In article <898012...@longley.demon.co.uk>,
> Da...@longley.demon.co.uk# wrote:
> >[...]
> > What counts in the final analysis is whether the case I make, and
> > the practical technology which is derived from it, solves the
> > problems it purports to and advances the illustrative field as I
> > propose.
>
> What really counts in the final analysis is whether the case
> you make explains how to build a general driving robot. If it
> does not, it has no relevance to AI or the philosophy of AI.
> We have existence proof that human beings can learn to drive cars
> reliably and our robots will mimic that learning capability
> whether you like it or not. After all is said and done, your
> "contribution" to AI and robot making will simply vanish into
> oblivion.
>
> Louis Savain
>
I have no specific interest in building 'robots' except to the
extent that thee take the form of rule based systems which help
us to manage our problem spaces more effectively than we are able
to do via our natural heuristics.
Bill Taylor
ric...@cs.niu.edu (Neil Rickert) writes:
|> >A. Which do you consider to be more frequent, words having the
|> >form 1) -----n- or words with the form 2) ----ing ?
|> >B. Consider a group of 10 people. Make an intuitive estimate of
|> >how many committees of 1) 2 people can be made from these people,
|> >and 2) have many committees of 8 people can be made from these people.
|> But these are not real questions. They are academic questions with
|> little relevance to everyday life. That many people get these wrong
|> is of little consequence.
Exactly! Good heavens, is this really an important phenomenon underlying
the Longley-bot religion? No wonder he/it won't come clean with a summary
of its basic tenets! If they're going to be as trivial as these. Phew!
|> In everyday life, the number of committees you can form with 8 people
|> is considerably smaller than the number of committees you can form
|> with 2 people. For example, with 8 people it is very hard to find a
|> meeting time that fits all of their schedules, and it is very likely
|> that there will be disruptive personality clashes
Good calls all. These things may well be on people's minds who answer these
things via common sense, rather than rigorous math.
Similarly, the wording of question A almost certainly means most plain
folk will interpret it as meaning
"which is more common: ----ing; or -----n- OTHER THAN ----ing".
Again a sensible interpretation - most ordinary folk (sensibly) do not
consider that reasonable people will ask questions of this form any
pernicketty intent. The man-in-the-street's charity is being turned
against him! He assumes the tester is being sensible, when he is not.
Sigh. We mathies have got a lot to answer for! ;-)
However, as Neil observes, this has nothing to do with the irrationality
of the common man, or of common sense. To insist it does, as does Longwind,
is a folly without warrant.
|> In my opinion, it is inept psychology to judge the
|> rationality of responses to questions of this type.
AND HOW!
|> We are certainly seeing a lot of Longley responding with his
|> unnatural beliefs and his preconceptions.
AND his irrationality! I sincerely hope they never let him near those
prisoners with his electric cattle prod!!
NWR summed it up perfectly (as usual) with this excerpt from his other post...
>>>>================
There really isn't any mystery to this.
Experimental design:
Design some ambiguous questions which are likely to be interpreted
differently by technical people than by non-technical people.
Get a group of experimental subjects who have little experience
with these kinds of questions, and who have very little reason for
a committment to a particular answer.
Judge the answers on the basis of how technical people would
interpret the question.
Publish in a psychology research journal.
================<<<<
,
Touche!
>The real question is why does Longley give so much importance to this?
One of the profound mysteries of the universe of c.ai.p.
Maybe it has to do with that rumour - can anyone shed any light on it?
It was suggested that maybe Longley is autistic. What do folk think?
-------------------------------------------------------------------------------
Bill Taylor W.Ta...@math.canterbury.ac.nz
-------------------------------------------------------------------------------
Faith is what you fall back on when you've lost the argument
-------------------------------------------------------------------------------
Betty Gladwin
Neil Rickert wrote:
> > "Pace Aristotle, it can be argued that irrational
> > behaviour is the norm not the exception.
> Right. This can be argued. And Sutherland does argue it, aided by
> considerable self delusion on his part.
Well, empirical research does show that automatic processing is an
important part of cognition. Being automatic, such processing doesn't
have to be rational, the whole term "rational" is meaningless on
that level. It takes a translation to a different level of
description to involve rationality in such cases.
> > It turns out that the
> > decisions of doctors, generals, engineers, judges,
> > business men and others are no more rational than those
> > made by you or me though their effects are often more
> > calamitous.
> And anybody with just a little bit of common sense (or an iota of
> familiarity with the base rates of performance by doctors, general,
> engineers, judges) would recognize this as an absurd generalization
> inconsistent with the evidence.
>
> Of course we all know that Longley has no common sense, so it is no
> surprise that he fails to see the absurdity.
Which, paradoxically, would prove his point in any case. Although
I wonder why you react this way to his statement that doctors etc
are no more rational than you or me - you think perhaps being a
doctor bestows some kind of exra rationality on the indiviudal
involved?
>
> >...
> >why should those working within the GOFAI tradition endeavour to
> >incorporate "common sense" knowledge into their systems.
> Perhaps so that their systems will not make the same sort of
> irrational judgments as has Sutherland in what was quoted above.
Which does support the idea that the risk is present.
Thomas
Neil Rickert
Da...@longley.demon.co.uk (David Longley) writes:
>You clearly don't know much if anything about the standard
>research methodology within psychology. As I have said elsewhere,
>there is a problem with null hypothesis statistical testing as a
>method of testing.
There is no problem with the proper use of null hypothesis testing.
There is a problem with the misuse of null hypothesis testing by
practitioners who do not understand the underlying mathematics.
Unfortunately, an inadequate understanding of mathematical statistics
is altogether too common in the social sciences.
Neil Rickert
>> > "Pace Aristotle, it can be argued that irrational
>> > behaviour is the norm not the exception.
>> Right. This can be argued. And Sutherland does argue it, aided by
>> considerable self delusion on his part.
>Well, empirical research does show that automatic processing is an
>important part of cognition. Being automatic, such processing doesn't
>have to be rational, the whole term "rational" is meaningless on
>that level. It takes a translation to a different level of
>description to involve rationality in such cases.
Longley had written:
>> > It turns out that the
>> > decisions of doctors, generals, engineers, judges,
>> > business men and others are no more rational than those
>> > made by you or me though their effects are often more
>> > calamitous.
I replied:
>> And anybody with just a little bit of common sense (or an iota of
>> familiarity with the base rates of performance by doctors, general,
>> engineers, judges) would recognize this as an absurd generalization
>> inconsistent with the evidence.
>> Of course we all know that Longley has no common sense, so it is no
>> surprise that he fails to see the absurdity.
>Which, paradoxically, would prove his point in any case.
No, it wouldn't. Longley is making generalizations. That his
general point may apply to one specific case (Longley himself) is no
basis for the generalization.
>I wonder why you react this way to his statement that doctors etc
>are no more rational than you or me - you think perhaps being a
>doctor bestows some kind of exra rationality on the indiviudal
>involved?
The statement was not that doctors, etc, are no more rational.
Rather, it was a statement about decision made by doctors, etc. I
took this to be a comment on the quality of their professional
decisions in their fields of expertise, rather than decisions they
make in other aspects of their lives. This must be understood in a
context of Longley's repeated claims that all judgements, including
those of doctors, are inferior to what can be done with an actuarial
table.
David Longley
In article <6m8n13$d...@ux.cs.niu.edu> ric...@cs.niu.edu "Neil Rickert" writes:
>
> The statement was not that doctors, etc, are no more rational.
> Rather, it was a statement about decision made by doctors, etc. I
> took this to be a comment on the quality of their professional
> decisions in their fields of expertise, rather than decisions they
> make in other aspects of their lives. This must be understood in a
> context of Longley's repeated claims that all judgements, including
> those of doctors, are inferior to what can be done with an actuarial
> table.
>
demonstrates - see the footnotes to Dawes, Faust and Meehl's
(1989) SCIENCE review, or the book by Dowie and Elstein
"Professional Judgement" or "Judgement and Decision Making" by
Arkes and Hammond. Psychologists may well undertake the research,
but the decision making is by professionals as a group of human
beings - not just psychologists (who are of course not immune to
the same problems as Dawes has reviewed in "House of Cards..".
Rather than publicly show your ignorance of the facts - why not
spend some time reading the relevant literature?
The point being made seems, to date, to be far too subtle for you
to grasp. If you read the literature you might appreciate that it
only *seems* to be my *personal* view because 1) *you* are so
unfamiliar with the extent of the research work and 2) have an
irrational tendency to argue with everything I say in principle
(regardless of the facts).
David Longley
There's nothing wrong with the proper use of logic and
mathematics either. But then that's the whole point is it not?
With respect to NHST, it's all too common everywhere that such
technology is used - a point you don't seem to grasp. It's a
technology which is used more in the social sciences than
anywhere else. The fact that it is not used more widely, and that
there is an alternative, *is* of course, central to the point
made in "Fragments"
But in this newsgroup such points seem to count for little. What
counts instead is nefarious rhetoric - a pastime which all too
many with nothing substantial to offer spend their time
generating for want of having anything genuinely constructive to
say. A pervasive and chronic intellectual disorder of
methodological solipsism.
Neil Rickert
Da...@longley.demon.co.uk (David Longley) writes:
>In article <6m8mk7$d...@ux.cs.niu.edu> ric...@cs.niu.edu "Neil Rickert" writes:
>> Da...@longley.demon.co.uk (David Longley) writes:
>> >You clearly don't know much if anything about the standard
>> >research methodology within psychology. As I have said elsewhere,
>> >there is a problem with null hypothesis statistical testing as a
>> >method of testing.
>> There is no problem with the proper use of null hypothesis testing.
>> There is a problem with the misuse of null hypothesis testing by
>> practitioners who do not understand the underlying mathematics.
>> Unfortunately, an inadequate understanding of mathematical statistics
>> is altogether too common in the social sciences.
>There's nothing wrong with the proper use of logic and
>mathematics either. But then that's the whole point is it not?
>With respect to NHST, it's all too common everywhere that such
>technology is used - a point you don't seem to grasp.
It is often used to good end. The amount of misuse in biological
areas seems considerably smaller than in the social sciences.
> It's a
>technology which is used more in the social sciences than
>anywhere else.
Properly used, it is a conservative methodology which is supposed to
ensure that you do not make claims without strong statistical
evidence. The trouble in the social sciences is that strongly
supportable claims are often hard to come by, and so researchers try
to support claims even when the evidence is marginal.
> The fact that it is not used more widely, and that
>there is an alternative, *is* of course, central to the point
>made in "Fragments"
And what is your alternative? Perhaps you prefer the application of
Bayesian methods starting with highly subjective prior
probabilities? Most mathematical statisticians consider
Neyman-Pearson methods to be sounder than Bayesian methods.
The fact is, any statistical methodology is subject to easy
manipulation, often resulting from excessive eagerness of the
experimenter rather than by any intention to deceive. Banning a
particular methodology won't solve this problem. More commonly the
problem is with the overall experimental design (which includes the
collection of data), rather than with the particular methodology for
handling the data that has been collected.
Neil Rickert
Da...@longley.demon.co.uk (David Longley) writes:
>In article <6m8n13$d...@ux.cs.niu.edu> ric...@cs.niu.edu "Neil Rickert" writes:
>> The statement was not that doctors, etc, are no more rational.
>> Rather, it was a statement about decision made by doctors, etc. I
>> took this to be a comment on the quality of their professional
>> decisions in their fields of expertise, rather than decisions they
>> make in other aspects of their lives. This must be understood in a
>> context of Longley's repeated claims that all judgements, including
>> those of doctors, are inferior to what can be done with an actuarial
>> table.
>Which is *exactly* what the research literature I have reviewed
>demonstrates - see the footnotes to Dawes, Faust and Meehl's
>(1989) SCIENCE review, or the book by Dowie and Elstein
>"Professional Judgement" or "Judgement and Decision Making" by
>Arkes and Hammond. Psychologists may well undertake the research,
>but the decision making is by professionals as a group of human
>beings - not just psychologists (who are of course not immune to
>the same problems as Dawes has reviewed in "House of Cards..".
Nobody doubts that even the best professionals make mistakes. And
nobody doubts that, once you know the correct answer, you can cook up
an actuarial table to give that answer. But only a hopelessly
confused psychologist such as Longley could then jump to the
irrational conclusion that we should always replace human judgement
by actuarial methods. If nothing else, the AI research on expert
systems has shown how difficult it is to surpass expert human
judgement.
>Rather than publicly show your ignorance of the facts - why not
>spend some time reading the relevant literature?
>The point being made seems, to date, to be far too subtle for you
>to grasp.
The point being made is that Longley is an irrational incompetent
excuse for a psychologist, who uses brow beating in an unsuccessful
attempt to conceal his own confusion.
The next time you have a tooth ache, go ask your actuarial tables to
fix it.
Jim Balter
David Longley wrote:
> But in this newsgroup such points seem to count for little. What
> counts instead is nefarious rhetoric [...]
I see that Longley has backed off from his previous stance that rhetoric
is "invariably" nefarious ("insincere", "argument for argument's sake"),
so perhaps the dlbot can be subtly influenced after all.
--
<J Q B>
lsa...@aol.com
In article <898049...@longley.demon.co.uk>,
Da...@longley.demon.co.uk# wrote:
[...]
>
> I have no specific interest in building 'robots' except to the
> extent that thee take the form of rule based systems which help
> us to manage our problem spaces more effectively than we are able
> to do via our natural heuristics.
It seems to me that your interests lie mainly in computer programming
and has little to do with what AI scientists want to do, which is to
build a general learning machine. Why do you insist that AI
scientists should *not* try to emulate the learning capability of
biological systems if your interests and expertise have little to do
with emulating that capability? What makes you think you can change
the goals of an entire scientific community? This smacks of arrogance
and selfishness.
Oliver Sparrow
Da...@longley.demon.co.uk (David Longley) wrote:
"" Which is *exactly* what the research literature I have reviewed
"" demonstrates
But is does not.
Quote from "Strengthening the Front Line, the challenge of knowledge-based
primary care" John Hoare ed Oct 1996 Oxford Health Care Mgmt Institute.
"There may be a danger at present of an all-consuming emphasis on the role of
evidence in medicine..... Debate on the issue is bound to continue; although
it is probably a fruitless argument about chalk and cheese. One view
emphasises the art and individual skills of care and medicine, the other
stresses the need to improve its base of knowledge. Through a patient's eyes
both are obviously necessary and are complementary. '
______________________________________
Oliver Sparrow
David Longley
> In article <898049...@longley.demon.co.uk>,
> Da...@longley.demon.co.uk# wrote:
>
> [...]
> >
> > I have no specific interest in building 'robots' except to the
> > extent that thee take the form of rule based systems which help
> > us to manage our problem spaces more effectively than we are able
> > to do via our natural heuristics.
>
> It seems to me that your interests lie mainly in computer programming
> and has little to do with what AI scientists want to do, which is to
> build a general learning machine. Why do you insist that AI
> scientists should *not* try to emulate the learning capability of
> biological systems if your interests and expertise have little to do
> with emulating that capability? What makes you think you can change
> the goals of an entire scientific community? This smacks of arrogance
> and selfishness.
>
> Louis Savain
What I'm saying is that some of the traditional goals of those
working (mainly in the GOFAI) tradition may be naive if, as I
say, natural "common sense" is not what they originally conceived
it to be. I am suggesting that research over the past 30 years
has changed the way that psychologists think of common sense,
many of those working in AI have not. There is no rational basis
for this other than academic isolationism.
The work in ANNs is no alternative. Such work has been undertaken
by psychologists for decades too, but in the guise of learning
theory. There too, the message is that such structures can learn
just about anything - not just functionally approximate the
desired procedures which one may wish them to learn in pursuit of
intelligent behaviour - which is turning out to be a bit of a
holy grail in a somewhat "pythonesque" sense.
If, as I say, intelligent behaviour amounts to (poorly understood
and therefore revered) effective procedures, it is the province
of science and technology generally to develop this. If that is
the case, it would explain why AI 'discoveries' (which I am not
disputing) are so quickly absorbed by what seem to be other areas
of technology.
However, I don't think the point I'm making *can* be made in just
a couple of paragraphs like this. I've tried to illustrate what I
see as *one* form of AI in the applied work I have listed. It is
only supposed to be illustrative.
David Longley
In article <358dbdcf...@news.demon.co.uk>
oh...@chatham.demon.co.uk "Oliver Sparrow" writes:
I know precisely what I am referring to - I suggest you may be
referring to something quite different.
I suggest you read the SCIENCE review by Dawes, Faust and Meehl
1989 before continuing this line. What I am referring to is
explicitly about diagnostic judgement and the relative efficacy
of actuarial vs. clinical.
Don't generalise that to more vague issues to do with caring. I
once asked a group of medical students how they thought they
manifested care for patients. The standard response was, as one
would expect, psychological talk about listening and being
sensitive to their anxieties etc. But all such talk is empty if
they don't invest time and effort in getting their rule based
diagnostic skills learned, which means requires intensive study
of a host of disciplines, and, in many respects, a working
knowledge of probability theory and basic statistics.
David Longley
In article <898162...@longley.demon.co.uk>
Da...@longley.demon.co.uk# "David Longley" writes:
> In article <358dbdcf...@news.demon.co.uk>
> oh...@chatham.demon.co.uk "Oliver Sparrow" writes:
>
> > Da...@longley.demon.co.uk (David Longley) wrote:
> >
> > "" Which is *exactly* what the research literature I have reviewed
> > "" demonstrates
> >
> > But is does not.
> >
> > Quote from "Strengthening the Front Line, the challenge of knowledge-based
> > primary care" John Hoare ed Oct 1996 Oxford Health Care Mgmt Institute.
> >
> > "There may be a danger at present of an all-consuming emphasis on the role of
> > evidence in medicine..... Debate on the issue is bound to continue; although
> > it is probably a fruitless argument about chalk and cheese. One view
> > emphasises the art and individual skills of care and medicine, the other
> > stresses the need to improve its base of knowledge. Through a patient's eyes
> > both are obviously necessary and are complementary. '
> > ______________________________________
> >
> > Oliver Sparrow
> >
> I know precisely what I am referring to - I suggest you may be
> referring to something quite different.
This may be an an opportune moment to remind folk of the Ariadne
Thread which runs through "Fragments" (the 12 volume "System
Specification for Profiling Behaviour" (1994) has two motifs on
each cover, Leibniz's Law and a regression formula). Throughout,
the guiding principle is *relation* between the failure of
substitution of identicals within contexts of propositional
attitude but the critical nature of this in science. Natural,
epistemic contexts are problematic because one can not reliably
quantify in, or substitute what seems to be identicals. This,
fundamentally Leibnizian principle, much worked upon by Frege in
"Sense and Reference" (Intension and Extension) and then by Quine
(who I draw upon extensively) should encourage one look carefully
at the behaviourist alternative I am advocating. Instead of
meaning and its problems, one should think of variants of
behaviour, "fragments of behaviour" and their extensional
analysis.
To anyone who has read http://www.longley.demon.co.uk/Frag.htm
the above will probably all make sense. To those who have no
knowledge of Leibniz's Law, the nature of 'mental' sentences and
their logical anomalies, and what I mean by enlightened
empiricism (Quine 1951) - it may well make sense, if, as I have
said many times, you read the material I have referenced (or look
through past elaborations I have provided in this newsgroup over
the past 3 years).
lsa...@aol.com
In article <898162...@longley.demon.co.uk>,
Well I know of several psychologists who disagree with you
concerning common sense. Their work shows that, contrary to
the outmoded litterature that you seem to prefer, human causal
induction is in fact rational and statistical in nature.
> The work in ANNs is no alternative. Such work has been undertaken
> by psychologists for decades too, but in the guise of learning
> theory. There too, the message is that such structures can learn
> just about anything - not just functionally approximate the
> desired procedures which one may wish them to learn in pursuit of
> intelligent behaviour - which is turning out to be a bit of a
> holy grail in a somewhat "pythonesque" sense.
The work in ANN is the correct alternative because we have
existence proof that biological networks can learn reliable and
effective behavior.
> If, as I say, intelligent behaviour amounts to (poorly understood
> and therefore revered) effective procedures, it is the province
> of science and technology generally to develop this.
This is nonsense since I've taken pains to point out that human
beings do learn extremely complex and effective procedures on their
own, using trial and error induction. You refuse to consider this
because you are stuck with your pompous stance that unless someone
gets a formal education, that someone is stupid. I maintain that the
opposite is true. We are able get an education because we are in
fact rational and we have excellent heuristics for learning.
> If that is
> the case, it would explain why AI 'discoveries' (which I am not
> disputing) are so quickly absorbed by what seem to be other areas
> of technology.
None of the AI discoveries you are talking about have much to do with
creating systems with effective learning heuristics like those used
by humans and animals.
> However, I don't think the point I'm making *can* be made in just
> a couple of paragraphs like this. I've tried to illustrate what I
> see as *one* form of AI in the applied work I have listed. It is
> only supposed to be illustrative.
The point you are making has no relevance to AI or anything else
because it is based on the false premise that biological systems
are irrational and unreliable. The empirical record says otherwise.
Neil Rickert
Da...@longley.demon.co.uk (David Longley) writes:
>In article <358dbdcf...@news.demon.co.uk>
> oh...@chatham.demon.co.uk "Oliver Sparrow" writes:
>> Da...@longley.demon.co.uk (David Longley) wrote:
>> "" Which is *exactly* what the research literature I have reviewed
>> "" demonstrates
>> But is does not.
>> Quote from "Strengthening the Front Line, the challenge of knowledge-based
>> primary care" John Hoare ed Oct 1996 Oxford Health Care Mgmt Institute.
>> "There may be a danger at present of an all-consuming emphasis on the role of
>> ...
>I know precisely what I am referring to - I suggest you may be
>referring to something quite different.
But then perhaps you are talking only to yourself, completely unable
to communicate your valuable ideas (if you have any) and unable to
comprehend what others are saying.
Oliver Sparrow
Da...@longley.demon.co.uk (David Longley) wrote:
"" I suggest you read the SCIENCE review by Dawes, Faust and Meehl
"" 1989 before continuing this line.
With respect, I did, in 1989. It tells one about the importance of making use
of evidence, training and human resource development. I have no argument
about the efficacy of this. But - and it is a major 'but' - there are known
to be other factors which make up significant proportion of the vartiance
between practices and hospitals, once these are weighted for their
socioeconomic base. This dimension is predicted by responses to questionaires
which are connected with the very issues which you dismiss. The evidence is
there and it is silly - or worse - to dismiss it because it does not fit with
a mechanistic model. Essentially, the one thing that we have learned about
public policy in the past two decades is that simple prescriptions do not fit
complex situations: GOSPLAN does not work. Decision support tools are
helpful, but nothing more than helpful.
I write this hot from addressing a couple of hundred people at the 1998
Computing in Government event, where the entire issue under debate was how
such tools may be deployed. Your position is not supported by the
professionals.
______________________________________
Oliver Sparrow
David Longley
In article <6mba75$g...@ux.cs.niu.edu> ric...@cs.niu.edu "Neil Rickert" writes:
>
> But then perhaps you are talking only to yourself, completely unable
> to communicate your valuable ideas (if you have any) and unable to
> comprehend what others are saying.
>
And maybe not. You're quite clearly unable to read and understand
the published literature I have referred to. The literature on
decision making reviewed in the anthologies and reviews are drawn
mainly from medical practice.
I don't think matters can be made any clearer other than by
quoting from the papers themselves - and even that doesn't work
with you. If there is a failure of communication - I suggest it
isn't for want of *my* efforts. You might make more progress in
understanding by accepting in the first place what you tacitly
admit above - namely that at present you do not understand. And
the reason for that, I keep telling you, is because you have
something to learn.
You can lead a Rickert to evidence but you can't make it think.
David Longley
In article <358a37a...@news.demon.co.uk>
oh...@chatham.demon.co.uk "Oliver Sparrow" writes:
Leaving aside the dubious status of the first sentence (need I go
back through my archives?) - tell me Oliver - just how does
anyone arrive at the conclusion that these "other factors" have a
*significant* contribution? And what do you think you are doing
when they are "weighted for their socio-economic base"?
Perhaps you think this is all a matter of intuitive judgement?
Take some time out and wait for the penny to drop before replying
- and don't give me this nonsense about "the professionals"
disagreeing with me - if there's one thing that Civil Servants
traditionally resist it's the professional contribution of
experts. The system is premised on the indefensible notion that
an adminisitrator's "good sense" and natural intelligence is all
that is required to effectively manage in any area - despite
numerous official reports that this is not the case.
You should look to the scientific evidence as I keep advising -
Stop second guessing it and getting it wrong. What I am offering
is technology for specific areas of application, and like it or
not, the professionals are on my side, both in the literature &
in the business world.
If you have explicit evdidence to the contrary - and I emphasize
*specific* evidence that my position is not supported, you should
cite it - or withdraw it as nefarious rhetoric.
Neil Rickert
Da...@longley.demon.co.uk (David Longley) writes:
>In article <6mba75$g...@ux.cs.niu.edu> ric...@cs.niu.edu "Neil Rickert" writes:
>> But then perhaps you are talking only to yourself, completely unable
>> to communicate your valuable ideas (if you have any) and unable to
>> comprehend what others are saying.
>And maybe not. You're quite clearly unable to read and understand
>the published literature I have referred to.
No. Rather it is that I am unwilling to understand the literature in
the peculiarly irrational way that Longley wants me to understand
it.
>I don't think matters can be made any clearer other than by
>quoting from the papers themselves - and even that doesn't work
>with you. If there is a failure of communication - I suggest it
>isn't for want of *my* efforts.
There is something wrong here. You, Longley, have made no
significant effort to communicate. There is no evidence that you
even comprehend the word "communicate". A number of people have
asked you to communicate your ideas, and have even explained how
communication works. Yet you insist in your habit of repeatedly
dumping the same quoted text into the newsgroup. That may be a way
of having the authors of the quoted text communicate, albeit not a
very effective one. But it is not a way of having David Longley
communicate.
>You can lead a Rickert to evidence but you can't make it think.
Actually, Longley's problem is that you can lead a Rickert to
evidence, but you cannot prevent it from thinking for itself and from
questioning Longley's absurd interpretations of that evidence.
David Longley
> Da...@longley.demon.co.uk (David Longley) writes:
> >In article <6mba75$g...@ux.cs.niu.edu> ric...@cs.niu.edu "Neil Rickert" writes:>
> >> But then perhaps you are talking only to yourself, completely unable
> >> to communicate your valuable ideas (if you have any) and unable to
> >> comprehend what others are saying.
>
> >And maybe not. You're quite clearly unable to read and understand
> >the published literature I have referred to.
>
> No. Rather it is that I am unwilling to understand the literature in
> the peculiarly irrational way that Longley wants me to understand
> it.
Fine - it matters little to me what you understand and why.
>
> >I don't think matters can be made any clearer other than by
> >quoting from the papers themselves - and even that doesn't work
> >with you. If there is a failure of communication - I suggest it
> >isn't for want of *my* efforts.
>
> There is something wrong here. You, Longley, have made no
> significant effort to communicate. There is no evidence that you
> even comprehend the word "communicate". A number of people have
> asked you to communicate your ideas, and have even explained how
> communication works. Yet you insist in your habit of repeatedly
> dumping the same quoted text into the newsgroup. That may be a way
> of having the authors of the quoted text communicate, albeit not a
> very effective one. But it is not a way of having David Longley
> communicate.
But I'm not trying to communicate "ideas" - I'm asking those who
*are* intrigued to *read* some specific material I have provided.
If they can't understand the papers I have provided, there are
references to elaborate on what is said. If that does not
suffice, I advise them not to bother.
>
> >You can lead a Rickert to evidence but you can't make it think.
>
> Actually, Longley's problem is that you can lead a Rickert to
> evidence, but you cannot prevent it from thinking for itself and from
> questioning Longley's absurd interpretations of that evidence.
>
I have no wish to prevent you from "thinking for yourself"
(whatever that means), but if you want to understand what I have
written, I require you to read it and show that you have
understood what I have written.
You are free to disregard anything I have to say - and I think we
all know the extent to which you are inclined to do this.
However - here's a prediction. Sooner or later - you're going to
find yourself arguing my case <g>.
David Longley
> Well I know of several psychologists who disagree with you
> concerning common sense. Their work shows that, contrary to
> the outmoded litterature that you seem to prefer, human causal
> induction is in fact rational and statistical in nature.
>
The possibility that you can find someone who you think disagrees
with me doesn't surprise me, but whether or not the disagreement
is what you you think it is may be another matter. When research
is undertaken in psychology the authors do not necessarily
personally believe or not believe the hypotheses they are
testing, they just recognise them to be possible logical
alternatives which are worthy of test. The problem is, as I have
said elsewhere, the alternatives are not always logical
alternatives, and the empirical results do not always allow
decisions to be made.
You, unfortunately, are driven by a desire to falsify what I have
written - for reasons best known to yourself. If indeed you have
some alternative to provide which you consider to be at odds with
what I assert - why not demonstrate this?
> > The work in ANNs is no alternative. Such work has been undertaken
> > by psychologists for decades too, but in the guise of learning
> > theory. There too, the message is that such structures can learn
> > just about anything - not just functionally approximate the
> > desired procedures which one may wish them to learn in pursuit of
> > intelligent behaviour - which is turning out to be a bit of a
> > holy grail in a somewhat "pythonesque" sense.
>
> The work in ANN is the correct alternative because we have
> existence proof that biological networks can learn reliable and
> effective behavior.
>
We also have "existence proofs" that they can learn unreliable
and quite neurotic behaviours - and history shows this to have
been far more prevalent than you seem to want to acknowledge
(again for reasons best known to yourself - if not, as I have
said - accounted for by your own pre-occupations giving you an
unrepresentative conception of the range of natural human
behaviour).
> > If, as I say, intelligent behaviour amounts to (poorly understood
> > and therefore revered) effective procedures, it is the province
> > of science and technology generally to develop this.
>
> This is nonsense since I've taken pains to point out that human
> beings do learn extremely complex and effective procedures on their
> own, using trial and error induction. You refuse to consider this
> because you are stuck with your pompous stance that unless someone
> gets a formal education, that someone is stupid. I maintain that the
> opposite is true. We are able get an education because we are in
> fact rational and we have excellent heuristics for learning.
You misrepresent what I say as usual - which is why I keep
telling you that you need a formal education in this area. Human
beings can and do indeed learn extremely complex and effective
procedures under natural, often adventitious conditions of
reinforcement, and it is well known how these behaviours are then
reinforced and vicariously distributed. That does not vitiate any
of the points I have made.
As I say, you have a naive and inaccurate conception of what has
been written, and you have arrived at this probably by the same
process which provides your other distorted beliefs.
I usually get paid for giving this sort of training advice -
consider yourself lucky you're getting it for nothing.
>
> > If that is
> > the case, it would explain why AI 'discoveries' (which I am not
> > disputing) are so quickly absorbed by what seem to be other areas
> > of technology.
>
> None of the AI discoveries you are talking about have much to do with
> creating systems with effective learning heuristics like those used
> by humans and animals.
How do you know which ones I'm talking about? And how do you
think anyone creates these technologies? Like other folk in this
thread, you just don't think what you are saying through. How do
the researchers and developers instantiate any of their "ideas"?
How do they then *engineer*? How do they *publish*.....??
>
> > However, I don't think the point I'm making *can* be made in just
> > a couple of paragraphs like this. I've tried to illustrate what I
> > see as *one* form of AI in the applied work I have listed. It is
> > only supposed to be illustrative.
>
> The point you are making has no relevance to AI or anything else
> because it is based on the false premise that biological systems
> are irrational and unreliable. The empirical record says otherwise.
Go and do some reading rather than ranting.
David Longley
Here's a small contribution to the re-establishment of rational
exchange in this thread. Those who disagree with what I have said
should:
1) state my position.
2) substantiate that precis with verbatim quotation.
3) show where your disagreement is.
4) substantiate that with evidence, not mere speculation.
Seth Russell
David Longley wrote:
> But I'm not trying to communicate "ideas" - I'm asking
> those who
> *are* intrigued to *read* some specific material I have
> provided.
That is a lie.
--
Seth
See "Bozo's Conjecture" at
http://www.clickshop.com/ai/conjecture.htm
And then on to the AI Jump List ...
Omar Haneef
Folks:
This thread is EXTREMELY long, 120 posts when I entered it. And I only read the first and last 20. Most of it is just insults so a lot of skimming was involved. Let me try to clarify some points as best I can, and maybe it will help settle matters:
(1) "Folk psychology" is used to describe a number of things. Amongst,
the use of intensional terms like "belief" and "desire", which some
behaviourists (like Dennet) claim do not exist. His claim is that what
we perceive as beliefs and desires are the results of false ontology;
that when a mature science is in place, we will realise that these
things are incorrect terms and they will be replaced. Its a hefty
debate with a large literature. As opposed to that view, Fodor
(amongst others) claims that the terms are useful and that a mature
science will VINDICATE their use.
(2) The irrationality of common sense judgements has been well
documented by a number of psychologists -- hate them if you want to --
such as Nisbett & Ross, and Tversky & Kahneman. Their work has been
influential in philosophy, psychology and economics (where Tversky and
Kahneman were on short lists for Nobel Prize nominations). Anyway, the
upshot is their work HAS drastically changed a lot of thinking in
medicine and economics. To give a flavour of the sorts of results they
have, here is one great example: they point out that people do not use
Bayesian reasoning (the normative, statistical view) when computing
probabilities. Try this experiment:
1. 1/5 cars in a city is a yellow cab
2. Yellow cabs are 2x as likely to be in an accident
3. A person gets hit by a car
4. Q: What is the car likely to be, a yellow cab or some other car?
Most people (try it) will respond with yellow cab not accounting for
the fact that there are LESS of them, so the probability of a yellow
cab being involved is (based only on the information above) slightly
less than 2/5.
Turns out that doctors were NOT taking base rates into account when
signing people up for preventive surgery or test. For examply, if a
disease is rare, say 1 in 10000 people have it, then even a test that
is accurate 98% of the time is useless since it is more likely that
the test is false. (If you know Bayes Theoreom you can try it out for
yourself).
P(B/A) = P(A).P(B)
Neil Rickert
oha...@andromeda.rutgers.edu (Omar Haneef) writes:
>(2) The irrationality of common sense judgements has been well
>documented by a number of psychologists -- hate them if you want to --
>such as Nisbett & Ross, and Tversky & Kahneman.
I don't hate them. I simply reject the claim that their experiments
demonstrate irrationality.
> Their work has been
>influential in philosophy, psychology and economics (where Tversky and
>Kahneman were on short lists for Nobel Prize nominations). Anyway, the
There is a general view amongst us arrogant mathematicians that it is
trivially obvious that people don't always make the best decisions.
Some of us tend to the opinion that only in very weak sciences such
as psychology, would research on this be considered worthy of
publication in a professional journal. But Nobel nominations??? I
would be more inclined to question the rationality of those who
wanted to nominate.
>1. 1/5 cars in a city is a yellow cab
>2. Yellow cabs are 2x as likely to be in an accident
>3. A person gets hit by a car
>4. Q: What is the car likely to be, a yellow cab or some other car?
>Most people (try it) will respond with yellow cab not accounting for
>the fact that there are LESS of them, so the probability of a yellow
>cab being involved is (based only on the information above) slightly
>less than 2/5.
"Most people" are probably right. When we say that Yellow cabs are
2x as likely to be in an accident, normal usage would be that this
was accident rate per mile driven. Since cabs, on the average, are
driven many more miles than other cars, Bayesian reasoning would
support "Most people."
>Turns out that doctors were NOT taking base rates into account when
>signing people up for preventive surgery or test. For examply, if a
>disease is rare, say 1 in 10000 people have it, then even a test that
>is accurate 98% of the time is useless since it is more likely that
>the test is false. (If you know Bayes Theoreom you can try it out for
>yourself).
If the test errors are independent of whether a person is diseased,
then this does not obviously follow. If a disease is that rare, then
most doctors do not give tests unless they see other indicators which
suggested the possibility of that disease.
In any case, what basis is there for the claim that Bayesian
reasoning should be normative?
David Longley
> oha...@andromeda.rutgers.edu (Omar Haneef) writes:
>
> >(2) The irrationality of common sense judgements has been well
> >documented by a number of psychologists -- hate them if you want to --
> >such as Nisbett & Ross, and Tversky & Kahneman.
>
> I don't hate them. I simply reject the claim that their experiments
> demonstrate irrationality.
What the work illustrates is the modus operandi of heuristics
used under conditions of uncertainty. These are not normative.
>
> > Their work has been
> >influential in philosophy, psychology and economics (where Tversky and
> >Kahneman were on short lists for Nobel Prize nominations). Anyway, the
>
> There is a general view amongst us arrogant mathematicians that it is
> trivially obvious that people don't always make the best decisions.
It is not just that people don't always make the best decisions -
it's the the strategies they *do* use have a certain
predictability to them. The explication of the processes is the
very business of cognitive psychology. The models we now have are
designed to fit the empirical findings, down to the biases
characteristics of the heuristics.
> Some of us tend to the opinion that only in very weak sciences such
> as psychology, would research on this be considered worthy of
> publication in a professional journal. But Nobel nominations??? I
> would be more inclined to question the rationality of those who
> wanted to nominate.
If it is what people *do* it is the business of empirical
psychology to investigate it, and circumscribe the parameters of
their use. I have provided some of the context to this in
Fragments.
CASEY DURFEE S
Rickert:
: I don't hate them. I simply reject the claim that their experiments
: demonstrate irrationality.
More like they demonstrate the inability of prior models (such as
Keynes' or Lewis') to accurately predict what people actually do.
Whether that means their experiments demonstrate irrationality is
more a question of whether you think that what people actually do
is rational or not.
Rickert:
: In any case, what basis is there for the claim that Bayesian
: reasoning should be normative?
Durfee:
None. And I can't think of many decision theorists who have made such
a claim.
-Casey Durfee
David Longley
In article <3596b...@rcfnews.cs.umass.edu>
cdu...@el0.cs.umass.edu "CASEY DURFEE S" writes:
> Rickert:
> : I don't hate them. I simply reject the claim that their experiments
> : demonstrate irrationality.
>
> More like they demonstrate the inability of prior models (such as
> Keynes' or Lewis') to accurately predict what people actually do.
> Whether that means their experiments demonstrate irrationality is
> more a question of whether you think that what people actually do
> is rational or not.
What we see here is yet more examples of the way heuristics
operate under conditions of uncertainty.
We see more and more people pick up on what they think the work
is, often eclectically tacitly absorbing what have elsewhere been
explictly drawn conclusions - and yet because of the nature of
what amounts to implicit functional definition - assuming they
are offering a point of view on the issue...
The intensional contexts almost guarantee corruption of the
original whilst at the same time confering a sui generis air to
the posts. Because they still retain *something* or the original,
and because these serially reproduced offerings are so often
couched in the acceptable style of folk psychology, ohers feel
inclined to comment with *their* points of view and so on.
>
> Rickert:
> : In any case, what basis is there for the claim that Bayesian
> : reasoning should be normative?
>
>
> Durfee:
> None. And I can't think of many decision theorists who have made such
> a claim.
>
The T&K studies claim to illustrate that natural human reasoning
(under conditions of uncertainty) is *not* Bayesian. There were those
who asserted that it was.
Savage L The Foundations of Statistics John Wiley & Sons 1954
There is also work to illustrate that it is not extensional:
Tversky A & Kahneman D Extensional Versus Intuitive Reasoning: The
Conjunction Fallacy in Probability Judgment Psychological Review
v90(4) 1983
Kahneman D, Slovic P & Tversky A Judgment Under Uncertainty:
Heuristics and Biases Cambridge University Press 1982
Agnoli F & Krantz D. H. Suppressing Natural Heuristics by Formal
Instruction: The Case of the Conjunction Fallacy Cognitive Psychology
21, 515-550, 1989
The following is one of the 14 text extracts from "Fragments" (note,
the full system is essentialy a functional and technical specification
for a system for behaviour profiling - a relational database, 4GL, a
set of routines for an applied field of behaviour management.
'..the meaning of words are abstractions from the truth
conditions of sentences that contain them.'
W.V.O. Quine (1981)
The Five Milestones of Empiricism: Theories and Things p.69
If such a line is accepted, intensionalist practices may serve no
practical purpose other than to distract from more fruitful processes
of measuring, recording and contracting behaviour. That is,
intensional practices may serve no more than to limit, through poor
socialization and private record keeping, what could be learned from
extensional analysis of relations between classes of behaviours (e.g.
frequencies of problem behaviour, and the joint frequencies of these
classes with other classes of behaviour such as age, and index
offence. Intensional contexts can be identified as follows:
'Chisholm proposes three independently operating criteria for
Intentional sentences.
(1) A simple declarative sentence is Intentional if it uses
a substantival expression - a name or a description - in such
a way that neither the sentence nor its contradictory implies
either that there is or that there isn't anything to which
the substantival expression truly applies.
(2) Any noncompound sentence which contains a propositional
clause...is Intentional provided that neither the sentence
nor its contradictory implies either that the propositional
clause is true or that it is false.
(3) If A and B are two names or descriptions designating the
same thing or things, and sentence P differs from sentence Q
only in having A where Q has B, then sentences P and Q are
Intentional if the truth of one together with the truth that
A and B are co-designative does not imply the truth of the
other'
The going scheme of logic, the logic that both works and is
generally supposed to suffice for all scientific discourse
(and, some hold, all SIGNIFICANT discourse), is extensional.
That is, the logic is blind to intensional distinctions; the
intersubstitution of coextensive terms, regardless of their
intensions, does not affect the truth value (truth or
falsity) of the enclosing sentence. Moreover, the truth
value of a complex sentence is always a function of the truth
values of its component sentences.
The Intentionalist thesis of irreducibility is widely
accepted, in one form or another, and there are two main
reactions to the impasse: Behaviourism and Phenomenology. The
behaviourist argues that since the Intentional idioms cannot
be made to fit into the going framework of science, they must
be abandoned, and the phenomena they are purported to
describe are claimed to be chimerical.'
D. C. Dennett (1969)
Content and Consciousness p32.
The choice was clearly spelled out by Quine in 1960, but remains
poorly appreciated:
'One may accept the Brentano thesis as showing the
indispensability of intentional idioms and the importance of
an autonomous science of intention, or as showing the
baselessness of intentional idioms and the emptiness of a
science of intention. My attitude, unlike Brentano's, is the
second. To accept intentional usage at face value is, we saw,
to postulate translation relations as somehow objectively
valid though indeterminate in principle relative to the
totality of speech dispositions. Such postulation promises
little gain in scientific insight if there is no better
ground for it than that the supposed translation relations
are presupposed by the vernacular of semantics and
intention.'
W. V. O Quine
The Double Standard: Flight from Intension
Word and Object (1960), p218-221
The alternative, methodologically incompatible approach of evidential
behaviourism, is restricted to extensional, normative analysis and
management of behaviour, drawing on natural inmate-environment
interactions, i.e., behaviour with respect to day to day activities.
This is eliminativist with respect to intensions (properties,
meanings, senses or thoughts, Quine, 1960; 1990; 1992), not on the
grounds that they comprise a body of pre-scientific 'folk' theoretical
idioms (Stich 1983; Churchland 1989), but because such idioms violate
the basic axiom of valid inference, namely Leibniz's Law: for any
objects x and y, if x is identical to y, then if x has a certain
property F, so does y. Symbolically: (x)(y)[(x=y) (Fx Fy)]. This is
the indiscernibility of identicals upon which all inference is
premised. ("Things are the same as each other, of which one can be
substituted for the other without loss of truth" - [Eadam sunt, quorum
unum potest substitui alteri salva veritate].
'...it is useless to suggest, as some logicians have done,
that the variable x may take as its values intensions of some
sort. For if we admit intensions as possible values of our
variables, we must abandon the principle of the
indiscernibility of identicals, and then, because we have no
clear criterion of identity, we shall be unable to say what
we want to say about extensions.'
Problems of Intensionality
W. Kneale and M Kneale (1962)
The Development of Logic p.617
'The first-order predicate calculus is an extensional logic
in which Leibniz's Law is taken as an axiomatic principle.
Such a logic cannot admit 'intensional' or 'referentially
opaque' predicates whose defining characteristic is that they
flout that principle.'
U. T. Place (1987)
Skinner Re-Skinned P. 244
In B.F. Skinner Consensus and Controversy
Eds. S. Modgil & C. Modgil
'There is a counterpart in modern logic of the thesis of
irreducibility. The language of physical and biological
science is largely extensional. It can be formulated
(approximately) in the familiar predicate calculus. The
language of psychology, however, is intensional. For the
moment it is good enough to think of an intensional sentence
as one containing words for intentional attitudes such as
belief.
Roughly what the counterpart thesis means is that important
features of extensional, scientific language on which
inference depends are not present in intensional sentences.
In fact intensional words and sentences are precisely those
expressions in which certain key forms of logical inference
break down.'
R. J. Nelson (1992)
Naming and Reference p.40
Note, '..intensional words and sentences are precisely those
expressions in which certain key forms of logical inference break
down' and '..the language of psychology, however, is intensional'.
Whilst it is clearly the case that folk psychology is largely
concerned with properties, characteristics or qualities of
individuals, their beliefs, desires, thoughts, feelings etc., it is
also the case that this is now true of much of contemporary
professional psychology (Fodor 1980). However, it may also be true
that many contemporary psychologists are not aware of the full
implications and quandaries implied by the this stance (Stich 1980).
Whilst it has been persuasively argued (Quine 1951,1956) that
quantification into intensional contexts is indeterminate, leading
inevitably to 'indeterminacy of translation' (Quine 1960). Nelson
(1992), a one time IBM senior mathematician goes on to say:
'It is widely claimed today by philosophers of logic that
intensional sentences cannot be equivalently rephrased or
replaced by extensional sentences. Thus Brentano's thesis
reflected in linguistic terms asserts that psychology cannot
be framed in the extensional terminology of mathematics,
physics or biology'.
ibid p.42.
This point has not only been made by logicians. In fact it has been a
major, perhaps the major finding of research within Personality and
Social Psychology since the 1950s. Here is how Ross and Nisbett (1991)
put the matter:
'Finally, it should be noted that some commonplace
statistical failings help sustain the dispositional bias.
First, people are rather poor at detecting correlations of
the modest size that underlie traits (Chapman and Chapman
1967, 1969; Kunda and Nisbett 1986; Nisbett and Ross 1980).
Second, people have little appreciation of the relationship
of sample size to evidence quality. In particular, they have
little conception of the value of aggregated observations in
making accurate predictions about trait-related behavior
(Kahneman & Tversky 1973; Kunda & Nisbett 1986). The gaps in
people's statistical abilities create a vacuum that the
perceptual and cognitive biases rush in to fill.'
L. Ross and R. E. Nisbett (1991)
The Person and the Situation: Perspectives of Social
Psychology
and within Cognitive Psychology, Agnoli & Krantz, 1989:
'A basic principle of probability is the conjunction rule,
p(B) >= p(A&B). People violate this rule often, particularly
when judgements of probability are based on intensional
heuristics such as representativeness and availability.
Through other probabilistic rules are obeyed with increasing
frequency as people's levels of mathematical talent and
training increase, the conjunction rule generally does not
show such a correlation. We argue that this recalcitrance is
not due to inescapable "natural assessments"; rather, it
stems from the absence of generally useful problem-solving
designs that bring extensional principles to bear on this
class of problem. We predict that when helpful extensional
strategies are made available, they should compete well with
intensional heuristics. Two experiments were conducted, using
as subjects adult women with little mathematical background.
In Experiment I, brief training on concepts of algebra of
sets, with examples of their use in solving problems, reduced
conjunction-rule violations substantially, compared to a
control group. Evidence from similarity judgements suggested
that use of the representativeness heuristic was reduced by
the training....
...We conclude that such intensional heuristics can be
suppressed when alternative strategies are taught.
The development of formal thought does not culminate in
adolescence as Piaget (1928) held; rather, it depends on
education (Fong, Krantz, & Nisbett, 1986, Nisbett, Fong,
Lehmann & Cheng 1987) and may continue throughout adulthood.
Probabilistic reasoning has been an especially useful domain
in which to study the impact of training in adulthood on
formal thought. Probabilistic principles are cultural
inventions at most a few centuries old (Hacking 1975).....
Tversky and Kahneman (1983) focused on processes in which
people substitute intensional for extensional thinking. In
the latter mode, concepts are represented mentally in the
same way as sets, hence, rules of logic and probability are
followed in the main. By contrast, intensional thinking
represents concepts by prototypes, exemplars, or relations to
other concepts (Rosch, 1978, Smith & Medlin 1981). Processing
is affected strongly by imaginability of prototypes,
availability of exemplars, etc., and its results are not
constrained as strongly by logical relations. A prime example
is the representativeness heuristic (Kahneman & Tversky
1972), in which probability of a outcome is judged in terms
of the similarity of that outcome to a prototype.
Tversky and Kahneman (1983) drew far reaching conclusions
from the fact that, in most of their tests, the prevalence of
conjunction errors was not affected by statistical education.
They developed the concept of "natural assessment", a
computation that is 'routinely carried out as part of the
perception of events and the comprehension of messages......
even in the absence of a specific task set.' They defined a
"judgmental heuristic" as a 'strategy that relies on a
natural assessment to produce an estimation or a prediction.'
They compared such mechanisms to perceptual computations, and
cognitive errors to perceptual illusions. In their view,
people well trained in mathematics nonetheless perform
natural assessments automatically. The results of these
mental computations strongly influence probability judgement.
Therefore, statistics courses presumably affect probability
judgements, in problems such as "Linda," no more than
geometry courses affect geometric visual illusions, i.e.,
scarcely at all.
Agnoli & Krantz (1989)
Suppressing Natural Heuristics by Formal Instruction:
The Case of the Conjunction Fallacy [my emphasis]
Cognitive Psychology 21, 515-550 (1989)
David Longley
In article <899110...@longley.demon.co.uk>
Da...@longley.demon.co.uk# "David Longley" writes:
>
> 'One may accept the Brentano thesis as showing the
> indispensability of intentional idioms and the importance of
> an autonomous science of intention, or as showing the
> baselessness of intentional idioms and the emptiness of a
> science of intention. My attitude, unlike Brentano's, is the
> second. To accept intentional usage at face value is, we saw,
> to postulate translation relations as somehow objectively
> valid though indeterminate in principle relative to the
> totality of speech dispositions. Such postulation promises
> little gain in scientific insight if there is no better
> ground for it than that the supposed translation relations
> are presupposed by the vernacular of semantics and
> intention.'
>
> W. V. O Quine
> The Double Standard: Flight from Intension
> Word and Object (1960), p218-221
>
It may not be a palatable conclusion for some - but what I am
suggesting in "Fragments" and in much that I have posted over the
past few years to this newsgroup in an effort to clarify points
made in that project, is that to the extent that the implications
of the above are not understood, we can expect little improvement
in the prediction and control of behaviour (living or
artificial).
The intensional is an *impediment* to scientific advancement, but
it may make for more easy reading. That extensional analysis is
more difficult, both the learn and sustain, just makes the former
that much harder to replace. Intensional heuristics are readily
"available", are more easily reinforced, and are, probably, for
most, more socially desirable.
I think folk should bear this in mind when discussing philosophy
of AI.
CASEY DURFEE S
: > Rickert:
: > : I don't hate them. I simply reject the claim that their experiments
: > : demonstrate irrationality.
Durfee:
: > More like they demonstrate the inability of prior models (such as
: > Keynes' or Lewis') to accurately predict what people actually do.
: > Whether that means their experiments demonstrate irrationality is
: > more a question of whether you think that what people actually do
: > is rational or not.
Longley:
<a buncha high-falutin' talk>
: and because these serially reproduced offerings are so often
: couched in the acceptable style of folk psychology, ohers feel
: inclined to comment with *their* points of view and so on.
Durfee:
I was making an observation based on considerable study of Bayesian
decision theory. This is one of those rare cases where I actually
know what I'm talking about.
: > Rickert:
: > : In any case, what basis is there for the claim that Bayesian
: > : reasoning should be normative?
: > Durfee:
: > None. And I can't think of many decision theorists who have made such
: > a claim.
: >
Longley:
: The T&K studies claim to illustrate that natural human reasoning
: (under conditions of uncertainty) is *not* Bayesian. There were those
: who asserted that it was.
Durfee:
You might want to check a definition of the word 'normative'. There's
a big difference between saying, "Human reasoning is Bayesian"
(descriptive) and saying, "Human reasoning should be Bayesian" (normative).
Of course there were those who asserted that human reasoning is Bayesian.
It's hard to imagine why people would work on a decision model they didn't
think had any resemblance to real life. I still can't think of anybody
offhand who has made an argument for normative decision theory.
Kahneman and Tversky's work was an attempt to ratify Bayesian theory so
that it would account for what they had observed. They weren't merely
trying to refute the theory, but to refine it.
Longley:
: The following is one of the 14 text extracts from "Fragments" (note,
: the full system is essentialy a functional and technical specification
: for a system for behaviour profiling - a relational database, 4GL, a
: set of routines for an applied field of behaviour management.
<Bobbit>
Durfee:
The rest of your post had absolutely no relevance to what I was talking
about, but thanks for going out of your way to be mind-bogglingly
abstruse. Have you ever considered writing programming language
specifications?
-Casey Durfee
Omar Haneef
Neil Rickert (ric...@cs.niu.edu) wrote:
: oha...@andromeda.rutgers.edu (Omar Haneef) writes:
: >(2) The irrationality of common sense judgements has been well
: >documented by a number of psychologists -- hate them if you want to --
: >such as Nisbett & Ross, and Tversky & Kahneman.
: I don't hate them. I simply reject the claim that their experiments
: demonstrate irrationality.
I didn't mean to say you do hate them, I know some people
do. But surely you have to present an argument against their
opinions. There are arguments out there, some of which I buy quite
wholeheartedly. (There is that German statistician -- Gergesomething
or other -- who is brilliant at debunking T & K type base rate
stuff. But he had a great argument to go with it.)
: > Their work has been
: >influential in philosophy, psychology and economics (where Tversky and
: >Kahneman were on short lists for Nobel Prize nominations). Anyway, the
: There is a general view amongst us arrogant mathematicians that it is
: trivially obvious that people don't always make the best decisions.
: Some of us tend to the opinion that only in very weak sciences such
: as psychology, would research on this be considered worthy of
: publication in a professional journal. But Nobel nominations??? I
: would be more inclined to question the rationality of those who
: wanted to nominate.
I think what I find impressive about K & T's work is that they
don't simply say we are irrational, they go ahead and generate models
of what sort of irrationality people express, when they are likely to
express it, and what heuristic -- alternative to the normative one --
people do use. Maybe you're much smarter than me, but when I first
read this stuff I was totally floored, and I'm still excited by the
debate it opened up.
Base Rate Neglect is one. They also talk about the framing
effect, that we are risk seeking when it comes to gains but avoid risk
when it comes to losses, etc.. We calculate chages of state in terms
of difference etc. Take a look at this one:
1. Give your friend $20
2. Tell him he can keep it, or gamble $10 for a double or nothing coin toss.
and
1. Give your friend $10
2. Tell him he can either have another $10, or take a coin toss chance for $20
In terms of changes of states of wealth, the two questions are
identical, but people, it turns out, are overwhelmingly more likely to
risk in the second version. Clearly, they are not using the same
heuristic.
Or here is another one:
=======================
There are four cards. Each card has a number on one side and a
letter on the other. The cards show the following faces: A, B, 1,
2. You want to see if the cards follow this simple rule: "If there is
a vowel on one side of the card, the other will be even." Which cards, at minimum, do you have to turn to discover the rule?
Now most people -- perhaps not yourself, being so well trained
against your natural impulses -- will choose A and 2 (try it!). This
might be because they think the rule works both ways (its been pointed
out that, because of Grician implicature, the english if might imply
"if and only if") but in that case they ought to choose all of
them. In any case the correct answer is A and 1. A for obvious
reasons. B and 2 you don't check because regardless of what is on the
other side the rule is not violated and 1 because a non-vowel would
violate the rule.
Another kicker: try an experiment identical in form but one
that (as Gergenezzer or whatever would put it) occurs in an
environment closer to our evoluitionary one, and the effect
disappears.
Tell your friend she or he is a bouncer at a bar, and you want
to make sure that no one over 21 is drinking. You can see four people,
but some have theur drinks covered and some have their IDs
covered. You see: a beer, a coke, an age 16, and an age 25. Which
drinks and/or IDs do you have to check to make sure there is nothing
illegal occuring? The idea is ridiculously obvious and comes
instantly. Cosmides and Humphreys claim this is evidence that our
evolutionary modules are designed for specific tasks such as cheater
detection.
The literature is fascinating, at least to me.
: >1. 1/5 cars in a city is a yellow cab
: >2. Yellow cabs are 2x as likely to be in an accident
: >3. A person gets hit by a car
: >4. Q: What is the car likely to be, a yellow cab or some other car?
: >Most people (try it) will respond with yellow cab not accounting for
: >the fact that there are LESS of them, so the probability of a yellow
: >cab being involved is (based only on the information above) slightly
: >less than 2/5.
: "Most people" are probably right. When we say that Yellow cabs are
: 2x as likely to be in an accident, normal usage would be that this
: was accident rate per mile driven. Since cabs, on the average, are
: driven many more miles than other cars, Bayesian reasoning would
: support "Most people."
Surely you are nitpicking. The original experiment was far more
elegant than my hurried recollection, and I think they made it clear
what twice as likely meant.
: >Turns out that doctors were NOT taking base rates into account when
: >signing people up for preventive surgery or test. For examply, if a
: >disease is rare, say 1 in 10000 people have it, then even a test that
: >is accurate 98% of the time is useless since it is more likely that
: >the test is false. (If you know Bayes Theoreom you can try it out for
: >yourself).
: If the test errors are independent of whether a person is diseased,
: then this does not obviously follow. If a disease is that rare, then
: most doctors do not give tests unless they see other indicators which
: suggested the possibility of that disease.
No, it turns out that professional doctors actually were
ignoring base rates in the relevant sense and after this research
caught the public eye, doctors started getting trained differently. At
least, thats the story I'm told. I do not find it hard to believe that
doctors did ignore base rates untill they were told not to. That they
have to be trained to attend to base rates.
: In any case, what basis is there for the claim that Bayesian
: reasoning should be normative?
Yes, well this is a question worthy of bringing up. People do question
why Bayesian reasoning should be normative. I, for one, do not
understand the complaint, but have heard it often enough that I feel
they must be on to something. In the contexts given, though, isn't
Bayesian reasoning superioir to ignoring the base rates? The german
guy, Gergenezzer or something like that, claims that we evolved in an
evolutionary environment where base rates in the FORM given in
questions was not relevant. He ran experiments where people are asked
to picture the cabs, or claims that 20 out of 100 cars are cabs
etc. When he does this, the base rate neglect simply disappears! I was
equally floored by this. That simply telling people quantitites: 20
out of hundred v. probabilities 0.2 will eliminate base rate neglect.
But there are other cool effects, maybe you mathematicians can figure
out what the computation is behind it: (This from Ross and Nisbett)
when people are asked to choose amongst identical items, such as a row
of red socks, they overwhelmignly choose objects towards the right!
This piece of data is less impressive to me for exactly the same
reasons that it is more resistant to criticism: there is no model or
theory, just a nice piece of data, so who cares.
Ciao for now,
Omar
Neil Rickert
oha...@andromeda.rutgers.edu (Omar Haneef) writes:
>Neil Rickert (ric...@cs.niu.edu) wrote:
>: oha...@andromeda.rutgers.edu (Omar Haneef) writes:
>: >(2) The irrationality of common sense judgements has been well
>: >documented by a number of psychologists -- hate them if you want to --
>: >such as Nisbett & Ross, and Tversky & Kahneman.
>: I don't hate them. I simply reject the claim that their experiments
>: demonstrate irrationality.
> I didn't mean to say you do hate them, I know some people
>do. But surely you have to present an argument against their
>opinions. There are arguments out there, some of which I buy quite
>wholeheartedly. (There is that German statistician -- Gergesomething
>or other -- who is brilliant at debunking T & K type base rate
>stuff. But he had a great argument to go with it.)
I think you are referring to Gigerenzer.
According to common definitions of rationality, whether a particular
decision is rational depends on: the inputs, the knowledge, the
intended outcome. My objection is that T&K judge the rationality of
their experimental subjects on what T&K consider to be the inputs and
the intended outcome, and on T&Ks estimation of the subject's
knowledge. The experimental subjects might actually be solving a
very different problem from what T&K assume, and their actual
knowledge may be different from what is assumed.
>1. Give your friend $20
>2. Tell him he can keep it, or gamble $10 for a double or nothing coin toss.
>and
>1. Give your friend $10
>2. Tell him he can either have another $10, or take a coin toss chance for $20
These are two very different problems.
>In terms of changes of states of wealth, the two questions are
>identical, but people, it turns out, are overwhelmingly more likely to
>risk in the second version. Clearly, they are not using the same
>heuristic.
Sure. But you are assuming that people are treating this as a way of
changing wealth. That is roughly equivalent to saying that golf and
soccer are identical games, because in each the aim is to get a ball
into some sort of hole.
In fact the problem, as you stated it, is a sort of game which exists
in a social setting. Treating it as a question of change of wealth
ignores the social parameters.
> There are four cards. Each card has a number on one side and a
>letter on the other. The cards show the following faces: A, B, 1,
>2. You want to see if the cards follow this simple rule: "If there is
>a vowel on one side of the card, the other will be even." Which cards, at minimum, do you have to turn to discover the rule?
> Now most people -- perhaps not yourself, being so well trained
>against your natural impulses -- will choose A and 2 (try it!).
Nor surprising. This is not a problem that people typically need to
solve. So there is no reason to suggest that they have direct
knowledge on the best way to tackle it. Thinking about such a
problem in terms of combinatorial logic is familiar enough to
mathematicians and logicians but not to most people. For most
people, a problem like this stresses the limitations of short term
memory.
> Another kicker: try an experiment identical in form but one
>that (as Gergenezzer or whatever would put it) occurs in an
>environment closer to our evoluitionary one, and the effect
>disappears.
Right. Because, in a more familiar setting they have experience that
they can call upon, and thus the demands on short term memory are
considerably reduced.
> Tell your friend she or he is a bouncer at a bar, and you want
>to make sure that no one over 21 is drinking. You can see four people,
>but some have theur drinks covered and some have their IDs
>covered. You see: a beer, a coke, an age 16, and an age 25. Which
>drinks and/or IDs do you have to check to make sure there is nothing
>illegal occuring? The idea is ridiculously obvious and comes
>instantly. Cosmides and Humphreys claim this is evidence that our
>evolutionary modules are designed for specific tasks such as cheater
>detection.
My explanation is that our knowledge is not in the form of
propositions (or beliefs), but is more in the form of abilities to
solve specific problems. The bouncer problem is closer to
experience, and thus something for which a person might have specific
abilities. The card problem is more abstract, and most people have
little experience and thus little ability in dealing with abstract
problems.
>: >Turns out that doctors were NOT taking base rates into account when
>: >signing people up for preventive surgery or test. For examply, if a
>: >disease is rare, say 1 in 10000 people have it, then even a test that
>: >is accurate 98% of the time is useless since it is more likely that
>: >the test is false. (If you know Bayes Theoreom you can try it out for
>: >yourself).
>: If the test errors are independent of whether a person is diseased,
>: then this does not obviously follow. If a disease is that rare, then
>: most doctors do not give tests unless they see other indicators which
>: suggested the possibility of that disease.
> No, it turns out that professional doctors actually were
>ignoring base rates in the relevant sense and after this research
>caught the public eye, doctors started getting trained differently.
You still have to ask what problem were the doctors solving. They
may be solving a problem of dealing with anxiety by a patient who is
worried about having a particular disease. They might be solving the
problem of minimizing their exposure to malpractice law suits. The
publication of research on base rates gives doctors evidence that
they can use to defend themselves in law suits, and gives information
they can give to anxious patients as an alternative way to calm them
down. In other words, the publication changes the doctor's
knowledge, and changes the nature of the problem that the doctor is
solving. At best, the argument for irrationality is that if you
ignore all of these changes then the behavior of doctors seems
irrational.
>: In any case, what basis is there for the claim that Bayesian
>: reasoning should be normative?
>Yes, well this is a question worthy of bringing up. People do question
>why Bayesian reasoning should be normative. I, for one, do not
>understand the complaint, but have heard it often enough that I feel
>they must be on to something. In the contexts given, though, isn't
>Bayesian reasoning superioir to ignoring the base rates?
Here is the problem. If I give you a bucket full of bits, and ask
you to apply Bayesian reasoning, you are not going to come up with
much. In fact, you probably won't even know how to start. However,
if these bits have been structured in some way, then you can begin to
construct some statistical information about the structures.
Bayesian reasoning, or indeed other kinds of statistical reasoning,
are sensitive to the way in which the data is structured. Since we
have no way of knowing how a particular person's brain has structured
that person's experiential data, we cannot tell which particular
Bayesian model we might reasonably expect a particular person to be
using.
David Longley
In article <3597a...@rcfnews.cs.umass.edu>
cdu...@el0.cs.umass.edu "CASEY DURFEE S" writes:
>
> : > Rickert:
> : > : I don't hate them. I simply reject the claim that their experiments
> : > : demonstrate irrationality.
>
> Durfee:
> : > More like they demonstrate the inability of prior models (such as
> : > Keynes' or Lewis') to accurately predict what people actually do.
> : > Whether that means their experiments demonstrate irrationality is
> : > more a question of whether you think that what people actually do
> : > is rational or not.
>
> Longley:
> <a buncha high-falutin' talk>
>
> : and because these serially reproduced offerings are so often
> : couched in the acceptable style of folk psychology, ohers feel
> : inclined to comment with *their* points of view and so on.
>
> Durfee:
> I was making an observation based on considerable study of Bayesian
> decision theory. This is one of those rare cases where I actually
> know what I'm talking about.
>
> : > Rickert:
> : > : In any case, what basis is there for the claim that Bayesian
> : > : reasoning should be normative?
>
> : > Durfee:
> : > None. And I can't think of many decision theorists who have made such
> : > a claim.
> : >
>
> Longley:
> : The T&K studies claim to illustrate that natural human reasoning
> : (under conditions of uncertainty) is *not* Bayesian. There were those
> : who asserted that it was.
>
> Durfee:
> You might want to check a definition of the word 'normative'. There's
> a big difference between saying, "Human reasoning is Bayesian"
> (descriptive) and saying, "Human reasoning should be Bayesian" (normative).
> Of course there were those who asserted that human reasoning is Bayesian.
> It's hard to imagine why people would work on a decision model they didn't
> think had any resemblance to real life. I still can't think of anybody
> offhand who has made an argument for normative decision theory.
>
> Kahneman and Tversky's work was an attempt to ratify Bayesian theory so
> that it would account for what they had observed. They weren't merely
> trying to refute the theory, but to refine it.
>
> Longley:
> : The following is one of the 14 text extracts from "Fragments" (note,
> : the full system is essentialy a functional and technical specification
> : for a system for behaviour profiling - a relational database, 4GL, a
> : set of routines for an applied field of behaviour management.
> <Bobbit>
>
> Durfee:
> The rest of your post had absolutely no relevance to what I was talking
> about, but thanks for going out of your way to be mind-bogglingly
> abstruse. Have you ever considered writing programming language
> specifications?
>
> -Casey Durfee
>
I have three points to make in response to this post.
1. I take it to be an example of the muddle typical of many
posting to this group who either do not have the professional
training which tends to be a pre-requisite for a full
understanding of the significance of the work I have cited, or
are so self-opinionated that they do not know when to listen
carefully when someone brings the implications to their
attention.
2. If you ignore what is unfamiliar, particularly when the
author claims to have something rather subtle to say, you are
unlikely to be shown much charity when you reveal your ignorance
and assert that your minimal understanding of what has been said
*is* what has been said.
3. It takes some time for post graduate applied psychologists to
appreciate the practical implications of the points being made in
"Fragments". Those with little or no professional training in
behavioural science would do well to remember that.
I offer the following extracts for those who are prepared to read
and learn what *has* been said, rather than for those who are
quite content with their *belief* that they understand what has
been said. For those who *are* prepared to read carefully, I
recommend, once again that they read, and re-read the material
provided in "Fragments of Behaviour: The Extensional Stance"
which is available in different formats at the website below, and
is available with some material drawn from a potential field of
application in:
http://www.longley.demon.co.uk/Frag.htm
o o o
"Normative models and descriptive heuristics
The view has been expressed (see, e.g., W. Edwards,
1968, 25) that man, by and large, follows the correct
Bayesian rule, but fails to appreciate the full impact
of evidence, and is therefore conservative. Peterson and
Beach (1967), for example, concluded that the normative
model provides a good first approximation to the
behavior of the Ss who are "influenced by appropriate
variables and in appropriate directions" (p. 43). This
view has not been shared by all. In a more recent review
of the literature, Slovic and Lichtenstein (1971) argued
that the above evaluation of man's performance as an
intuitive statistician is far "too generous," while
Pitz, Downing, and Reinhold (1967) concluded, on the
basis of their data, that human performance in Bayesian
tasks is "nonoptimal in a more fundamental way than is
implied by discussions of conservatism" (p. 392).
The usefulness of the normative Bayesian approach to
the analysis and the modeling of subjective probability
depends primarily not on the accuracy of the subjective
estimates, but rather on whether the model captures the
essential determinants of the judgment process. The
research discussed in this paper suggests that it does
not. In particular, we have seen that sample size has no
effect on subjective sampling distributions, that
posterior binomial estimates are determined (in the
aggregate case, at least) by sample proportion rather
than by sample difference, and that they do not depend
on the population proportion. In his evaluation of
evidence, man is apparently not a conservative Bayesian:
he is not Bayesian at all.
It could be argued that the failure of the normative
model to describe human behavior is limited to naive Ss
faced with unfamiliar random processes, and that the
normative model could provide an adequate account of the
evaluation of the more familiar random processes that
people encounter in everyday life. There is very little
evidence, however, to support this view. First, it has
been shown (Tversky & Kahneman, 1971, 2) that the same
type of systematic errors that are suggested by
considerations of representativeness can be found in the
intuitive judgments of sophisticated scientists.
Apparently, acquaintance with the theory of probability
does not eliminate all erroneous intuitions concerning
the laws of chance. Second, in our daily life we
encounter numerous random processes (e.g., the birth of
a boy or a girl, hitting a red light at a given
intersection, getting a hand with no hearts in a card
game) which obey the binomial law, for example, to a
high degree of approximation. People, however, fail to
extract from these experiences an adequate conception of
the binomial process. Apparently, extensive exposure to
numerous examples alone does not produce optimal
behavior.
In their daily lives, people ask themselves and
others questions such as: What are the chances that this
12-year-old boy will grow up to be a scientist? What is
the probability that this candidate will be elected to
office? What is the likelihood that this company will go
out of business? These problems differ from those
discussed earlier in the paper in that, due to their
unique character, they cannot be readily answered either
in terms of frequency of occurrence in the past, or in
terms of some well-defined sampling process.
In this paper, we investigated in-some detail one
heuristic according to which the likelihood of an event
is evaluated by the degree to which it is representative
of the major characteristics of the process or
population from which it originated. Although our
experimental examples were confined to well-defined
sampling processes (where objective probability is
readily computable), we conjecture that the same
heuristic plays an important role in the evaluation of
uncertainty in essentially unique situations where no
"correct" answer is available. The likelihood that a
particular 12-year-old boy will become a scientist, for
example, may be evaluated by the degree to which the
role of a scientist is representative of our image of
the boy. Similarly, in thinking about the chances that a
company will go out of business, or that a politician
will be elected for office, we have in mind a model of
the company, or of the political situation, and we
evaluate as most likely those outcomes which best
represent the essential features of the corresponding
model...."
From:
A Tversky & D Kahneman
Subjective probability: A Judgment of representativeness
Cognitive Psychology, 1972,3 430-454
extract 2
"Discussion
This article has been concerned with cognitive biases
that stem from the reliance on judgmental heuristics.
These biases are not attributable to motivational
effects such as wishful thinking or the distortion of
judgments by payoffs and penalties. Indeed, several of
the severe errors of judgment reported earlier occurred
despite the fact that subjects were encouraged to be
accurate and were rewarded for the correct answers
(Kahneman & Tversky, 1972b, 3; Tversky & Kahneman,
1973,11).
The reliance on heuristics and the prevalence of
biases are not restricted to laymen. Experienced
researchers are also prone to the same biases - when
they think intuitively. For example, the tendency to
predict the outcome that best represents the data, with
insufficient regard for prior probability, has been
observed in the intuitive judgments of individuals who
have had extensive training in statistics (Kahneman &
Tversky, 1973, 4; Tversky & Kahneman, 1971, 2). Although
the statistically sophisticated avoid elementary errors,
such as the gambler's fallacy, their intuitive judgments
are liable to similar fallacies in more intricate and
less transparent problems.
It is not surprising that useful heuristics such as
representativeness and availability are retained, even
though they occasionally lead to errors in prediction or
estimation. What is perhaps surprising is the failure of
people to infer from lifelong experience such
fundamental statistical rules as regression toward the
mean, or the effect of sample size on sampling
variability. Although everyone is exposed, in the normal
course of life, to numerous examples from which these
rules could have been induced, very few people discover
the principles of sampling and regression on their own.
Statistical principles are not learned from everyday
experience because the relevant instances are not coded
appropriately. For example, people do not discover that
successive lines in a text differ more in average word
length than do successive pages, because they simply do
not attend to the average word length of individual
lines or pages. Thus, people do not learn the relation
between sample size and sampling variability, although
the data for such learning are abundant.
The lack of an appropriate code also explains why
people usually do not detect the biases in their
judgments of probability. A person could conceivably
learn whether his judgments are externally calibrated by
keeping a tally of the proportion of events that
actually occur among those to which he assigns the same
probability. However, it is not natural to group events
by their judged probability. In the absence of such
grouping it is impossible for an individual to discover,
for example, that only 50 percent of the predictions to
which he has assigned a probability of .9 or higher
actually come true.
The empirical analysis of cognitive biases has
implications for the theoretical and applied role of
judged probabilities. Modern decision theory (de
Finetti, 1968; Savage, 1954) regards subjective
probability as the quantified opinion of an idealized
person. Specifically, the subjective probability of a
given event is defined by the set of bets about this
event that such a person is willing to accept. An
internally consistent, or coherent, subjective
probability measure can be derived for an individual if
his choices among bets satisfy certain principles, that
is, the axioms of the theory. The derived probability is
subjective in the sense that different individuals are
allowed to have different probabilities for the same
event. The major contribution of this approach is that
it provides a rigorous subjective interpretation of
probability that is applicable to unique events and is
embedded in a general theory of rational decision.
It should perhaps be noted that, while subjective
probabilities can sometimes be inferred from preferences
among bets, they are normally not formed in this
fashion. A person bets on team A rather than on team B
because he believes that team A is more likely to win;
he does not infer this belief from his betting
preferences. Thus, in reality, subjective probabilities
determine preferences among bets and are not derived
from them, as in the axiomatic theory of rational
decision (Savage, 1954).
The inherently subjective nature of probability has
led many students to the belief that coherence, or
internal consistency, is the only valid criterion by
which judged probabilities should be evaluated. From the
standpoint of the formal theory of subjective
probability, any set of internally consistent
probability judgments is as good as any other. This
criterion is not entirely satisfactory, because an
internally consistent set of subjective probabilities
can be incompatible with other beliefs Aeld by the
individual. Consider a person whose subjective
probabilities for all possible outcomes of a coin-
tossing game reflect the gambler's fallacy. That is, his
estimate of the probability of tails on a particular
toss increases with the number of consecutive heads that
preceded that toss. The judgments of such a person could
be internally consistent and therefore acceptable as
adequate subjective probabilities according to the
criterion of the formal theory. these probabilities,
however, are incompatible with the generally held belief
that a coin has no memory and is therefore incapable of
generating sequential dependencies. For judged
probabilities to be considered adequate or rational,
internal consistency is not enough. The judgments must
be compatible with the entire web of beliefs held by the
individual. Unfortunately, there can be no simple formal
procedure for assessing the compatibility of a set of
probability judgments with the judge's total system of
beliefs. The rational judge will nevertheless strive for
compatibility, even though internal consistency is more
easily achieved and assessed. In particular, he will
attempt to make his probability judgments compatible
with his knowledge about the subject matter, the laws of
probability, and his own judgmental heuristics and
biases.
Summary
This article described three heuristics that are
employed in making judgments under uncertainty: (i)
representativeness, which is usually employed when
people are asked to judge the probability that an object
or event A belongs to class or process B; (ii)
availability of instances or scenarios, which is often
employed when people are asked to assess the frequency
of a class or the plausibility of a particular
development; and (iii) adjustment from an anchor, which
is usually employed in numerical prediction when a
relevant value is available. These heuristics are highly
economical and usually effective, but they lead to
systematic and predictable errors. A better
understanding of these heuristics and of the biases to
which they lead could improve judgments and decisions in
situations of uncertainty."
David Longley
In article <6n8nro$1un$1...@andromeda.rutgers.edu>
oha...@andromeda.rutgers.edu "Omar Haneef" writes:
>
> I didn't mean to say you do hate them, I know some people
> do. But surely you have to present an argument against their
> opinions. There are arguments out there, some of which I buy quite
> wholeheartedly. (There is that German statistician -- Gergesomething
> or other -- who is brilliant at debunking T & K type base rate
> stuff. But he had a great argument to go with it.)
It was Gerd Gigerenzer - perhaps you'd like to outline his case.
I for one don't find it all that persuasive and think it
*fundamentally* misses the point...
(References provided in "Fragments")
CASEY DURFEE S
Longley:
: 1. I take it to be an example of the muddle typical of many
: training which tends to be a pre-requisite for a full
: understanding of the significance of the work I have cited, or
: are so self-opinionated that they do not know when to listen
: carefully when someone brings the implications to their
: attention.
Durfee:
Before responding to my post, please note that I merely
pointed out that there are few, if any, philosophers who have
posited a normative rather than a descriptive decision theory.
If you can come up with a cite of a counterexample, I would appreciate it.
If not, then please quit slandering me.
As for the rest of your post (which had nothing to do with what I
was talking about, by the way), I noticed a couple of factual errors
that you might want to straighten out before you go ranting on about
how ignorant people will not be shown charity.
: Finetti, 1968; Savage, 1954) regards subjective
: probability as the quantified opinion of an idealized
: person. Specifically, the subjective probability of a
Subjective probability does not have to be the opinion of an
idealized person. Beliefs merely have to be consistent; they
don't have to be reasonable to conform to the axioms of decision
theory.
: individual. Unfortunately, there can be no simple formal
: procedure for assessing the compatibility of a set of
: probability judgments with the judge's total system of
: beliefs. The rational judge will nevertheless strive for
There is a simple formal procedure: If you can make a Dutch
Book against someone, their beliefs are inconsistent.
-Casey Durfee
David Longley
In article <35991...@rcfnews.cs.umass.edu>
cdu...@el0.cs.umass.edu "CASEY DURFEE S" writes:
> Longley:
> : 1. I take it to be an example of the muddle typical of many
> : posting to this group who either do not have the professional
> : training which tends to be a pre-requisite for a full
> : understanding of the significance of the work I have cited, or
> : are so self-opinionated that they do not know when to listen
> : carefully when someone brings the implications to their
> : attention.
>
> Durfee:
> Before responding to my post, please note that I merely
> pointed out that there are few, if any, philosophers who have
> posited a normative rather than a descriptive decision theory.
> If you can come up with a cite of a counterexample, I would appreciate it.
> If not, then please quit slandering me.
Prima facie this looks odd. Predicate logic is normative, so is
mathematics. Computer science is in large part a development of
that normative, Leibnizian, Boolean or Fregian line of work. Even
"The Laws of Thought" was a normative not descriptive work.
But as I have said elsewhere, on second thoughts, what you say is
grist to my mill, as I wrote "Fragments"
http://www.longley.demon.co.uk/Fragment.zip
and
http://www.longley.demon.co.uk/Frag.htm
mainly because so few of the people I was working with or
teaching teaching seemed aware of the research (from the
perspective I was presenting it).
Philosophers, mathematicians and computer scientists develop
normative models of decision making. Engineers use those models
as do most other scientists.
Psychologists develop descriptive models. They develop models of
conditioning, attribution, and explicate heuristics and their
biases. Historically they have modelled these with neural
networks (albeit in different guises).
Since most of my library is on decision theory and its history I
won't cite more examples here, suffice it to say that what isn't
explicitly referenced in "Fragments" in answer to the above
question is there in the references to the references.
I offer the following extracts in the hope that they will
interest you and clarify the point of my posts more generally:
'My conclusion from these experiments, and a hundred
more of which I have read, is that they CAN be taken, if
we want, as testing whether decision theory is true. But
it is at least as plausible to take them as testing how
good one or another criterion of preference is, on the
assumption that decision theory is true.
Amos Tversky, a tireless and skilled worker in this
area, seems to have come to much the same conclusion. He
tells of some experiments done by him and Daniel
Kahneman in Jerusalem. They asked subjects to choose
between A and B, and C and D, in the following
situation:
Choice I: A = (1000,1/2,0) B= (400)
Choice II: C = (1000,1/10,0) D= (400,1/5,0)
Here A, for example, is a wager that is equally apt
(subjectively) to pay $1,000 or nothing; B is simply
receiving $400 (no risk). Tversky says almost all
subjects picked B over A and C over D; the result held
for naive and educated subjects, and analogous results
were obtained for different payoffs and probabilities.
The trouble is, this pattern of preferences is
incompatible with decision theory.'
If length is not transitive, what does it mean to use a
number to measure a length at all? We could find or
invent an answer, but unless or until we do we must
strive to interpret 'longer than' so that it comes out
transitive. Similarly for 'preferred to'.
I think I have an argument to show that the main
empirical thrust of an explanation of an action in
decision theory, or of a reason explanation, does not
come from the axioms of decision theory, or 'the
assumption of rationality', but rather from the
attributes of desires, preferences, or beliefs.
D Davidson (1976)
Hempel on Explaining Action
'Davidson is very familiar with the work of Amos Tversky
(see Tversky & Kahneman, 1975, 1982). Thus he is
extremely aware that PREFERENCES cannot be read off from
even SINCERE VERBAL REPORTS. As Tversky's very careful
empirical research shows, people's sincere verbal
reports of their own preferences are totally incoherent.
If we acted on the maxim of ascribing to people all the
preferences they say (sincerely) they have then we would
be unable to interpret their behavior at all, for
EXPRESSED preferences are totally contradictory (e.g.,
they violate the logical property of the transitivity of
preference very badly).
Putnam (1983)
Computational Psychology and Interpretation p153
REALISM AND REASON
Philosophical Papers Volume 3
"The best, though by no means the only, evidence for
desires and beliefs is action, and this suggests the
possibility of a theory that deals directly with the
relations between actions, and treats wants and thoughts
as theoretical constructs. A sophisticated theory along
these lines was proposed by Frank Ramsey [#1]. Ramsey
was primarily interested in providing a foundation in
behaviour for the idea that a person accords one or
another degree of credence to a proposition. Ramsey was
able to show that if the pattern of an individual's
preferences or choices among an unlimited set of
alternatives meets certain conditions, then that
individual can be taken to be acting so as to maximize
expected utility, that is, he acts as if he assigns
values to the outcomes on an interval scale, judges the
plausibility of the truth of propositions on a ratio
scale, and chooses the alternative with the highest
computed expected yield.
Ramsey's theory suggests an experimental procedure for
disengaging the roles of subjective probability (or
degree of belief) and subjective value in choice
behaviour. Clearly, if it may be assumed that an agent
judges probabilities in accord with frequencies or so
called objective probabilities, it is easy to compute
from his choices among gambles what his values are; and
similarly one can compute his degree of belief in
various propositions if one can assume what his values
are, say, linear in money. But neither assumption seems
justified in advance of evidence, and since choices are
the resultant of both factors, how can either factor be
derived from choices until the other is known? Here, in
effect, is Ramsey's solution: we can tell that a man
judges an event as likely to happen as not if he doesn't
care whether an attractive or an unattractive outcome is
tied to it, if he is indifferent, say, between these two
options:
- Option 1 Option 2
If it rains you get: $1000 a kick
If it doesn't rain: a kick $1000
Using this event with a subjective probability of one
half, it is possible to scale values generally, and
using these values, to scale probabilities.
In many ways, this theory takes a long step towards
scientific respectability. It gives up trying to explain
actions one at a time by appeal to something more basic,
and instead postulates a pattern in behaviour from which
beliefs and attitudes can be inferred. This
simultaneously removes the need for establishing the
existence of beliefs and attitudes apart from behaviour,
and takes into systematic account (as a construct) the
whole relevant network of cognitive and motivational
factors. The theory assigns numbers to measure degrees
of belief and desire, as is essential if it is to be
adequate to prediction, and yet it does this on the
basis of purely qualitative evidence (preferences or
choices between pairs of alternatives). Can we accept
such a theory of decision as a scientific theory of
behaviour on a par with a physical theory?
Well first we must notice that a theory like Ramsey's
has no predictive power at all unless it is assumed that
beliefs and values do not change over time. The theory
merely puts restrictions on a temporal cross-section of
an agent's disposition to choose. If we try
experimentally to test the theory, we run into the
difficulty that the testing procedure disturbs the
pattern we wish to examine. After spending several years
testing variants of Ramsey;s theory on human subjects, I
tried the following experiment (with Merrill Carlsmith).
Subjects made all possible pairwise choices within a
small field of alternatives, and in a series of
subsequent sessions, were offered the same set of
options over and over. The alternatives were complex
enough to mask the fact of repetition, so that subjects
could not remember their previous choices, and pay-offs
were deferred to the end of the experiment so that there
was no normal learning or conditioning. The choices for
each session and each subject were then examined for
inconsistencies - cases where someone had chosen A over
B, B over C, and C over A. It was found that as time
went on, people became steadily more consistent;
intransitivities were gradually eliminated; after six
sessions, all subjects were close to being perfectly
consistent. This was enough to show that a static theory
like Ramsey's could not, even under the most carefully
controlled conditions, yield accurate predictions:
merely making choices (with no reward or feedback)
alters future choices. There was also and entirely
unexpected result. If the choices of an individual over
all trials were combined, on the assumption that his
'real' preference was for the alternative of a par he
chose most often, then there were almost no
inconsistencies at all. Apparently, from the start there
were underlying and consistent values which were better
and better realised in choice. I found it impossible to
construct a formal theory that could explain this, and
gave up my career as an experimental psychologist.
Davidson (1980)
Psychology as Philosophy
ESSAYS ON ACTIONS AND EVENTS
"Before drawing a moral from this experiment, let me
return to Ramsey's ingenious method for abstracting
subjective values and probabilities simultaneously from
choice behaviour. Application of the theory depends, it
will be remembered, on finding a proposition with a
certain property: it must be such that the subject does
not care whether its truth or its falsity is tied to the
more attractive of the two outcomes. In the context of
theory, it is clear that this means ANY two outcomes.
So, if the theory is to operate at all, if it is to be
used to measure degrees of belief and the relative force
of desire, it is first necessary that there be a
proposition of the required sort. Apparently, this is an
empirical question; yet the claim that the theory is
true is then a very sweeping empirical claim. If it is
ever correct, according to the theory, to say that for a
given person a certain event has some specific
subjective probability, it must be the case that a
detailed and powerful theory is true concerning the
pattern of that person's choice behaviour. And if it is
ever reasonable to assert, for example, that one event
has a higher subjective probability than another for a
given person, then there must be good reason to believe
that a very strong theory is true rather than false.
From a formal point of view, the situation is
analogous to fundamental measurement in physics, say of
length, temperature, or mass. The assignment of numbers
to measure any of these assumes that a very tight set of
conditions holds. And I think that we can treat the
cases as parallel in the following respect. Just as the
satisfaction of the conditions for measuring length or
mass may be viewed as constitutive of the range of
application of the sciences that employ these measures,
so the satisfaction of conditions of consistency and
rational coherence may be viewed as constitutive of the
range of applications of such concepts as those of
belief, desire, intention and action. It is not easy to
describe in convincing detail an experiment that would
presuade us that the transitivity of the relation of
HEAVIER THAN had failed. Though the case is not as
extreme, I do not think we can clearly say what should
convince us that a man at a given time (without change
of mind) preferred A to B, B to C, and C to A. The
reason for our difficulty is that we cannot make good
sense of an attribution of preference except against a
background of coherent attitudes.
The significance of the experiment I described a page or
so back is that it demonstrates how easy it is to
interpret choice behaviour so as to give a consistent
and rational pattern. When we learn that apparent
inconsistency fades with repetition but no learning, we
are apt to count the inconsistency as merely apparent.
When we learn that frequency of choice may be taken as
evidence for an underlying consistent disposition, we
may decide to write off what what seem to be
inconsistent choices as failures of perception or
execution. My point is not merely that the data are open
to more than one interpretation, though this is
obviously true. My point is that if we are intelligibly
to attribute attitudes and beliefs, or usefully to
describe motions as behaviour, then we are committed to
finding, in the pattern of behaviour, belief, and
desire, a large degree of rationality and consistency.
A final consideration may help to reinforce this claim.
In the experiments I have been describing, it is common
to offer the subject choices verbally, and for him to
respond by saying what he chooses. We assume that the
subject is choosing between the alternatives described
by the experimenter, i.e. that the words used subject
and experimenter have the same interpretation. A more
satisfying theory would drop the assumption by
incorporating in decision theory a theory of
communication. This is not a peripheral issue, because
except in the case of the most primitive beliefs and
desires, establishing the correctness of an attribution
of belief or desire involves much the same problems as
showing that we have understood the words of another.
Suppose I offer a person an apple and a pear. He points
to the apple, and I record that he has chosen the apple.
By describing his action in this way, I imply that he
intended to point to the apple, and that by pointing he
intended to indicate his choice. I also imply that he
believed he was choosing an apple. In attributing
beliefs we can make very fine distinctions, as fine as
our language provides. Not only is there a difference
between his believing he is choosing an apple and his
believing he is choosing a pear. There is even a
difference between his believing he is choosing the
best apple in the box and his believing he is choosing
the largest apple, and this can happen when the largest
is the best.
All the distinctions available in our language are used
in the attribution of belief (and desire and intention);
this is perhaps obvious from the fact that we can
attribute a belief by putting any declarative sentence
after the words, 'He believes that'. There is every
reason to hold, then, that establishing the correctness
of an attribution of belief is no easier than
interpreting a man's speech. But I think we can go
further, and say that the problems are identical.
Beliefs cannot be ascertained in general without command
of a man's language; and we cannot master a man's
language without knowing much of what he believes.
Unless someone could talk with him, it would not be
possible to know that a man believed Fermat's last
theorem to be true, or that he believed Napoleon had all
the qualities of a great general.
'The reason we cannot understand what a man means by
what he says without knowing a good deal about his
beliefs is this. In order to interpret verbal behaviour,
we must be able to tell when a speaker holds a sentence
he speaks to be true. But sentences are held to be true
partly because of what is believed, and partly because
of what the speaker means by his words. The problem of
interpretation therefore is the problem of abstracting
simultaneously the roles of belief and meaning from the
pattern of sentences to which a speaker subscribes over
time. The situation is like that in decision theory:
just as we cannot infer beliefs from choices without
also inferring desires, so we cannot decide what a man
means by what he says without at the same time
constructing a theory about what he believes.
In the case of language, the basic strategy must be to
assume that by and large a speaker we do not yet
understand is consistent and correct in his beliefs -
according to our own standards, of course. Following
this strategy makes it possible to pair up sentences the
speaker utters with sentences of our own that we hold
true under like circumstances. When this is done
systematically, the result is a method of translation.
Once the project is under way, it is possible, and
indeed necessary, to allow some slack for error or
difference of opinion. But we cannot make sense of error
until we have established a base of agreement.
The interpretation of verbal behaviour thus shows the
salient features of the explanation of behaviour
generally: we cannot profitably take the parts one by
one (the words and sentences), for it is only in the
context of the system (language) that their role can be
specified. When we turn to the task of interpreting the
pattern, we notice the need to find it in accord, within
limits, with the standards of rationality. In the case
of language, this is apparent, because understanding it
is TRANSLATING it into our own system of concepts. But
in fact the case is no different with beliefs, desires,
and actions.
The constitutive force in the realm of behaviour derives
from the need to view others, nearly enough, as like
ourselves. As long as it is behaviour and not something
else we want to explain and describe, we must warp the
evidence to fit this frame. Physical concepts have
different constitutive elements. Standing ready, as we
must, to adjust psychological terms to one set of
standards and physical terms to another' we know that we
cannot insist on a sharp and law-like connection between
them. Since psychological phenomena do not constitute a
closed system, this amounts to saying they are not, even
in theory, amenable to precise prediction or subsumption
under deterministic laws. The limit thus placed on the
social sciences is set not by nature, but by us when we
decide to view men as rational agents with goals and
purposes, and as subject to moral evaluation.
Davidson (1980)
Psychology as Philosophy
ESSAYS ON ACTIONS AND EVENTS
[#1] 'Truth and Probability'. Ramsey's theory, in a less
interesting form, was later, and independently,
rediscovered by von Neumann and Morgenstern, and is
sometimes called a theory of decision under uncertainty,
or simply decision theory, by economists and
psychologists.
'The conclusion defended in this paper is a familiar
one, and is shared by many philosophers and, probably,
psychologists. The position might be put this way: the
study of human action, motives, desires, beliefs, memory
and learning, at least so far as these are logically
tied to the so called 'propositional attitudes', cannot
employ the same methods as, or be reduced to, the more
precise physical sciences. Many would agree, too, that
we cannot expect to find strict psychophysical laws. If
there is anything new in what I say on this topic, it is
in the details of the reasons that I give for saying
that generalizations that combine psychological and
physical predicates are not lawlike in the strong sense
that physical laws can be. What apparently arouses the
most doubt and opposition is my attempt to combine the
view that psychological concepts have an autonomy
relative to the physical with a monistic ontology and a
causal analysis of action.
I thought, then, that my CONCLUSIONS (in contrast,
perhaps, to my arguments) concerning the nature of
cognitive psychology as a science were neither new nor
apt to excite much debate. But hearing Professor
Peters's generous and sensitive remarks, as well as
listening to the discussion of my paper at the
Canterbury conference, made me realise I had given the
impression that I was making some sort of attack on
psychology generally, or at least on its right to be
called a science. That is certainly not what I intended,
but I do see how things I wrote could bear that
interpretation. So here I will briefly try to set
matters straight.
First, let me re-emphasize the fact that my arguments
are limited in application to branches of psychology
that make essential reference to 'propositional
attitudes' such as belief, desire, and memory, or use
concepts logically tied to these, such as perception,
learning, and action. (Some of these concepts may not
always show intensionality, and in such cases are also
exempt.)
Second, I made much of the fact that psychophysical
generalizations must be treated as irreducibly
statistical in character, in contrast to sciences where
in principle exceptions can be taken care of by
refinements couched in a homogeneous vocabulary. This is
not a REPROACH to psychology, nor does it mean its
predictions and explanations are in fact less precise
than those of many other sciences. I assume that in
application meteorology and geology, for example, are
far less precise than much work on perception. The point
is not the actual degree of looseness in psychology, but
what guarantees that it can't be eliminated, namely the
conceptually hermaphroditic character of its
generalizations.
Third, I argued that the part of psychology with which I
was concerned cannot be, or be incorporated in, a closed
science. This is due to the irreducibility of
psychological concepts, and to the fact that
psychological events and states often have causes that
have no natural psychological descriptions. I do not
want to say that analogous remarks may not hold for some
other sciences, for example, biology. But I do not know
how to show that the concepts of biology are
nomologically irreducible to the concepts of physics.
What sets apart certain psychological concepts - their
intensionality - does not apply to the concepts of
biology.
'Science' being the honorific word it is in some
quarters, it would be meretricious to summarise these
points by saying that psychology (the part with which we
are concerned) is not a science; the conclusion is
rather that psychology is set off from other sciences in
an important and interesting way. The argument against
the existence of strict psychophysical laws provides the
key to psychology's uniqueness: the argument led from
the necessarily holistic character of interpretations of
propositional attitudes to the recognition of an
irreducably normative element in all attributions of
attitude. In the formation of hypotheses and the reading
of evidence, there is no way psychology can avoid
consideration of the nature of rationality, of coherence
and consistency. At one end of the spectrum, logic and
rational decision theory are psychological theories from
which the obviously empirical has been drained. At the
other end, there is some form of behaviourism better
imagined than described from which all taint of the
normative has been subtracted. Psychology, if it deals
with propositional attitudes, hovers in between. This
branch of the subject cannot be divorced from such
questions as what constitutes a good argument, a valid
inference, a rational plan, or a good reason for acting.
These questions also belong to the traditional concerns
of philosophy, which is my excuse for my title.'
Donald Davidson
Psychology as Philosophy, pp.239-241
ESSAYS ON ACTIONS AND EVENTS (1980).
>
> As for the rest of your post (which had nothing to do with what I
> was talking about, by the way), I noticed a couple of factual errors
> that you might want to straighten out before you go ranting on about
> how ignorant people will not be shown charity.
>
> : Finetti, 1968; Savage, 1954) regards subjective
> : probability as the quantified opinion of an idealized
> : person. Specifically, the subjective probability of a
>
> Subjective probability does not have to be the opinion of an
> idealized person. Beliefs merely have to be consistent; they
> don't have to be reasonable to conform to the axioms of decision
> theory.
You're arguing with Tversky and Kahneman.
>
> : individual. Unfortunately, there can be no simple formal
> : procedure for assessing the compatibility of a set of
> : probability judgments with the judge's total system of
> : beliefs. The rational judge will nevertheless strive for
>
> There is a simple formal procedure: If you can make a Dutch
> Book against someone, their beliefs are inconsistent.
>
> -Casey Durfee
>
Perhaps you need to reassess you familiarity with the literature
and your appraisal of the relevance of my posts in response to
yours, and in the context of the title of this thread.
Omar Haneef
A word before I go on: Rickert's (Neil?) responses indicate to me that
our disagreement is NOT of the sort that I had previously thought it
was. It is not that we disagree that people are irrational, or that T
& K had useful things to contribute, but rather on in the
interpretation of the T & K work. This is, I am glad to say, a healthy
sort of disagreement. Anyway, lets go on.
Neil Rickert (ric...@cs.niu.edu) wrote:
: oha...@andromeda.rutgers.edu (Omar Haneef) writes:
: >Neil Rickert (ric...@cs.niu.edu) wrote:
: >: oha...@andromeda.rutgers.edu (Omar Haneef) writes:
: >: >(2) The irrationality of common sense judgements has been well
: >: >documented by a number of psychologists -- hate them if you want to --
: >: >such as Nisbett & Ross, and Tversky & Kahneman.
: >: I don't hate them. I simply reject the claim that their experiments
: >: demonstrate irrationality.
: > I didn't mean to say you do hate them, I know some people
: >do. But surely you have to present an argument against their
: >opinions. There are arguments out there, some of which I buy quite
: >wholeheartedly. (There is that German statistician -- Gergesomething
: >or other -- who is brilliant at debunking T & K type base rate
: >stuff. But he had a great argument to go with it.)
: I think you are referring to Gigerenzer.
Ah. I thought I didn't forget names. Thats it.
: According to common definitions of rationality, whether a particular
: decision is rational depends on: the inputs, the knowledge, the
: intended outcome. My objection is that T&K judge the rationality of
: their experimental subjects on what T&K consider to be the inputs and
: the intended outcome, and on T&Ks estimation of the subject's
: knowledge. The experimental subjects might actually be solving a
: very different problem from what T&K assume, and their actual
: knowledge may be different from what is assumed.
I think in the above formulation, the distance between
yourself and T&K is greatly diminished. I'm not clear where you
disagree, then. Suppose we set this formalism up:
T&K think they are setting up problem P, for which solution S is
optimal in world W arrived at by heuristic H (not the only one for S,
but lets say its the normative process for arriving at S). Then you
say the subject is actually working on problem P' for which S' is
optimal using H'. However, the subject is in the same world W where S
is optimal and the "real" problem is P.
Upshot: What is the big difference between saying that the error is in
attempting to use H' to find S' or attempting to use H' to find S? You
seem to be saying that the error is not in the fact that they use the
wrong technique to solve the problem, but that they are attempting to
solve the wrong problem because of constraints on their
problem-solving capacity.
Is that really a disagreement?
: >1. Give your friend $20
: >2. Tell him he can keep it, or gamble $10 for a double or nothing coin toss.
: >and
: >1. Give your friend $10
: >2. Tell him he can either have another $10, or take a coin toss chance for $20
: These are two very different problems.
In a certain sense, of course they are different, and in a certain
sense they are not. I think its to T&K's credit that they come up with
an psychological specification of the difference.
: >In terms of changes of states of wealth, the two questions are
: >identical, but people, it turns out, are overwhelmingly more likely to
: >risk in the second version. Clearly, they are not using the same
: >heuristic.
: Sure. But you are assuming that people are treating this as a way of
: changing wealth. That is roughly equivalent to saying that golf and
: soccer are identical games, because in each the aim is to get a ball
: into some sort of hole.
: In fact the problem, as you stated it, is a sort of game which exists
: in a social setting. Treating it as a question of change of wealth
: ignores the social parameters.
Look at what you just said above! You are a natural T&K kinda guy!
They (and I) would agree with you! Absolutely, thinking of them in
terms of changes of wealth is NOT how people do it, people do it
differently. How? And why? Why would they not be sensitive to changes
in wealth? What constraints does that put on the system? How is this
way cheaper? These are questions they want answers to.
: > There are four cards. Each card has a number on one side and a
: >letter on the other. The cards show the following faces: A, B, 1,
: >2. You want to see if the cards follow this simple rule: "If there is
: >a vowel on one side of the card, the other will be even." Which cards, at minimum, do you have to turn to discover the rule?
: > Now most people -- perhaps not yourself, being so well trained
: >against your natural impulses -- will choose A and 2 (try it!).
: Nor surprising. This is not a problem that people typically need to
: solve. So there is no reason to suggest that they have direct
: knowledge on the best way to tackle it. Thinking about such a
: problem in terms of combinatorial logic is familiar enough to
: mathematicians and logicians but not to most people. For most
: people, a problem like this stresses the limitations of short term
: memory.
The STM/social familiarity hypothesis is an interesting one, but there
is reason to believe it is incorrect. Oh, I think some members of the
crowd -- T&K, or Cosmides and Humphreys or Girgenezzer -- went ahead
and tried the same sort of problem in a social, familiar setting
WITHOUT referring to cheaters. The idea was that you had to go to an
exotic tribe and figure out what the rule was. Then they tried a
different version where they specifically clued in that social
CHEATING was going on. Subjects are much worse at the no-cheating
version than the cheating version by some enormous quantity.
: > Another kicker: try an experiment identical in form but one
: >that (as Gergenezzer or whatever would put it) occurs in an
: >environment closer to our evoluitionary one, and the effect
: >disappears.
: Right. Because, in a more familiar setting they have experience that
: they can call upon, and thus the demands on short term memory are
: considerably reduced.
Again, some people thought that familiarity had a role, but this has
been ruled out since.
: > Tell your friend she or he is a bouncer at a bar, and you want
: >to make sure that no one over 21 is drinking. You can see four people,
: >but some have theur drinks covered and some have their IDs
: >covered. You see: a beer, a coke, an age 16, and an age 25. Which
: >drinks and/or IDs do you have to check to make sure there is nothing
: >illegal occuring? The idea is ridiculously obvious and comes
: >instantly. Cosmides and Humphreys claim this is evidence that our
: >evolutionary modules are designed for specific tasks such as cheater
: >detection.
: My explanation is that our knowledge is not in the form of
: propositions (or beliefs), but is more in the form of abilities to
: solve specific problems. The bouncer problem is closer to
: experience, and thus something for which a person might have specific
: abilities. The card problem is more abstract, and most people have
: little experience and thus little ability in dealing with abstract
: problems.
see above.
: >: >Turns out that doctors were NOT taking base rates into account when
: >: >signing people up for preventive surgery or test. For examply, if a
: >: >disease is rare, say 1 in 10000 people have it, then even a test that
: >: >is accurate 98% of the time is useless since it is more likely that
: >: >the test is false. (If you know Bayes Theoreom you can try it out for
: >: >yourself).
: >: If the test errors are independent of whether a person is diseased,
: >: then this does not obviously follow. If a disease is that rare, then
: >: most doctors do not give tests unless they see other indicators which
: >: suggested the possibility of that disease.
: > No, it turns out that professional doctors actually were
: >ignoring base rates in the relevant sense and after this research
: >caught the public eye, doctors started getting trained differently.
: You still have to ask what problem were the doctors solving. They
: may be solving a problem of dealing with anxiety by a patient who is
: worried about having a particular disease. They might be solving the
: problem of minimizing their exposure to malpractice law suits. The
: publication of research on base rates gives doctors evidence that
: they can use to defend themselves in law suits, and gives information
: they can give to anxious patients as an alternative way to calm them
: down. In other words, the publication changes the doctor's
: knowledge, and changes the nature of the problem that the doctor is
: solving. At best, the argument for irrationality is that if you
: ignore all of these changes then the behavior of doctors seems
: irrational.
Interesting, I hadn't thought of that. I do think, though, that the
doctors who hadn't thought of the base rate neglect problem actually
WERE surprised that 95% or 99% accurate procedures were useless
against very rare diseases. I think this surprise -- which I relate
to, because I felt it -- is what makes me think that, in fact, people
do ignore base rates usually. And, in addition, they ignore it in
domains they ought not to.
: >: In any case, what basis is there for the claim that Bayesian
: >: reasoning should be normative?
: >Yes, well this is a question worthy of bringing up. People do question
: >why Bayesian reasoning should be normative. I, for one, do not
: >understand the complaint, but have heard it often enough that I feel
: >they must be on to something. In the contexts given, though, isn't
: >Bayesian reasoning superioir to ignoring the base rates?
: Here is the problem. If I give you a bucket full of bits, and ask
: you to apply Bayesian reasoning, you are not going to come up with
: much. In fact, you probably won't even know how to start. However,
: if these bits have been structured in some way, then you can begin to
: construct some statistical information about the structures.
: Bayesian reasoning, or indeed other kinds of statistical reasoning,
: are sensitive to the way in which the data is structured. Since we
: have no way of knowing how a particular person's brain has structured
: that person's experiential data, we cannot tell which particular
: Bayesian model we might reasonably expect a particular person to be
: using.
I think T&K would agree that the brain does NOT structure data in such
a way that Bayesian reasoning can be used on it. But that is their
point. The world does afford situations where it is better, on
occasion to use Bayesian reasoning (for instance) and so, in those
situations, the brain is likely to provide sub-optimal solutions
unabetted by external aids.
-Omar
Sergio Navega
Sorry to interfere in such elaborate discussion, guys, but
I thought I could show a slightly different view.
Disclaimer:
- I haven't read the whole thread (I'm not *that* mad!)
- Tversky & Kahneman is still in my reading list
- Gigerenzer is unread, below Tversky in that list
(Maybe it's better start with him)
>
> Amos Tversky, a tireless and skilled worker in this
> area, seems to have come to much the same conclusion. He
> tells of some experiments done by him and Daniel
> Kahneman in Jerusalem. They asked subjects to choose
> between A and B, and C and D, in the following
> situation:
>
> Choice I: A = (1000,1/2,0) B= (400)
> Choice II: C = (1000,1/10,0) D= (400,1/5,0)
>
> Here A, for example, is a wager that is equally apt
> (subjectively) to pay $1,000 or nothing; B is simply
> receiving $400 (no risk). Tversky says almost all
> subjects picked B over A and C over D; the result held
> for naive and educated subjects, and analogous results
> were obtained for different payoffs and probabilities.
> The trouble is, this pattern of preferences is
> incompatible with decision theory.'
>
Of course it is incompatible with decision theory! Decision theory
does not take into account EMOTIONS! Bayesian belief networks never
cares if one lost money previously on life.
If this is the kind of reasoning that supports Tversky & Kahneman's
point, I think I will stick with their oppositors!
In case this isn't obvious enough for some, almost every decision
we make has *deep* emotional reasons. An addicted gambler, that which
spends lots of money in Las Vegas, decides *very differently* (and
in all aspects) from that family-type, law-respecting, tax-paying guy.
A cold, stock trader guy will decide even differently. A politician may
offer some money to the Tversky guy to "bend" a little the results,
because it will be good for the next elections :)
Personality traits affects everything.
Ok, science should *not* be done on purely emotional basis, but *any*
scientist will have profound influences of his emotions. Those who
fail to recognize (and understand!) this will be driven by emotions
without knowing it. To say that human decision making is flawed is
the same baloney as saying that your beloved car is sad because you
had just sold it.
Regards,
Sergio Navega.
CASEY DURFEE S
Durfee:
: > Subjective probability does not have to be the opinion of an
: > don't have to be reasonable to conform to the axioms of decision
: > theory.
Longley:
: You're arguing with Tversky and Kahneman.
Durfee:
First of all, I wasn't "arguing" with anyone, I was pointing out a
factual error in the document that you posted. Secondly,
Kahneman and Tversky's studies had nothing to do with coming up with
criteria for whether beliefs are reasonable or not. Kahneman and
Tversky tried to explain why agents who thought they were engaging
in reasonable behaviour were violating the Sure Thing Principle.
The factual error that I pointed out was that you can hold any sort
of unreasonable beliefs you want and still be considered a Bayesian as long
as you are consistent. What Kahneman and Tversky pointed out was
that a vast majority of people in certain decision situations will
make what looks like a reasonable choice but violate decision theory.
There's no argument there.
Longley (from Fragments):
: > : individual. Unfortunately, there can be no simple formal
: > : procedure for assessing the compatibility of a set of
: > : probability judgments with the judge's total system of
: > : beliefs. The rational judge will nevertheless strive for
Durfee:
: > There is a simple formal procedure: If you can make a Dutch
: > Book against someone, their beliefs are inconsistent.
Longley:
: Perhaps you need to reassess you familiarity with the literature
: and your appraisal of the relevance of my posts in response to
: yours, and in the context of the title of this thread.
Durfee:
Let's do a quick summary here so people can see how silly this all is:
I responded to something that Neil Rickert said about decision theory.
My original post was not addressed to you at all. You, butting in,
claimed that decision theory was primarily normative; I noted that this
was factually untrue. In response you did not address the fact that you
had made a factual error but instead doubted my qualifications to point
out mistakes you had made. I then pointed out two more factual errors
that I, as a philospher who has done graduate work in decision theory,
found in one quick, casual reading of your work. You then responded by
not acknowledging that you were wrong, again deciding to attack my
credentials. I've made one last attempt here to point out the error in
your work. If you choose not to fix these factual errors, then there's
nothing more I can do.
I'm through speaking to you. I find it impossible to have a conversation
with you, since you try to change the topic every time I note an
inconsistency in your work. Instead of answering intellectual criticism
in this newsgroup, you post long, irrelevant excerpts and slander anyone
who points out that said excerpts are both irrelevant and rife with
factual errors. I detect many of the symptoms of borderline
schitzophrenia in you: wild cognitive leaps, disregard for the facts,
inability to stay focused on the conversation at hand, and a general
poverty of speech. Although I am not a licensed psychologist, I advise that
you see someone who is.
-Casey Durfee
David Longley
The above is folk psychological reasoning at work. The problem
with it is that it is, you will find, resistant to logical
inference, and therefore resistant to falsification (test in
general). You may not think this is the case, but that's what I
try to bring out in "Fragments" and through illustrative examples
of the peculiarities of the intensional idioms such as "said
that".
> Personality traits affects everything.
Appeal to personality "traits" often serves to do little more
than distract one from useful analysis. The consistency across
contexts is characteristically intensional.
Neil Rickert
oha...@andromeda.rutgers.edu (Omar Haneef) writes:
>: According to common definitions of rationality, whether a particular
>: decision is rational depends on: the inputs, the knowledge, the
>: intended outcome. My objection is that T&K judge the rationality of
>: their experimental subjects on what T&K consider to be the inputs and
>: the intended outcome, and on T&Ks estimation of the subject's
>: knowledge. The experimental subjects might actually be solving a
>: very different problem from what T&K assume, and their actual
>: knowledge may be different from what is assumed.
> I think in the above formulation, the distance between
>yourself and T&K is greatly diminished. I'm not clear where you
>disagree, then. Suppose we set this formalism up:
>T&K think they are setting up problem P, for which solution S is
>optimal in world W arrived at by heuristic H (not the only one for S,
>but lets say its the normative process for arriving at S). Then you
>say the subject is actually working on problem P' for which S' is
>optimal using H'. However, the subject is in the same world W where S
>is optimal and the "real" problem is P.
I agree with everything except the notion that there is a single
"real" problem. I don't see any basis for this.
Perhaps the problem for the test subject is: "I'm stuck in a room
where a crazy professor is asking all kinds of silly questions. How
do I get out of here?" What counts as the problem is bound to be
relative to the person considering the situation.
>Upshot: What is the big difference between saying that the error is in
>attempting to use H' to find S' or attempting to use H' to find S? You
>seem to be saying that the error is not in the fact that they use the
>wrong technique to solve the problem, but that they are attempting to
>solve the wrong problem because of constraints on their
>problem-solving capacity.
I don't agree that this is always a matter of problem-solving
capacity. And I don't agree with the notion of a "wrong problem."
What is a right problem depends very much on contextual
considerations. And the context for the experimenter is very
different from the context for the subject.
>Is that really a disagreement?
>: >1. Give your friend $20
>: >2. Tell him he can keep it, or gamble $10 for a double or nothing coin toss.
>: >and
>: >1. Give your friend $10
>: >2. Tell him he can either have another $10, or take a coin toss chance for $20
>: These are two very different problems.
>In a certain sense, of course they are different, and in a certain
>sense they are not. I think its to T&K's credit that they come up with
>an psychological specification of the difference.
If you are faced with this choice, then the state of wealth is only
one consideration. Another consideration is that you are cooperating
with a friend, and you don't want to seem uncooperative. The way
that the problem is phrased changes the expectation as to what kind
of behavior would be considered cooperative.
>: >In terms of changes of states of wealth, the two questions are
>: >identical, but people, it turns out, are overwhelmingly more likely to
>: >risk in the second version. Clearly, they are not using the same
>: >heuristic.
>: Sure. But you are assuming that people are treating this as a way of
>: changing wealth. That is roughly equivalent to saying that golf and
>: soccer are identical games, because in each the aim is to get a ball
>: into some sort of hole.
>: In fact the problem, as you stated it, is a sort of game which exists
>: in a social setting. Treating it as a question of change of wealth
>: ignores the social parameters.
>Look at what you just said above! You are a natural T&K kinda guy!
>They (and I) would agree with you! Absolutely, thinking of them in
>terms of changes of wealth is NOT how people do it, people do it
>differently. How? And why? Why would they not be sensitive to changes
>in wealth? What constraints does that put on the system? How is this
>way cheaper? These are questions they want answers to.
There are things we value from our social network which are more
important than wealth. We are members of a social species, and thus
biology gives particular importance to our social relations. In
fact, most of us could not survive in under the law of the jungle, so
that our very existence depends on social relations. Moreover, we
are all a little bit different, so we will be a little different in
what we consider important.
>The STM/social familiarity hypothesis is an interesting one, but there
>is reason to believe it is incorrect. Oh, I think some members of the
>crowd -- T&K, or Cosmides and Humphreys or Girgenezzer -- went ahead
>and tried the same sort of problem in a social, familiar setting
>WITHOUT referring to cheaters. The idea was that you had to go to an
>exotic tribe and figure out what the rule was. Then they tried a
>different version where they specifically clued in that social
>CHEATING was going on. Subjects are much worse at the no-cheating
>version than the cheating version by some enormous quantity.
I don't pretend to be able to read minds. It is my experience that
most people have trouble with abstract thinking, and the "correct"
solution to this card problem requires some kind of abstract
thinking.
The natural thing for a person to do, is to use trial and error. The
limitation to only two trials is what makes this an artificial
problem.
>: You still have to ask what problem were the doctors solving. They
>: may be solving a problem of dealing with anxiety by a patient who is
>: worried about having a particular disease. They might be solving the
>: problem of minimizing their exposure to malpractice law suits. The
>: publication of research on base rates gives doctors evidence that
>: they can use to defend themselves in law suits, and gives information
>: they can give to anxious patients as an alternative way to calm them
>: down. In other words, the publication changes the doctor's
>: knowledge, and changes the nature of the problem that the doctor is
>: solving. At best, the argument for irrationality is that if you
>: ignore all of these changes then the behavior of doctors seems
>: irrational.
>Interesting, I hadn't thought of that. I do think, though, that the
>doctors who hadn't thought of the base rate neglect problem actually
>WERE surprised that 95% or 99% accurate procedures were useless
>against very rare diseases. I think this surprise -- which I relate
>to, because I felt it -- is what makes me think that, in fact, people
>do ignore base rates usually. And, in addition, they ignore it in
>domains they ought not to.
I would expect this to be a surprise. But I don't see anything that
indicates irrationality. The doctor saw a problem, and had a way of
solving it which generally worked well enough. Now he has been shown
that there are circumstances where a different way of solving the
problem can be better. The mistake is to frame this as a case of
base rate neglect. The doctor had a way of dealing with the problem
for which base rate information was not required, and thus it is not
neglect to not consider that information.
>I think T&K would agree that the brain does NOT structure data in such
>a way that Bayesian reasoning can be used on it. But that is their
>point. The world does afford situations where it is better, on
>occasion to use Bayesian reasoning (for instance) and so, in those
>situations, the brain is likely to provide sub-optimal solutions
>unabetted by external aids.
Ok. But I don't see anything irrational about using well tried and
proven methods, just because those methods are occasionally
suboptimal. I don't worry about whether the route I take, driving to
work, is suboptimal. Why do we consider intellectual problems
different than ordinary problems, and consider it irrational to use a
suboptimal solution to a intellectual problem?
Neil Rickert
Da...@longley.demon.co.uk (David Longley) writes:
> Predicate logic is normative,
Why?
> , so is
>mathematics.
Again, why?
Is there some reason to claim these are normative, or is it merely a
piece of doctrine that is to be tossed out?
If a problem is explicitely stated in predicate logic, I can agree
that the logic solution might reasonably be considered normative.
And if a problem is explicitely presented as a mathematics problem,
the mathematical solution might reasonably be taken as normative.
But every day problems are not explicitly logic problems, nor are
they explicitly mathematics problems.
David Longley
In article <35996...@rcfnews.cs.umass.edu>
> Let's do a quick summary here so people can see how silly this all is:
> I responded to something that Neil Rickert said about decision theory.
> My original post was not addressed to you at all. You, butting in,
> claimed that decision theory was primarily normative; I noted that this
> was factually untrue. In response you did not address the fact that you
> had made a factual error but instead doubted my qualifications to point
> out mistakes you had made. I then pointed out two more factual errors
> that I, as a philospher who has done graduate work in decision theory,
> found in one quick, casual reading of your work. You then responded by
> not acknowledging that you were wrong, again deciding to attack my
> credentials. I've made one last attempt here to point out the error in
> your work. If you choose not to fix these factual errors, then there's
> nothing more I can do.
From: Decision. Probability, and Utility
Selected Readings
Ed P Gardenfors & N-E Sahlin (1988)
Cambridge Press
(1. From the introduction - re: Normative models)
o o o
"We can now formulate the fundamental decision rule of
Bayesian theory:
The principle of maximizing expected utility (MEU): In a
given decision situation the decision maker should
choose the alternative with maximal expected utility (or
one of the alternatives with maximal expected utility if
there are more than one).
At this point we must introduce a distinction
between a normative and a descriptive decision theory. A
normative theory gives rules for what a rational
decision maker should do in various situations. A
descriptive theory aims at describing how people in fact
(rationally or not) make decisions in different
situations. The principle of maximizing expected utility
can be given both a normative and a descriptive reading.
* The normative reading of the MEU principle has been
* prevalent among economists who are interested in making
* rational decisions in the market (e.g., see Sen, 1977)
* and among philosophers who are interested in general
* foundations of rationality. Among psychologists the MEU
principle has served as a theory about people's actual
behavior (e.g., see Slovic, Lichtenstein, and Fischhoff,
1983, Kahneman, Slovic, and Tversky, 1982, for surveys
of this area). In Section 3, we return to some of the
problems that the MEU principle has encountered as a
descriptive theory. It should be noted, however, that
the distinction between a normative and a descriptive
interpretation of a decision theory is slippery: A
proposed normative theory is tested against our
intuitive understanding of rational decisions, and
people's actual behavior may be influenced by existing
normative theories. For a discussion of the interplay
between normative and descriptive interpretations of
psychological experiments in general and decision
theoretic experiments in particular, cf. Sahlin (1987).
2. From the same book. Re: purpose of T&Ks work on
decisions under *RISK*.
o o o
Prospect theory: An analysis of decision under risk
Daniel Kahneman and Amos Tversky
1. Introduction
* Expected utility theory has dominated the analysis of
* decision making under risk. It has been generally
* accepted as a normative model of rational choice (Keeney
* and Raiffa, 1976), and widely applied as a descriptive
* model of economic behavior (e.g., Friedman and Savage,
* 1948, and Arrow, 1971). Thus, it is assumed that all
reasonable people would wish to obey the axioms of the
theory (von Neumann & Morgenstern, 1944, and Savage,
1954), and that most people actually do, most of the
time.
The present paper describes several classes of
choice problems in which preferences systematically
violate the axioms of expected utility theory. In the
light of these observations we argue that utility
theory, as it is commonly interpreted and applied, is
not an adequate descriptive model and we propose an
alternative account of choice under risk."
>
> I'm through speaking to you. I find it impossible to have a conversation
> with you, since you try to change the topic every time I note an
> inconsistency in your work. Instead of answering intellectual criticism
> in this newsgroup, you post long, irrelevant excerpts and slander anyone
> who points out that said excerpts are both irrelevant and rife with
> factual errors. I detect many of the symptoms of borderline
> schitzophrenia in you: wild cognitive leaps, disregard for the facts,
> inability to stay focused on the conversation at hand, and a general
> poverty of speech. Although I am not a licensed psychologist, I advise that
> you see someone who is.
>
> -Casey Durfee
>
Well, my postgraduate training was under two psychiatrists, and
the applied context was the etiology and treatment of the
functional psychoses - including schizophrenia.
After over a decade of applied work in what I claim to be
normative decision making, *as* a professional psychologist, I
reckon I'm in a pretty good position to comment on what's
relevant and irrelevant - but despite that, prefer to just cite
my sources when it comes exchanges such as these.
(Incidentally, most of what has been covered to date has been
implicitly concerned with decision making under conditions of
uncertainty - not that this matters ultimately).
David Longley
The problem partially comes down to this (for more extensive
coverage/context see http://www.longley.demon.co.uk/Frag.htm and
other papers at that site.
Psychologists study how animals (including humans) learn/behave.
They have done so most extensively this century using
conditioning paradigms. Within neuroscience, small systems models
have been tested which provide quite detailed physiological
mechanisms which mediate these conditioning processes. This is
all good, descriptive science.
Since the mid 1980s, the neural network approach has had
something of a renaissance. For years these models existed in
different guises in the literature, but (IMO) largely because of
the rise of neuroscience in the late 70s and 80s (I remember
being part of this) this new look was soon regarded as a
revolutionary alternative to the earlier, basically logicist or
information theoretic approach.
One of the slippery issues in this whole area is the distinction
between the descriptive and the normative. It's such a slippery
issue that I felt I had to review material from several areas of
contemporary philosophy and psychology to bring out some
important implications for students I was teaching a course in
applied criminological psychology to. As the years went past, it
became clear to me that the implications were not at all obvious
- even to graduate psychologists, so I put the material together
as an introduction to an operational set of helpful strategies
within an relational database infrastructure running in the
English Prison System.
Such environments require a degree of control which is rare in
human society, and this is especially so in the maximum security
estate where my work was focused. Decisions have to be made about
the management of behaviour. The risks involved can be high,
forcing, one might imagine, extreme conservatism and certainly
putting a premium on accountability.
Yet, despite the great demands for observation in support of
security and control - and despite the very high staff to inmate
ratios, the extent to which decisions are made without reference
to data, base-rates, professional analysis - ie on clinical
judgement, call it what you will, is remarkable - and the
consequences are, on the basis of known research, somewhat
unsurprising.
In recent years, I have discussed this in the context of
Artificial Intelligence more generally. That is, I have
contrasted the natural, clinical judgements of common-sense or
folk-psychology with normative decision making procedures, and I
have made a case that these are coextensive with the web of
scientific knowledge we have empirically built up to date. In
some areas of application we do call this "Artificial
Intelligence", but one of my points has been that might just be
an administrative convenience (as Popper once said in another
context).
When we work to expand/apply our scientific knowledge base, we do
so from what I have called "the extensional stance". It's value
lies in the fact that naturally, limitations on how we make sense
of the world lead us to use intensional heuristics (as described
by many psychologists since the 1940s, most famously by Tversky
and Kahneman 1973). What characterises the intensional is failure
of certain axioms of logic, axioms which are the sine qua non for
logical inference - critically, quantification in and
substitution salva veritate.
The problem, (and it is much wider than one or two GOFAI
objectives, such as the attempt to incorporate common sense into
large databases or the intensional idioms such as belief into the
logicist programme), is that many in the field of AI are, I
suggest, guilty of having committed the genetic fallacy and not
having understood the nature of the cognitive revolution in
psychology. Too many fail to appreciate the descriptive nature of
psychology - as a consequence, they go on to take empirical
findings as grounds for normative principles.
When people use "effective procedures", they use them as acquired
skills, and much of life is about developing that repertoire of
skills, most formally through education. The best way we have of
measuring or even talking about intelligence, is through
relational testing of an individuals skills in what are,
fundamentally, in culture fair tests, *logical* operations.
It is a mistake, a medieval dogma, therefore, to equate common-
sense with such skills. Scientific procedures are normatively
validated procedures, learned, and refined through collective
experience and test *from the extensional stance*. This is to be
contrasted with intensional heuristics which are
characteristically neglectful of evidence and non-extensional.
The philosophical significance of these issues can, I think be
fruitfully traced back through the history of modern philosophy,
with Quine's 1951 "Two Dogmas of Empiricism" representing an, if
not the, important 20th century milestone, and through the recent
history of psychology.
I have sketched some of that history in "Fragments of Behaviour:
The Extensional Stance" http://www.longley.demon.co.uk/Frag.htm
and in supporting papers at that website.
Omar Haneef
As of this response, I think the difference between our views
-- steadily diminishing over the last few posts -- is negligible. I
think we've expounded our positions.
Neil Rickert (ric...@cs.niu.edu) wrote:
: oha...@andromeda.rutgers.edu (Omar Haneef) writes:
[snip]
: >T&K think they are setting up problem P, for which solution S is
: >optimal in world W arrived at by heuristic H (not the only one for S,
: >but lets say its the normative process for arriving at S). Then you
: >say the subject is actually working on problem P' for which S' is
: >optimal using H'. However, the subject is in the same world W where S
: >is optimal and the "real" problem is P.
: I agree with everything except the notion that there is a single
: "real" problem. I don't see any basis for this.
[snip]
: capacity. And I don't agree with the notion of a "wrong problem."
: What is a right problem depends very much on contextual
: considerations. And the context for the experimenter is very
: different from the context for the subject.
So here is another fundamental disagreement. You think that:
1. It is not the heuristic that ought to be used to explain the
difference in performance, but the target problem.
2. You disagree that a problem can be criticized as wrong.
I think, in opposition, that there isn't much difference between
saying that it is the heuristic of problem attempted that is the to
explain the performance. I do think there is such a thing as a "wrong"
problem.
[snip]
: If you are faced with this choice, then the state of wealth is only
: one consideration. Another consideration is that you are cooperating
: with a friend, and you don't want to seem uncooperative. The way
: that the problem is phrased changes the expectation as to what kind
: of behavior would be considered cooperative.
I think that is not what is going on here, but that is of
course an alternative explanation that is involved in this tradition
of explanation and counter-explanation. I'm willing to bet if you ran
the experiment so that the two states call for the same amount of
cooperation or not, you would get the same results that T&K did. This
is, however, an empirical prediction and there isn't much to say about
it untill the results are in.
: >: >In terms of changes of states of wealth, the two questions are
: >: >identical, but people, it turns out, are overwhelmingly more likely to
: >: >risk in the second version. Clearly, they are not using the same
: >: >heuristic.
: >: Sure. But you are assuming that people are treating this as a way of
: >: changing wealth. That is roughly equivalent to saying that golf and
: >: soccer are identical games, because in each the aim is to get a ball
: >: into some sort of hole.
: >: In fact the problem, as you stated it, is a sort of game which exists
: >: in a social setting. Treating it as a question of change of wealth
: >: ignores the social parameters.
: >Look at what you just said above! You are a natural T&K kinda guy!
: >They (and I) would agree with you! Absolutely, thinking of them in
: >terms of changes of wealth is NOT how people do it, people do it
: >differently. How? And why? Why would they not be sensitive to changes
: >in wealth? What constraints does that put on the system? How is this
: >way cheaper? These are questions they want answers to.
: There are things we value from our social network which are more
: important than wealth. We are members of a social species, and thus
: biology gives particular importance to our social relations. In
: fact, most of us could not survive in under the law of the jungle, so
: that our very existence depends on social relations. Moreover, we
: are all a little bit different, so we will be a little different in
: what we consider important.
These are beliefs that I myself adhere to. In GENERAL, they
would not consitute an alternative explanation to T&K. T&K's work is
supposed to be a CRITICISM of economists who think that utility curves
and rational choices is the way to treat human beings. T&K would agree
that we DO value things other than wealth, that we are involved in
social relations. None of this is to say that, in spite of this
awareness, they claim their models are the best match for their data.
: >The STM/social familiarity hypothesis is an interesting one, but there
: >is reason to believe it is incorrect. Oh, I think some members of the
: >crowd -- T&K, or Cosmides and Humphreys or Girgenezzer -- went ahead
: >and tried the same sort of problem in a social, familiar setting
: >WITHOUT referring to cheaters. The idea was that you had to go to an
: >exotic tribe and figure out what the rule was. Then they tried a
: >different version where they specifically clued in that social
: >CHEATING was going on. Subjects are much worse at the no-cheating
: >version than the cheating version by some enormous quantity.
: I don't pretend to be able to read minds. It is my experience that
: most people have trouble with abstract thinking, and the "correct"
: solution to this card problem requires some kind of abstract
: thinking.
Right, and what I am saying is that even when the problem is made less
abstract, it is very hard to solve when "social cheating" isn't
cued. This means that the abstract quality of the problem, although it
accounts for some of the performance difference, doesn't account for
most of it.
: The natural thing for a person to do, is to use trial and error. The
: limitation to only two trials is what makes this an artificial
: problem.
I think the reason it is tough is because people don't usually think
to check for the null hypothesis. It is, as you may know, NOT natural
to check the null hypothesis. Many experiments have shown people do
not do this and practice only makes things worse. Alison Gopnick
claims that this is because nature will hit the subject over the head
with a null hypothesis so you don't usually have to check for it
untill you're pushing the limits.
: >: You still have to ask what problem were the doctors solving. They
I don't think that "the doctor had a way of dealing with the
problem for which base rate information was not required." That is
where we disagree. The doctor had a way of dealing with it where the
base rate information was not used. Had it been used, the doctor would
have seen a better way of treating his or her patients. This is a good
example, I think, of (1) a very small difference between saying the
doctor was using the wrong heuristic for the right problem or the
right heuristic for the wrong problem.
You know, some of the procedure people undertook as a result
of BRN were very painful. Imagine a doctor prescribing chemotherapy as
a precaution against a small lump for a very rare cancer. I think
there is clearly an argument to be made that the doctor would have
done a better job in many cases if the risk of the patient was better
assessed.
: >I think T&K would agree that the brain does NOT structure data in such
: >a way that Bayesian reasoning can be used on it. But that is their
: >point. The world does afford situations where it is better, on
: >occasion to use Bayesian reasoning (for instance) and so, in those
: >situations, the brain is likely to provide sub-optimal solutions
: >unabetted by external aids.
: Ok. But I don't see anything irrational about using well tried and
: proven methods, just because those methods are occasionally
: suboptimal. I don't worry about whether the route I take, driving to
: work, is suboptimal. Why do we consider intellectual problems
: different than ordinary problems, and consider it irrational to use a
: suboptimal solution to a intellectual problem?
I think, somewhere along the line, "irrationality" has become
a very loaded word. I'm not sure if its because it gets bandied around
in debates or what, but it seems to have some proprietary meaning that
I am not privy to which sets of debates. If you are willing to
substitute "sub-optimal" for "irrational", then I have no problem. I
am not wedded to the word "irrational", and I don't know what it would
mean other than sub-optimal.
You know there are philosophical arguments that one has to
ascribe rationality in order to ascribe mental states at all. I wonder
if the two debates got tangled up or something.
Ciao
Omar
Neil Rickert
oha...@andromeda.rutgers.edu (Omar Haneef) writes:
> As of this response, I think the difference between our views
>-- steadily diminishing over the last few posts -- is negligible. I
>think we've expounded our positions.
>Neil Rickert (ric...@cs.niu.edu) wrote:
>: oha...@andromeda.rutgers.edu (Omar Haneef) writes:
>[snip]
>: >T&K think they are setting up problem P, for which solution S is
>: >optimal in world W arrived at by heuristic H (not the only one for S,
>: >but lets say its the normative process for arriving at S). Then you
>: >say the subject is actually working on problem P' for which S' is
>: >optimal using H'. However, the subject is in the same world W where S
>: >is optimal and the "real" problem is P.
>: I agree with everything except the notion that there is a single
>: "real" problem. I don't see any basis for this.
>[snip]
>: capacity. And I don't agree with the notion of a "wrong problem."
>: What is a right problem depends very much on contextual
>: considerations. And the context for the experimenter is very
>: different from the context for the subject.
>So here is another fundamental disagreement. You think that:
>1. It is not the heuristic that ought to be used to explain the
>difference in performance, but the target problem.
>2. You disagree that a problem can be criticized as wrong.
You cannot say a problem is wrong, unless you have standards for
deciding what makes a problem right or wrong.
Certainly, there are occasions when there can be standards. I don't
see that there were any that applied to the T&K work.
> These are beliefs that I myself adhere to. In GENERAL, they
>would not consitute an alternative explanation to T&K. T&K's work is
>supposed to be a CRITICISM of economists who think that utility curves
>and rational choices is the way to treat human beings. T&K would agree
>that we DO value things other than wealth, that we are involved in
>social relations. None of this is to say that, in spite of this
>awareness, they claim their models are the best match for their data.
I am certainly a critic of rational decision theories coming from
economists, and of politicians who espouse those theories.
Let me try to explain my position in a different way:
We can identify 3 kinds of problems:
(A) formal problems. A problem is presented in formal language,
or is described informally in a way that implies a specific
formalization. The answer is likewise expected to be
formal.
(B) natural language problems. The language is presented in
natural language sentences, and the answer is expected to be
given in natural language sentences.
(C) non-linguistic problems. The problem occurs in ordinary
real world experience. For example when crossing the street
you have to judge your timing so as to not be hit by a
passing car. We might say that the problem is specified in
terms of motions and shapes, and the answer is to be given
in terms of motions and shapes.
We might reasonably say that T&K presented type (B) problems. I
would claim that we evolved to solve type (C) problems. Kosslyn
might suggest that we have some kind of geometric representation
system for directly solving type (C) problems.
We can then ask the question of how we solve the various problems.
The assumptions of symbolic AI, of cognitivism, and of most
philosophers are that all problems are first converted to type (A)
problems, and then solved with formal methods.
I would argue that all problems are first converted to type (C)
problems, and solved in that way. But let me first explain more of
what I mean. When I am doing mathematics and similar type (A)
problems (my Ph.D. is in math, so I am tolerably good at it), then I
am not following Aristotle's rules of logic. Rather, I do logic in
terms of combinatorial experiments where I move imaginary objects
around and test cases. Learning mathematics entailed expanding my
abilities at type (C) problems to handle these combinatorial
experiments. Some mathematics students try to solve type (A)
problems purely within type (A). That is, those students memorize
rules, and attempt to apply them. In my experience, most students
who do mathematics in that way fail the class.
If a person solves type (B) problems by converting them to type (C)
problems, and if that person has no way of incorporating
linguistically presented base rate information into their system for
type (C) problems, then they are likely to ignore the base rate
information.
David Longley
> You cannot say a problem is wrong, unless you have standards for
> deciding what makes a problem right or wrong.
>
> Certainly, there are occasions when there can be standards.
> I don't see that there were any that applied to the T&K work.
^^^^^^^^^^^^^^^^
In that case, read (and study) the Dawes, Faust and Meehl SCIENCE
review (1989). It is all part of the same research tradition.
Jim Balter
CASEY DURFEE S wrote:
> : individual. Unfortunately, there can be no simple formal
> : procedure for assessing the compatibility of a set of
> : probability judgments with the judge's total system of
> : beliefs. The rational judge will nevertheless strive for
>
> There is a simple formal procedure: If you can make a Dutch
> Book against someone, their beliefs are inconsistent.
There may be a simple formal procedure for determining whether
someone's beliefs are inconsistent, but that doesn't describe one,
any more than "If you can prove that a given system can prove
its own Godel sentence, then the system is inconsistent" is a
simple formal procedure for showing that a system is inconsistent.
You would need a procedure for determining whether you can make
a Dutch Book against their beliefs.
--
<J Q B>
David Longley
Methodological Solipsist/
Folk Psychologist: "I don't believe this challenge to
common sense and rationality has any
good basis.
Methodological Behaviourist: "I understand that, a good degree in
psychology is probably a pre-requisite.
Methodological Solipsist/ "So you think you're better than the
Folk Psychologist: rest of us do you?"
Methodological Behaviourist: "No, I think I'm better at psychology
than those without a good degree in
the subject..."
Methodological Solipsist/ "But I see you making the same mistakes
Folk Psychologist: you say I and other folk make - you are
a hypocrite - I'm going to ignore you."
Methodological Behaviourist: "Would you refuse cancer-treatment from an
oncologist who has cancer?
David Longley
See Tversky and Kahneman's (1982) of the use of the term
"heuristic" and especially in the context of the earlier Popper
extract.
FRAGMENTS OF BEHAVIOUR:
EXTRACT 3: FROM 'A System Specification for PROfiling BEhaviour'
Full text is available at:
http://www.uni-hamburg.de/~kriminol/TS/tskr.htm
http://www.longley.demon.co.uk/Frag.htm
http://www.longley.demon.co.uk/Fragjn97.pdf
There is a logical possibility that in restricting the subject matter
of psychology, and thereby the deployment of psychologists, to what
can only be analysed and managed from a Methodological Solipsistic
(cognitive) perspective, one will render some very significant results
of research in psychology irrelevant to applied *behaviour* science
and technology, unless taken as a vindication of the stance that
behaviour is essentially context specific. As explicated above,
intensions are not, in principle, amenable to quantitative analysis.
They are, in all likelihood, only domain or context specific. A few
further examples should make these points clearer.
Many Cognitive Psychologists study 'Deductive Inference' from the
perspective of 'psychologism', a doctrine, which, loosely put,
equates the principles of logic with those of thinking. Yet the work
of Church (1936), Post (1936) and Turing (1937) clearly established
that the principles of 'effective' computation are not psychological,
and can in fact be mechanically implemented. However, researchers in
'Cognitive Science' such as Johnson-Laird and Byrne (1992) have
reviewed 'mental models' which provide an account for some of the
difficulties and some of the errors observed in human deductive
reasoning (Wason 1966). Throughout the 1970s, substantial empirical
evidence began to accumulate to refute the functionalist (Putnam 1967)
thesis that human cognitive processes were formal and computational.
Even well educated subjects it seems, have considerable difficulty
with relatively simple deductive Wason Selection tasks such as the
following:
_____ _____ _____ _____
| | | | | | | |
| A | | T | | 4 | | 7 |
|_____| |_____| |_____| |_____|
Where the task is to test the rule "if a card has a vowel on one side
it has an even number on the other".
Or in the following:
_____ _____ _____ _____
| | | | | | | |
| A | | 7 | | D | | 3 |
|_____| |_____| |_____| |_____|
where subjects are asked to test the rule 'each card that has an A on
one side will have a 3 on the other'. In both problems they can only
turn over a maximum of two cards to ascertain the truth of the rule.
Similarly, the majority have difficulty with the following,
similar problem, where the task is to reveal up to two hidden
halves of the cards to ascertain the truth or falsehood of the rule
'whenever there is a O on the left there is a O on the right':
_____________ _____________ ____________ ____________
| ||||||| | ||||||| ||||||| | ||||||| |
| O ||||||| | ||||||| ||||||| O | ||||||| |
|______||||||| |______||||||| |||||||______| |||||||______|
(a) (b) (c) (d)
Yet computer technology has no difficulty with these examples of the
application of basic deductive inference rules (modus ponens and modus
tollens). The above require the application of the material
conditional. [1] is falsified by turning cards A and 9, [2] by turning
cards A and 7, and [3] by turning cards (a) and (d). Logicians, and
others trained in the formal rules of deductive logic often fail to
solve such problems:
'Time after time our subjects fall into error. Even some
professional logicians have been known to err in an
embarrassing fashion, and only the rare individual takes
us by surprise and gets it right. It is impossible to
predict who he will be. This is all very puzzling....'
P. C. Wason and P. N. Johnson-Laird (1972)
Psychology of Reasoning
Furthermore, there is impressive empirical evidence that formal
training in logic does not generalise to such problems (Nisbett et al
1987). Yet why is this so if, in fact, human reasoning is, as the
cognitivists, have claimed, essentially logical and computational?
Wason (1966) also provided subjects with numbers which increased in
series, asking them to identify the rule. In most cases, the simple
fact that all examples shared no more than simple progression was
skipped, and whatever hypotheses they created were held onto even
though the actual rule was subsequently made clear. This persistence
of belief, and rationalisation of errors despite debriefing and
exposure to contrary evidence, is well documented in psychology, and
is a phenomenon which methodologically is, as Popper makes clear in
the leading quote to this paper, at odds with the formal advancement
of knowledge. Here is what Sir Karl Popper (1965) had to say about
this matter:
'My study of the CONTENT of a theory (or of any
statement whatsoever) was based on the simple and
obvious idea that the informative content of the
CONJUNCTION, ab, of any two statements, a, and b, will
always be greater than, or at least equal to, that of
its components.
Let a be the statement 'It will rain on Friday'; b the
statement 'It will be fine on Saturday'; and ab the
statement 'It will rain on Friday and it will be fine on
Saturday': it is then obvious that the informative
content of this last statement, the conjunction ab, will
exceed that of its component a and also that of its
component b. And it will also be obvious that the
probability of ab (or, what is the same, the probability
that ab will be true) will be smaller than that of
either of its components.
Writing Ct(a) for 'the content of the statement a', and
Ct(ab) for 'the content of the conjunction a and b', we
have
(1) Ct(a) <= Ct(ab) => Ct(b)
This contrasts with the corresponding law of the
calculus of probability,
(2) p(a) => p(ab) <= p(b)
where the inequality signs of (1) are inverted. Together
these two laws, (1) and (2), state that with increasing
content, probability decreases, and VICE VERSA; or in
other words, that content increases with increasing
IMprobability. (This analysis is of course in full
agreement with the general idea of the logical CONTENT
of a statement as the class of ALL THOSE STATEMENTS
WHICH ARE LOGICALLY ENTAILED by it. We may also say that
a statement a is logically stronger than a statement b
if its content is greater than that of b - that is to
say, if it entails more than b.)
This trivial fact has the following inescapable
consequences: if growth of knowledge means that we
operate with theories of increasing content, it must
also mean that we operate with theories of decreasing
probability (in the sense of the calculus of
probability). Thus if our aim is the advancement or
growth of knowledge, then a high probability (in the
sense of the calculus of probability) cannot possibly be
our aim as well: THESE TWO AIMS ARE INCOMPATIBLE.
I found this trivial though fundamental result about
thirty years ago, and I have been preaching it ever
since. Yet the prejudice that a high probability must be
something highly desirable is so deeply ingrained that
my trivial result is still held by many to be
'paradoxical'.
K. Popper
Truth, Rationality, and the Growth of Knowledge
Ch. 10, p 217-8
CONJECTURES AND REFUTATIONS (1965)
Modus tollens and the extensional principle that a compound event can
only be less probable (or equal) to its component events independently
is fundamental to the logic of scientific discovery, and yet this,
along with other principles of extensionality (deductive logic) seem
to be principles which are in considerable conflict with intuition, as
Kahneman and Tversky (1983) demonstrated with their illustration of
the 'Linda Problem'. In conclusion, the above authors wrote, twenty
years after Wason's experiments on deductive reasoning and Popper's
(1965) remarks on Conjectures and Refutation':
'In contrast to formal theories of belief, intuitive
judgments of probability are generally not extensional.
People do not normally analyse daily events into
exhaustive lists of possibilities or evaluate compound
probabilities by aggregating elementary ones. Instead,
they use a limited number of heuristics, such as
representativeness and availability (Kahneman et al.
1982). Our conception of judgmental heuristics is based
on NATURAL ASSESSMENTS that are routinely carried out as
part of the perception of events and the comprehension
of messages. Such natural assessments include
computations of similarity and representativeness,
attributions of causality, and evaluations of the
availability of associations and exemplars. These
assessments, we propose, are performed even in the
absence of a specific task set, although their results
are used to meet task demands as they arise. For
example, the mere mention of "horror movies" activates
instances of horror movies and evokes an assessment of
their availability. Similarly, the statement that Woody
Allen's aunt had hoped that he would be a dentist
elicits a comparison of the character to the stereotype
and an assessment of representativeness. It is
presumably the mismatch between Woody Allen's
personality and our stereotype of a dentist that makes
the thought mildly amusing.. Although these assessments
are not tied to the estimation of frequency or
probability, they are likely to play a dominant role
when such judgments are required. The availability of
horror movies may be used to answer the question "What
proportion of the movies produced last year were horror
movies?", and representativeness may control the
judgement that a particular boy is more likely to be an
actor than a dentist.
The term JUDGMENTAL HEURISTIC refers to a strategy -
whether deliberate or not - that relies on natural
assessment to produce an estimation or a prediction.
.Previous discussions or errors of judgement have
focused on deliberate strategies and on
misinterpretations of tasks. The present treatment calls
special attention to the processes of anchoring and
assimilation, which are often neither deliberate nor
conscious. An example from perception may be
instructive: If two objects in a picture of a three-
dimensional scene have the same picture size, the one
that appears more distant is not only seen as "really"
larger but also larger in the picture. The natural
computation of real size evidently influences the (less
natural) judgement of picture size, although observers
are unlikely to confuse the two values or to use the
former to estimate the latter.
The natural assessments of representativeness and
availability do not conform to the extensional logic of
probability theory.'
A. Tversky and D. Kahneman
Extensional Versus Intuitive Reasoning:
The Conjunction Fallacy in Probability Judgment.
Psychological Review Vol 90(4) 1983 p.294
The study of Natural Deduction (Gentzen 1935;Prawitz 1971; Tenant
1990) as a psychological process (1983) is really just the study of
the performance of a skill (like riding a bicycle in fact), which
attempts to account for why some of the difficulties with deduction
per se occur. The best models here may turn out to be connectionist,
where each individual's model ends up being almost unique in its fine
detail. There is a problem for performance theories, as Johnson Laird
and Byrne (1991) point out:
'A major difficulty for performance theories based on
formal logic is that people are affected by the content
of a deductive system..yet formal rules ought to apply
regardless of content. That is what they are: rules that
apply to the logical form of assertions, once it has
been abstracted from their content.'
P. N. Johnson-Laird and R. M. J. Byrne (1991)
Deduction p.31
The theme of this volume up to this point has been that methodological
solipsism is unlikely to reveal much more than the shortcomings and
diversity of social and personal judgment and the context specificity
of behaviour. It took until 1879 for Frege to discover the Predicate
Calculus (Quantification Theory), and a further half century before
Church (1936), Turing (1937) and others laid the foundations for
computer and cognitive science through their collective work on
recursive function theory. From empirical evidence, and developments
in technology, it looks like human and other animal reasoning is
primarily inductive and heuristic, not deductive and algorithmic.
Human beings have considerable difficulties with the latter, and this
is normal. It has taken considerable intellectual effort to discover
formal, abstract, extensional principles, often only with the support
of logic, mathematics and computer technology itself. The empirical
evidence, reviewed in this volume is that extensional principles are
not widely applied except in specific professional capacities which
are domain-specific. In fact, on the simple grounds that the discovery
of such principles required considerable effort should perhaps make us
more ready to accept that they are unlikely to be spontaneously
applied in everyday reasoning and problem solving.
For further coverage of the 'counter-intuitive' nature of deductive
reasoning (and therefore its low frequency in everyday practice) see
Sutherland's 1992 popular survey 'Irrationality', or Plous (1993) for
a recent review of the psychology of judgment and decision making. For
a thorough survey of the rise (and possibly the fall) of Cognitive
Science, see Putnam 1986, or Gardner 1987. The latter concluded his
survey of the Cognitive Revolution within psychology with a short
statement which he referred to as the 'computational paradox'. One
thing that Cognitive Science has shown us is that the computer or
Turing Machine is not a good model of how people reason, at least not
in the Von-Neumann Serial processing sense. Similarly, people do not
seem to think in accordance with the axioms of formal, extensional
logic. Instead, they learn rough and ready heuristics which they which
they try to apply to problems in a very rough, approximate way.
Accordingly, Cognitive Science may well turn to the work of Church,
Turing and other mathematical logicians who, in the wake of Frege,
have worked to elaborate what effective processing is. We will then be
faced with the strange situation of human psychology being of little
practical interest, except as a historical curiosity, an example of
pre-Fregian logic and pre-Church (1936) computation. Behaviour science
will pay as little attention to the 'thoughts and feelings' of 'folk
psychology' as contemporary physics does to quaint notions of 'folk
physics'. For some time, experimental psychologists working within the
information processing (computational) tradition have been working to
replace such concepts such as 'general reasoning capacity' with more
mechanistic notions such as 'Working Memory' (Baddeley 1986):
'This series of studies was concerned with determining
the relationship between general reasoning ability (R)
and general working-memory capacity (WM). In four
studies, with over 2000 subjects, using a variety of
tests to measure reasoning ability and working-memory
capacity, we have demonstrated a consistent and
remarkably high correlation between the two factors. Our
best estimates of the correlation between WM and R were
.82, .88., .80 and .82 for studies 1 through 4
respectively.
...
The finding of such a high correlation between these two
factors may surprise some. Reasoning and working-memory
capacity are thought of differently and they arise from
quite different traditions. Since Spearman (1923),
reasoning has been described as an abstract, high level
process, eluding precise definition. Development of good
tests of reasoning ability has been almost an art form,
owing more to empirical trial-and-error than to a
systematic delineation of the requirements such tests
must satisfy. In contrast, working memory has its roots
in the mechanistic, buffer-storage model of information
processing. Compared to reasoning, short-term storage
has been thought to be a more tractable, demarcated
process.'
P. C. Kyllonen & R. E. Christal (1990)
Reasoning Ability Is (Little More Than) Working-Memory
Capacity
Intelligence 14, 389-433
Such evidence stands well with the logical arguments of Cherniak which
were introduced in Section A, and which are implicit in the following
introductory remarks of Shinghal (1992) on automated reasoning:
'Suppose we are given the following four statements:
1. John awakens;
2. John brings a mop;
3. Mother is delighted, if John awakens and cleans his room;
4. If John brings a mop, then he cleans his room.
The statements being true, we can reason intuitively to
conclude that Mother is delighted. Thus we have deduced
a fact that was not explicitly given in the four
statements. But if we were given many statements, say a
hundred, then intuitive reasoning would be difficult.
Hence we wish to automate reasoning by formalizing it
and implementing it on a computer. It is then usually
called automated theorem proving. To understand
computer-implementable procedures for theorem proving,
one should first understand propositional and predicate
logics, for those logics form the basis of the theorem
proving procedures. It is assumed that you are familiar
with these logics.'
R. Shinghal (1992)
Formal Concepts in Artificial Intelligence: Fundamentals
Ch.2 Automated Reasoning with Propositional Logic p.8
Automated report writing and automated reasoning drawing on actuarial
data is fundamental to the PROBE project. In contrast to such work
using deductive inference, Gluck and Bower (1988) have modelled human
inductive reasoning using artificial neural network technology (which
are heuristic, based on constraint satisfaction/approximation, or
'best fit' rather than being 'production rule' based). That is, it is
unlikely that anyone spontaneously reasons using truth-tables or the
Resolution Rule (Robinson 1965). Furthermore, Rescorla (1988), perhaps
the dominant US spokesman for research in Pavlovian Conditioning, has
drawn attention to the fact that Classical Conditioning should perhaps
be seen as the experimental modelling of inductive inferential
'cognitive' heuristic processes. Throughout this paper, it is being
argued that such inductive inferences are in fact best modelled using
artificial neural network technology, and that such processing is
intensional, with all of the traditional problems of intensionality:
'Connectionist networks are well suited to everyday
common sense reasoning. Their ability to simultaneously
satisfy soft constraints allows them to select from
conflicting information in finding a plausible
interpretation of a situation. However, these networks
are poor at reasoning using the standard semantics of
classical logic, based on truth in all possible models.'
M. Derthick (1990)
Mundane Reasoning by Settling on a Plausible Model
Artificial Intelligence 46,1990,107-157
and perhaps even more familiarly:
'Induction should come with a government health warning.
A baby girl of sixteen months hears the word 'snow' used
to refer to snow. Over the next months, as Melissa
Bowerman has observed, the infant uses the word to refer
to: snow, the white tail of a horse, the white part of a
toy boat, a white flannel bed pad, and a puddle of milk
on the floor. She is forming the impression that 'snow'
refers to things that are white or to horizontal areas
of whiteness, and she will gradually refine her concept
so that it tallies with the adult one. The underlying
procedure is again inductive.'
P. N. Johnson-Laird (1988)
Induction, Concepts and Probability p.238: The Computer
and The Mind
Neil Rickert
Da...@longley.demon.co.uk (David Longley) writes:
>Methodological Solipsist/
> Folk Psychologist: "I don't believe this challenge to
> common sense and rationality has any
> good basis.
The so-called "methodological solipsist" understands that the success
of science depends on claims being subject to criticism and
challenge.
>Methodological Behaviourist: "I understand that, a good degree in
> psychology is probably a pre-requisite.
The behaviorist, by contrast, claims authority on the basis of
certification, and thus sees no need to defend his position and no
need for evidence. So it is `the behaviorist who is actually a
solipsist.
>Methodological Solipsist/ "So you think you're better than the
> Folk Psychologist: rest of us do you?"
>Methodological Behaviourist: "No, I think I'm better at psychology
> than those without a good degree in
> the subject..."
The behaviorist continues to insist that knowledge = certification.
>Methodological Solipsist/ "But I see you making the same mistakes
> Folk Psychologist: you say I and other folk make - you are
> a hypocrite - I'm going to ignore you."
>Methodological Behaviourist: "Would you refuse cancer-treatment from an
> oncologist who has cancer?
The behaviorist uses absurd and inappropriate analogies to dismiss
valid criticism.
Seth Russell
CASEY DURFEE S wrote:
> There is a simple formal procedure: If you can make a Dutch
> Book against someone, their beliefs are inconsistent.
What's a "Dutch Book" ?
--
Seth
See "Bozo's Conjecture" at http://www.clickshop.com/ai/conjecture.htm
And then on to the AI Jump List ...
Phil Roberts, Jr.
Seth Russell wrote:
>
> CASEY DURFEE S wrote:
>
> > There is a simple formal procedure: If you can make a Dutch
> > Book against someone, their beliefs are inconsistent.
>
> What's a "Dutch Book" ?
>
ditto
And while your at it, how about a little illumination on "hermeneutic".
--
Phil Roberts, Jr.
Feelings of Worthlessness and So-Called Cognitive Science
http://www.geocities.com/Athens/5476
Phil Roberts, Jr.
Phil Roberts, Jr. wrote:
>
> Seth Russell wrote:
> >
> > CASEY DURFEE S wrote:
> >
> > > There is a simple formal procedure: If you can make a Dutch
> > > Book against someone, their beliefs are inconsistent.
> >
> > What's a "Dutch Book" ?
> >
>
> ditto
>
> And while your at it, how about a little illumination on "hermeneutic".
>
Oops! I meant "heuristic". Sorry.
Sergio Navega
Wason's Selection Task is a typical example of one experiment
that can support two different conclusions. The act of taking
side with one of them and denying the value of the other is,
in my view, the fundamental problem of those who fail to absorb
good lessons from the things that do not conform with his
expectations.
For the sake of my argument here, I will repeat the test as
I know it and then retell both conclusions it raises:
A subject is presented to a set of 4 cards in which, it
is informed, each card have a letter in one side and
a number in the other. There's only one rule to be verified:
A card that have an 'A' in one side, must have a '4'
on the other. The question is: what are the cards
that must be turned over to see if this rule holds?
Here are the cards:
A B 4 7
Under casual inspection, the majority of the subjects
tested relate that he/she would turn the 'A' (correct)
and the card with '4' (irrelevant). Most fail to
see that the card with '7' should be turned over too,
because if it contains an 'A' it would be clearly a
contradiction with the furnished rule. Thus, the
conclusion is that the majority of the unprepared
subjects don't reason logically (in this case fail
to do modus tollens) and so human reasoning is,
normally, flawed.
This is conclusion ONE, and is where some behaviorists
stop.
But there's more:
Thagard (and others) proposes to continue with the
experiment in a different way. Suppose you're the
owner of an adult's bar where minors are not
allowed to get in. The cards have on one side
the age of a person and on the other the "status"
of that person (inside or outside the bar).
The question is mainly the same: what are
the cards we must turn over to see if a minor
is not disobeying the rule of being outside
the bar?
INSIDE OUTSIDE 25 16
It is clear ("logical") that we should turn
cards 'INSIDE' *and* card '16'. Almost everybody
find this test easy. Almost everybody get it right.
Nevertheless, this is EXACTLY the same test as
before, with the only difference of being
inside a *context*, where the mechanism in use
may be not logical inferences but another
(Jonhson-Laird's mental models, Cheng & Holyoak
pragmatic reasoning schemas, etc).
This is conclusion TWO, made by "folk psychologists".
So, which one is right? Which one we should
choose and which should be dismissed?
My conclusions:
a) Human mind does not work using logical principles.
Logical reasoning is *simulated* using other
mechanisms.
b) There is great benefit in reeducating people
to use critical thinking. Modus tollens, in
particular, is often neglected. AI systems
should be designed with this in mind, too.
c) If AI is to be useful for mankind, it
MUST replicate some of our reasoning methods,
*including* those that are said to be "flawed".
d) Evolution could have done better for
our minds. Or not? Maybe the way we reason
today is the *most adequate* to handle our
complex and dynamic world.
e) I believe (nasty word, huh? :-) that
AI is not successful today because it is
using the methods that explain the first
conclusion. The "frame problem" that
scares GOFAI cannot be solved unless we
use "soft", non-logical methods (I'm working
on this).
f) Fluid, inexact, vague, relative, uncertain
forms of reasoning are where we excel. AI should
pursue the same route too, with the added advantage
of clear-cut logic and fast arithmetic ability. AI made
with these principles will be MUCH better than us.
What is funny is that my conclusions here don't demand
that I choose one of the two options. Coexistence in
our mind of antagonist viewpoints may help a lot in
our search for the truth.
Regards,
Sergio Navega.
David Longley
> Da...@longley.demon.co.uk (David Longley) writes:
>
> >Methodological Solipsist/
> > Folk Psychologist: "I don't believe this challenge to
> > common sense and rationality has any
> > good basis.
>
> The so-called "methodological solipsist" understands that the success
> of science depends on claims being subject to criticism and
> challenge.
>
> >Methodological Behaviourist: "I understand that, a good degree in
> > psychology is probably a pre-requisite.
>
> The behaviorist, by contrast, claims authority on the basis of
> certification, and thus sees no need to defend his position and no
> need for evidence. So it is `the behaviorist who is actually a
> solipsist.
>
> >Methodological Solipsist/ "So you think you're better than the
> > Folk Psychologist: rest of us do you?"
>
> >Methodological Behaviourist: "No, I think I'm better at psychology
> > than those without a good degree in
> > the subject..."
>
> The behaviorist continues to insist that knowledge = certification.
You have it the wrong way round, and demonstrably so as I show
below. The methodological solipsist doesn't basically care about
empirical evidence, because he is concerned about how the world
seems to him. He is only interested in the processes, not the
can be seen as a final criticism of such phenomenalist
empiricism.
'Explores the distinction between 2 doctrines, both of
which inform theory construction in much of modern
cognitive psychology: the representational theory of
mind and the computational theory of mind. According to
the former, propositional attitudes are viewed as
relations that organisms bear to mental representations.
According to the latter, mental processes have access
only to formal (nonsemantic) properties of the mental
representations over which they are defined. The
following claims are defended: (1) The traditional
dispute between rational and naturalistic psychology is
plausibly viewed as an argument about the status of the
computational theory of mind. (2) To accept the
formality condition is to endorse a version of
methodological solipsism. (3) The acceptance of some
such condition is warranted, at least for that part of
psychology that concerns itself with theories of the
mental causation of behavior. A glossary and several
commentaries are included.'
J A Fodor (1980)
Methodological solipsism considered as a research
strategy in cognitive psychology.
Massachusetts Inst of Technology
Behavioral and Brain Sciences; 1980 Mar Vol 3(1) 63-109
'If psychological explanation is a matter of describing
computational processes, then the references of our
thoughts do not matter to psychological explanation.
This is Fodor's main argument.....Notice that Fodor's
argument can be taken a step further. For not only are
the references of our thoughts not mentioned in
cognitive psychology; nothing that DETERMINES their
references, like Fregian senses, is mentioned
either....Neither reference nor reference-determining
sense have a place in the description of computational
processes.'
B. F. Loar
Ibid p.89
'Fodor thinks that when we explain behaviour by mental
causes, these causes would be given "opaque"
descriptions "true in virtue of the way the agent
represents the objects of his wants (intentions,
beliefs, etc.) to HIMSELF" (his emphasis). But what an
agent intends may be widely different from the way he
represents the object of his intention to himself. A man
cannot shuck off the responsibility for killing another
man by just 'directing his intention' at the firing of a
gun:
"I press a trigger - Well, I'm blessed!
he's hit my bullet with his chest!"'
P. Geach
ibid p80
Here's how Fodor contrasted Methodological Solipsism with the
naturalistic approach:
'..there's a tradition which argues that - epistemology
to one side - it is at best a strategic mistake to
attempt to develop a psychology which individuates
mental states without reference to their environmental
causes and effects...I have in mind the tradition which
includes the American Naturalists (notably Pierce and
Dewey), all the learning theorists, and such
contemporary representatives as Quine in philosophy and
Gibson in psychology. The recurrent theme here is that
psychology is a branch of biology, hence that one must
view the organism as embedded in a physical environment.
The psychologist's job is to trace those
organism/environment interactions which constitute its
behavior.'
J. Fodor (1980) ibid. p.64
>
> >Methodological Solipsist/ "But I see you making the same mistakes
> > Folk Psychologist: you say I and other folk make - you are
> > a hypocrite - I'm going to ignore you."
>
> >Methodological Behaviourist: "Would you refuse cancer-treatment from an
> > oncologist who has cancer?
>
> The behaviorist uses absurd and inappropriate analogies to dismiss
> valid criticism.
>
No, the behaviourist isn't fundamentally interested in a priori,
purely philosophical or even mathematical criticism, or debate
over the meaning of words on a priori grounds. The tribunal is
experimental evidence.
You still have the two positions the wrong way round - and that's
NOT a matter of interpretation. What you are defending is pre-
Quinean positivism.
I suspect this is because mathematicians *are* methodological
solipsists, which is why it was so attractive to Husserl and the
other phenomenologists. He thought it was a sound basis for
empiricism, and Carnap tried it in his "Aufbau" (1928). From
primities he used the technology of logic and mathematics as
developed in Principia Mathematica to build *an* internally
consistent system to show how it *could* be done. He failed. The
alternative is universally understood to be Quine's external
empiricism.
It is *you* who uses absurd and inappropriate analogies to
*evidence*. You don't even write as a scientist, you write as if
it can all be achieved a priori. You should look to the evidence,
the research literature I keep pointing you towards.
David Longley
In article <359BBC...@ix.netcom.com>
phi...@ix.netcom.com "Phil Roberts, Jr." writes:
> Seth Russell wrote:
> >
> > CASEY DURFEE S wrote:
> >
> > > There is a simple formal procedure: If you can make a Dutch
> > > Book against someone, their beliefs are inconsistent.
> >
> > What's a "Dutch Book" ?
> >
>
> ditto
>
> And while your at it, how about a little illumination on "hermeneutic".
>
> --
>
> Phil Roberts, Jr.
>
> Feelings of Worthlessness and So-Called Cognitive Science
> http://www.geocities.com/Athens/5476
>
The following is an extract from a paper by Kyburg (1968) which
gives a rather good account of what the Dutch Book Theorem is.
However, I strongly advise those who are unfamiliar with the
literature in this field to do a considerable amount of reading
around this subject before drawing and strong conclusions.
I also suggest they read the extract I posted a few days back
from Davidson.
o o o
'Ramsey proves as "fundamental laws of probable belief":
1. The degree of belief in P, plus the degree of belief
in not-P equals 1.
2. The degree of belief in P given Q, plus the degree of
belief in not-P given Q equals 1.
3. The degree of belief in P and Q equals the degree of
belief in P multiplied by the degree of belief in Q
given P.
2 Ramsey actually employs seven other axioms; these
concern only requirements of consistency in assigning
values to total states of affairs.
4. The degree of belief in P and Q plus the degree of
belief in P and not-Q equals the degree of belief in P.
Ramsey writes:
These are the laws ot probability, which we
have proved to be necessarily true of any
consistent set of degrees of belief Any
definite set of degrees of beIief which broke
them would be inconsistent in the sense that
it violated the laws of preference between
options, such as that preferability is a
transitive asymmetrical relation, and that if
alpha is preferable to beta, beta for certain
cannot be preferable to alpha if p, beta if
not-p If anyone's mental condition violated
these laws, his choice would depend on the
precise form in which the options were offered
him, which is absurd He could have a book made
against him by a cunning bettor, and would
then stand to lose in any event. (p. 182) IP
36, this volume]
This has been christened the Dutch Book Theorem, by
Isaac Levi (Paper read at the 1965 meeting of the
American Philosophical Association, Eastern Division)
This is as far as the formal logic of subjective
probability takes us. From the point of view of the
"logic of consistency" (which for Ramsey includes the
probability calculus), no set of beliefs is more
rational than any other, so long as they both satisfy
the quantitative relationships expressed by the
fundamental laws of probability. Thus I am free to
assign the number l/3 to the probability that the sun
will rise tomorrow; or, more cheerfully, to take the
probability to be 9/10 that I have a rich uncle in
Australia who will send me a telegram tomorrow informing
me that he has made me his sole heir. Neither Ramsey,
nor Savage, nor de Finetti, to name three leading
figures in the personalistic movement, can find it in
his heart to detect any logical shortcomings in anyone,
or to find anyone logically culpable, whose degrees of
belief in various propositions satisfy the laws of the
probability calculus, however odd those degrees of
belief may otherwise be. Reasonableness, in which Ramsey
was also much interested, he considered quite another
matter.4 The connection between rationality (in the
sense of conformity to the rules of the probability
calculus) and reasonableness (in the ordinary inductive
sense) is much closer for Savage and de Finetti than it
was for Ramsey, but it is still not a strict connection;
one can still be wildly unreasonable without sinning
against either logic or probability.
Now this seems patently absurd. It is to suppose that
even the most simple statistical inferences have no
logical weight where my beliefs are concerned.....
..... '
H E Kyburg (1968)
Bets and beliefs
American Philosophical Quarterly, 5 , pp63-78
Hermeneutics has its origin in the interpretation of texts, where
the assumption was that misunderstanding would be the norm... In
this century it has been most closely associated with existential
phenomenology (cf Gadamer 1976 "Philosophical Hermeneutics".
For most purposes here - see Methodological Solipsism.
David Longley
In article <359bc...@news2.ibm.net> sna...@ibm.net "Sergio Navega" writes:
>
> So, which one is right? Which one we should
> choose and which should be dismissed?
>
I cover this in Fragments, and posted it again a couple of days
ago. Both are right, what the research shows is that human
reasoning is CONTENT bound, it is shaped by context and is
exemplar or stereotype driven.
The second example does NOT contradict the first. The point is
that outside of familiar contexts people do not reason reliably.
They attribute, or use heuristics, a priori schemas.
Have a look at section 3 of "Fragments".
Neil Rickert
>> Da...@longley.demon.co.uk (David Longley) writes:
No, that is nonsense. The methodological solipsist is sampling the
evidence with perception. But between each such sampling, he only
considers internal representations. The idea that he altogether
ignores empirical evidence is complete nonsense.
We do have on the word of David Longley (quoted above), that the
behaviorist only cares about his certification. So the behaviorist
is really a solipsist.
Neil Rickert
"Sergio Navega" <sna...@ibm.net> writes:
>a) Human mind does not work using logical principles.
>Logical reasoning is *simulated* using other
>mechanisms.
I agree with this. We need to be clear that "does not work using
logical principles" carries no implication of being illogical.
Rather, one might call it alogical or extra-logical.
>c) If AI is to be useful for mankind, it
>MUST replicate some of our reasoning methods,
>*including* those that are said to be "flawed".
I suppose this depends on the ways in which we want to use AI.
>e) I believe (nasty word, huh? :-) that
>AI is not successful today because it is
>using the methods that explain the first
>conclusion. The "frame problem" that
>scares GOFAI cannot be solved unless we
>use "soft", non-logical methods (I'm working
>on this).
I would say that the frame problem is not much of a problem for
humans. I don't mean that our brains have solved the problem.
Rather, what are brains are doing is such that instances of the frame
problem are relatively infrequent and usually unimportant.
Sergio Navega
David Longley wrote in message <899405...@longley.demon.co.uk>...
>In article <359bc...@news2.ibm.net> sna...@ibm.net "Sergio Navega"
writes:
>>
>> So, which one is right? Which one we should
>> choose and which should be dismissed?
>>
>
>I cover this in Fragments, and posted it again a couple of days
>ago. Both are right, what the research shows is that human
>reasoning is CONTENT bound, it is shaped by context and is
>exemplar or stereotype driven.
>
>The second example does NOT contradict the first. The point is
>that outside of familiar contexts people do not reason reliably.
>They attribute, or use heuristics, a priori schemas.
>
Well, this is a good reason for us to investigate seriously how
this is supposed to work in humans. It is not to say that AI have "failed"
(some systems produced so far are really impressive), it is that
AI does not "understand" our strange way of thinking. I see no other
way for AI to succeed (for us, humans) than to put a little of our
"flawed" way or reasoning into their silicon brains.
Regards,
Sergio Navega.
Sergio Navega
Neil Rickert wrote in message <6ngrud$k...@ux.cs.niu.edu>...
>"Sergio Navega" <sna...@ibm.net> writes:
>
>>a) Human mind does not work using logical principles.
>>Logical reasoning is *simulated* using other
>>mechanisms.
>
>I agree with this. We need to be clear that "does not work using
>logical principles" carries no implication of being illogical.
>Rather, one might call it alogical or extra-logical.
>
Good point. Today's machines are logical and that's not enough.
We don't want one illogical machine, we want that extra-logical.
>>e) I believe (nasty word, huh? :-) that
>>AI is not successful today because it is
>>using the methods that explain the first
>>conclusion. The "frame problem" that
>>scares GOFAI cannot be solved unless we
>>use "soft", non-logical methods (I'm working
>>on this).
>
>I would say that the frame problem is not much of a problem for
>humans. I don't mean that our brains have solved the problem.
>Rather, what are brains are doing is such that instances of the frame
>problem are relatively infrequent and usually unimportant.
>
I would add that even if we find something unusual, our brain seems
to find a way to solve this new problem, on the majority of the
circumstances. But we are limited, our neural mechanism cannot be
augmented (at least with current technology). AI, on the other
hand, and once it reasons the way we do, can be expanded easily.
This is what I think is the most amazing thing that awaits us
in the next century: the possibility of going beyond human
thinking.
Regards,
Sergio Navega.
David Longley
> No, that is nonsense.
Read what Geach says.....then think about it - don't just reject
that it is true of *your* thinking just because it is being so
effectively lampooned.
'Fodor thinks that when we explain behaviour by mental
causes, these causes would be given "opaque"
descriptions "true in virtue of the way the agent
represents the objects of his wants (intentions,
beliefs, etc.) to HIMSELF" (his emphasis). But what an
agent intends may be widely different from the way he
represents the object of his intention to himself. A man
cannot shuck off the responsibility for killing another
man by just 'directing his intention' at the firing of a
gun:
"I press a trigger - Well, I'm blessed!
he's hit my bullet with his chest!"'
P. Geach
ibid p80
'..there's a tradition which argues that - epistemology
to one side - it is at best a strategic mistake to
attempt to develop a psychology which individuates
mental states without reference to their environmental
causes and effects...I have in mind the tradition which
includes the American Naturalists (notably Pierce and
Dewey), all the learning theorists, and such
contemporary representatives as Quine in philosophy and
Gibson in psychology. The recurrent theme here is that
psychology is a branch of biology, hence that one must
view the organism as embedded in a physical environment.
The psychologist's job is to trace those
organism/environment interactions which constitute its
behavior.'
J. Fodor (1980) ibid. p.64
> We do have on the word of David Longley (quoted above), that the
> behaviorist only cares about his certification.
There's NO quotation.... They're your words not mine.
Neil Rickert
Da...@longley.demon.co.uk (David Longley) writes:
>In article <6ngqma$k...@ux.cs.niu.edu> ric...@cs.niu.edu "Neil Rickert" writes:
>> >> The behaviorist continues to insist that knowledge = certification.
>> >You have it the wrong way round, and demonstrably so as I show
>> >below. The methodological solipsist doesn't basically care about
>> >empirical evidence, because he is concerned about how the world
>> >seems to him.
>> No, that is nonsense.
>Read what Geach says.....then think about it - don't just reject
>that it is true of *your* thinking just because it is being so
>effectively lampooned.
But your quote from Geach doesn't apply to my thinking at all. It
may be a good description of Longley's thinking about what he takes
to be my thinking.
> .... The recurrent theme here is that
> psychology is a branch of biology, hence that one must
> view the organism as embedded in a physical environment.
> The psychologist's job is to trace those
> organism/environment interactions which constitute its
> behavior.'
> J. Fodor (1980) ibid. p.64
Now there is a description that fits my view. It is what Fodor gives
as the alternative to methodological solipsism.
>> We do have on the word of David Longley (quoted above), that the
>> behaviorist only cares about his certification.
>There's NO quotation.... They're your words not mine.
>> >> >Methodological Behaviourist: "No, I think I'm better at psychology
>> >> > than those without a good degree in
>> >> > the subject..."
That is quoted from Longley. I read it as a clear claim for the
authority of certification. Others can read it as they wish.
Neil Rickert
"Sergio Navega" <sna...@ibm.net> writes:
>>I would say that the frame problem is not much of a problem for
>>humans. I don't mean that our brains have solved the problem.
>>Rather, what are brains are doing is such that instances of the frame
>>problem are relatively infrequent and usually unimportant.
>I would add that even if we find something unusual, our brain seems
>to find a way to solve this new problem, on the majority of the
>circumstances.
I'm driving to work. Half way there, I begin to worry about
whether I remembered to close the garage door.
A person calls the host to decline an invitation to a party. The
surprised host mentions that the person had already declined
yesterday.
These sorts of thing seem to be examples of frame problem failures.
I don't think we are particularly good at solving it. We do pretty
well on arranging our affairs so that these problems don't often
arise.
Phil Roberts, Jr.
Neil Rickert wrote:
>
> Da...@longley.demon.co.uk (David Longley) writes:
>
> >> >> >Methodological Behaviourist: "No, I think I'm better at psychology
> >> >> > than those without a good degree in
> >> >> > the subject..."
>
Philosophy of science community: "SHOW ME THE BEEF" (Theoretical Anomalies)
>
> That is quoted from Longley. I read it as a clear claim for the
> authority of certification. Others can read it as they wish.
I read it as a claim that a science of psychology actually exists.
Even Dennett has the good sense to put 'science' in quotes when
talking about cognitive "science". One of my heroes, Thomas Hardy
Leahey ('A History of Psychology') would, no doubt, agree:
Psychology seems, therefore, to be a science of perpetual crisis.
It has never been able to get past what Kuhn calls the pre-paradigm
phase of science. Psychologists have never ceased to debate
basic issues about the nature, goals, methods and definitions of
their field. Only with the schools has there ever been much
agreement on these problems, and even the schools usually
contain warring factions. (Thomas Hardy Leahey)
...behaviorism is irrefutable; it is excessively technological;
it has no place for meaning; its problems are all borrowed, for
it reacts to questions set by mentalistic psychology but then
reduces them to S-R temrs, being unable itself to pose creative
puzzles (Leahey)
Koch argues that the label "science" functions as a kind of
security blanket, desperately clutched as a talisman agains
doubt. (Leahey)
David Longley
> Da...@longley.demon.co.uk (David Longley) writes:
> >In article <6ngqma$k...@ux.cs.niu.edu> ric...@cs.niu.edu "Neil Rickert" writes:>
> >> >> The behaviorist continues to insist that knowledge = certification.
>
> >> >You have it the wrong way round, and demonstrably so as I show
> >> >below. The methodological solipsist doesn't basically care about
> >> >empirical evidence, because he is concerned about how the world
> >> >seems to him.
>
> >> No, that is nonsense.
>
> >Read what Geach says.....then think about it - don't just reject
> >that it is true of *your* thinking just because it is being so
> >effectively lampooned.
>
> But your quote from Geach doesn't apply to my thinking at all. It
> may be a good description of Longley's thinking about what he takes
> to be my thinking.
>
> > .... The recurrent theme here is that
> > psychology is a branch of biology, hence that one must
> > view the organism as embedded in a physical environment.
> > The psychologist's job is to trace those
> > organism/environment interactions which constitute its
> > behavior.'
>
> > J. Fodor (1980) ibid. p.64
>
> Now there is a description that fits my view. It is what Fodor gives
> as the alternative to methodological solipsism.
Which is behaviourism - the very position which "fragments"
emphasises. Now you may well have come round to this position
over the past three years, but in he past you've been as
dismissive oif that as you have almost everything else. And with
no good basis.
>
> >> We do have on the word of David Longley (quoted above), that the
> >> behaviorist only cares about his certification.
>
> >There's NO quotation.... They're your words not mine.
>
> >> >> >Methodological Behaviourist: "No, I think I'm better at psychology
> >> >> > than those without a good degree in
> >> >> > the subject..."
>
> That is quoted from Longley. I read it as a clear claim for the
> authority of certification. Others can read it as they wish.
>
Now that *is* a quotation, but what you say can not be
substituted for what is written. 1) You have left out the context
and secondly it is not the same (in anyone's book) as what you
have written.
The whole problem of psychological contexts is that you can not
substitute in this sort of way and draw valid inferences. I keep
drawing your attention to the fact that in trying to do so, you
fabricate, or attribute - and they are *your* attributions and
fabrications. To continue drawing conclusions on the basis of
these bizarre constructions and claim yet that they belong in the
objective world is, in your case, delusional. It's a
consequence of your methodological solipsism that you don't make
a distinction between what you believe and what is the case.
Asserting that this is not so, does not make it not so. And to
claim that I am advocating methodological solipsism when the
whole of "Fragments" is written contrasting its deficiencies with
respect to the Quinean alternative jut compounds that delusion
and reduces you to having to post desperate nefarious rhetoric.
Spend some time thinking this through rather than dashing off any
more inane "rebuttals".
David Longley
What's *annoying* about so many of your posts is that you don't
cite your sources. All these "I thinks" might lead the unwary to
believe that these are your insights, rather than insights you
have found attractive.
Is this just one more example of your lack of insight?
Neil Rickert
Da...@longley.demon.co.uk (David Longley) writes:
>In article <6ngvlr$k...@ux.cs.niu.edu> ric...@cs.niu.edu "Neil Rickert" writes:
>> > .... The recurrent theme here is that
>> > psychology is a branch of biology, hence that one must
>> > view the organism as embedded in a physical environment.
>> > The psychologist's job is to trace those
>> > organism/environment interactions which constitute its
>> > behavior.'
>> > J. Fodor (1980) ibid. p.64
>> Now there is a description that fits my view. It is what Fodor gives
>> as the alternative to methodological solipsism.
>Which is behaviourism - the very position which "fragments"
>emphasises. Now you may well have come round to this position
>over the past three years, but in he past you've been as
>dismissive oif that as you have almost everything else. And with
>no good basis.
Oh, bull. I have held that position since long before I ever had the
misfortune to encounter Longley. I am far more of a behaviorist than
is Longley, but he is too stupid to be able to tell. What I
criticise of Longley's position is not that it is behaviorist, but
that it an absurd and extreme version of behaviorism.
Neil Rickert
Da...@longley.demon.co.uk (David Longley) writes:
>In article <6nh1if$k...@ux.cs.niu.edu> ric...@cs.niu.edu "Neil Rickert" writes:
>> I'm driving to work. Half way there, I begin to worry about
>> whether I remembered to close the garage door.
>> A person calls the host to decline an invitation to a party. The
>> surprised host mentions that the person had already declined
>> yesterday.
>> These sorts of thing seem to be examples of frame problem failures.
>> I don't think we are particularly good at solving it. We do pretty
>> well on arranging our affairs so that these problems don't often
>> arise.
>What's *annoying* about so many of your posts is that you don't
>cite your sources. All these "I thinks" might lead the unwary to
>believe that these are your insights, rather than insights you
>have found attractive.
What is irritating about Longley, is that after several years
involvement with c.a.p. he still has not discovered that it is not a
research journal, and that citation is never required and often
inappropriate.
lsa...@aol.com
In article <899405...@longley.demon.co.uk>,
Da...@longley.demon.co.uk# wrote:
>
> In article <359bc...@news2.ibm.net> sna...@ibm.net "Sergio Navega" writes:
> >
> > So, which one is right? Which one we should
> > choose and which should be dismissed?
> >
>
> I cover this in Fragments, and posted it again a couple of days
> ago. Both are right, what the research shows is that human
> reasoning is CONTENT bound, it is shaped by context and is
> exemplar or stereotype driven.
>
> The second example does NOT contradict the first. The point is
> that outside of familiar contexts people do not reason reliably.
> They attribute, or use heuristics, a priori schemas.
It's like saying that if you don't know something, you don't know it.
What a revolutionary insight! Not! Faced with unfamiliar contexts, we
attribute the best we can. Brittle extensional systems would just
stop working altogether.
Louis Savain
-----== Posted via Deja News, The Leader in Internet Discussion ==-----
http://www.dejanews.com/rg_mkgrp.xp Create Your Own Free Member Forum
lsa...@aol.com
In article <359bc...@news2.ibm.net>,
"Sergio Navega" <sna...@ibm.net> wrote:
>
>[...]
> My conclusions:
>
> a) Human mind does not work using logical principles.
> Logical reasoning is *simulated* using other
> mechanisms.
I agree with this. Humans discover logic through the
process of inducing causal correlations and through
internal and sensory feedback. IOW, logic emerges
from causality. Our coming adaptive intelligent machines
will work on the same principle.
> b) There is great benefit in reeducating people
> to use critical thinking. Modus tollens, in
> particular, is often neglected. AI systems
> should be designed with this in mind, too.
There is no need to specifically design AI systems
with the ability to use "critical thinking" from the
start. If the system is designed to use causality or
contingency-based covariation, it too, can be trained
to use "formal" methods and rules.
> c) If AI is to be useful for mankind, it
> MUST replicate some of our reasoning methods,
> *including* those that are said to be "flawed".
More often than not, the "flaw" is nothing worse
than a lack of knowledge. All knowledge systems
suffer from this, including so-called "extensional"
systems.
> d) Evolution could have done better for
> our minds. Or not? Maybe the way we reason
> today is the *most adequate* to handle our
> complex and dynamic world.
I agree with you that we are optimally designed by
nature to handle change and complexity. Change is
just another way of saying cause and effect. IMO,
we learn everything we know about the world through
trial and error, by generating expectations
regarding how sensory events change with respect
to one another. This is forcibly a statistical,
covariational process.
> e) I believe (nasty word, huh? :-) that
> AI is not successful today because it is
> using the methods that explain the first
> conclusion. The "frame problem" that
> scares GOFAI cannot be solved unless we
> use "soft", non-logical methods (I'm working
> on this).
Glad to see you're working on it. What is your
take on it so far?
> f) Fluid, inexact, vague, relative, uncertain
> forms of reasoning are where we excel.
This requires the use of statistical methods. It
is an inductive process as is any type of sensory
learning. Having said that, there is no reason why
the same mechanism cannot handle discreteness and
certainty. The fact that we can recognize the
difference between the two, suggests that we can
learn to handle both.
> AI should
> pursue the same route too, with the added advantage
> of clear-cut logic and fast arithmetic ability. AI made
> with these principles will be MUCH better than us.
I have to partially disagree with this. Logic and
arithmetic are not and should not be a fundamental part
of intelligence. They must be learned by the intelligent
entity just like everything else. However, there is good
reason to suppose that an intelligent machine endowed
with a direct sensory link to a logic/arithmetic module,
will be better at logical tasks than it would be
otherwise. But in either case, logic and arithmetic
must be learned by the machine. The learning part is
the intelligent part. Not the logic module.
> What is funny is that my conclusions here don't demand
> that I choose one of the two options. Coexistence in
> our mind of antagonist viewpoints may help a lot in
> our search for the truth.
I'm not so sure I can agree with that. One cannot
logically accept truth and its opposite.
Regards,
David Longley
In article <359C1B...@ix.netcom.com>
phi...@ix.netcom.com "Phil Roberts, Jr." writes:
> Neil Rickert wrote:
> >
> > Da...@longley.demon.co.uk (David Longley) writes:
>
> >
> > >> >> >Methodological Behaviourist: "No, I think I'm better at psychology
> > >> >> > than those without a good degree in
> > >> >> > the subject..."
> >
>
> Philosophy of science community: "SHOW ME THE BEEF" (Theoretical Anomalies)
>
> >
> > That is quoted from Longley. I read it as a clear claim for the
> > authority of certification. Others can read it as they wish.
>
> I read it as a claim that a science of psychology actually exists.
> Even Dennett has the good sense to put 'science' in quotes when
> talking about cognitive "science". One of my heroes, Thomas Hardy
> Leahey ('A History of Psychology') would, no doubt, agree:
Until you learn to face a few real world facts you will continue
to be impressed and distracted by rhetoric, and you won't be able
to write about anything of consequence.
Look to the context of "A System Specification for PROfiling
BEhaviour" (1994) and think about the alternatives discussed in
the early sections and why they are contrasted.
Most people only come to fully appreciate the significance of
many academic arguments in psychology once they have begun work
in applied fields. Until then, much of what they know is truly
"academic" and inconsequential - and they choose sides almost
like choosing a football team...I think that's all you are doing.
Take that as well intended criticism (and don't respond with a
backlash).
Phil Roberts, Jr.
Neil Rickert wrote:
> ...... I am far more of a behaviorist than
> is Longley.....
why?
Bill Taylor
Da...@longley.demon.co.uk (posing as a human) writes:
|> What's *annoying*
Bots can't get annoyed.
|> about so many of your posts is that you don't cite your sources.
He's interested in bandwidth and readability. Unlike bots.
|> All these "I thinks" might lead the unwary to
|> believe that these are your insights, rather than insights you
|> have found attractive.
Gotta be human before you can
(a) have insights
(b) find anything attractive.
|> Is this just one more example of your lack of insight?
Now now - no cheating! Do your own dirty work. If you want to learn
something about insight you've gotta develope some of your own.
Ask your handlers about it...
Bill Taylor
Da...@longley.demon.co.uk (or his handlers) writes:
|> Which is behaviourism - the very position which "fragments"
Well it's such a long-winded wittering rambnle, it's bound to allude
to almost anything.
|> Now that *is* a quotation, but what you say can not be
|> substituted for what is written.
If only you could substitute some thought for some quotation! Sigh...
|> I keep drawing your attention to the fact that in trying to do so,
If only we could all draw YOUR attention to the fact that NO-ONE here is
interested in what you have to say, no-one understands you, no-one finds you
the least bit enlightening, and everyone wishes you would piss off
back to the prisons where you so obviously belong...
David Longley
In article <6nhpd4$hhu$9...@cantuc.canterbury.ac.nz>
mat...@math.canterbury.ac.nz "Bill Taylor" writes:
> Da...@longley.demon.co.uk (or his handlers) writes:
>
> |> Which is behaviourism - the very position which "fragments"
>
> Well it's such a long-winded wittering rambnle, it's bound to allude
> to almost anything.
That's your uneducated view - it's obviously well beyond your
comprehension.
>
> |> Now that *is* a quotation, but what you say can not be
> |> substituted for what is written.
>
> If only you could substitute some thought for some quotation! Sigh...
There are good reasons for using quotation rather than "thought"
- as I explain in "Fragments".
>
>
> |> I keep drawing your attention to the fact that in trying to do so,
>
> If only we could all draw YOUR attention to the fact that NO-ONE here is
> interested in what you have to say, no-one understands you, no-one finds you
> the least bit enlightening, and everyone wishes you would piss off
> back to the prisons where you so obviously belong...
>
Well if you did some of the reading I have recommended you might
understand what is being said, and you might be enlightened. BUt
like several others in this group, you just won't be told
anything.
David Longley
In article <6nhpd4$hhu$9...@cantuc.canterbury.ac.nz>
mat...@math.canterbury.ac.nz "Bill Taylor" writes:
>
> If only we could all draw YOUR attention to the fact that NO-ONE here is
> interested in what you have to say, no-one understands you, no-one finds you
> the least bit enlightening, and everyone wishes you would piss off
> back to the prisons where you so obviously belong...
>
Ok.. let's see whether this is true:
If you agree with the general with the remarks made above send an
e-mail to me with "AGREE" in the body of the text. If you don't
agree, write "DO NOT AGREE".
Copy the e-mail to Taylor.
David Longley
> In article <899405...@longley.demon.co.uk>,
> Da...@longley.demon.co.uk# wrote:
> >
> > In article <359bc...@news2.ibm.net> sna...@ibm.net "Sergio Navega" writes:
> > >
> > > So, which one is right? Which one we should
> > > choose and which should be dismissed?
> > >
> >
> > I cover this in Fragments, and posted it again a couple of days
> > ago. Both are right, what the research shows is that human
> > reasoning is CONTENT bound, it is shaped by context and is
> > exemplar or stereotype driven.
> >
> > The second example does NOT contradict the first. The point is
> > that outside of familiar contexts people do not reason reliably.
> > They attribute, or use heuristics, a priori schemas.
>
> It's like saying that if you don't know something, you don't know it.
> What a revolutionary insight! Not! Faced with unfamiliar contexts, we
> attribute the best we can. Brittle extensional systems would just
> stop working altogether.
>
> Louis Savain
It's not "like saying that if you don't know something, you don't
know it" - stop thinking what it's "like" and study the context
itself.
'the modern.....position is that learned problem-solving
skills are, in general, idiosyncratic to the task.'
A. Newell 1980.
Newell, A. (1980). One final word.
In D. T. Tuma & F. Reif (Eds.),
Problem solving and education: Issues in teaching
and research.
Hillsdale, N.J.: Lawrence Erlbaum.
David Longley
In article <6nhnpi$hhu$3...@cantuc.canterbury.ac.nz>
mat...@math.canterbury.ac.nz "Bill Taylor" writes:
> Da...@longley.demon.co.uk (bot) writes:
>
> |> Whilst there is scope to discuss the heuristics which Tversky and
>
> Discuss them then and stop wittering!
>
>
> |> In passing, it might be of interest to some
>
> Little you say is of interest to anyone!
>
> |> that Simon is
> |> generally credited with the introduction of heuristics in
> |> computer science.
>
> What utter bilge.
>
"[we] now have the elements of a theory of heuristic (as
contrasted with algoritmic) problem solving; and we can
use this theory to understand humasn heuristic processes
and to simulate such with digital computers. Intuition,
insight, and learning are no longer exclusive
possessions of humans: any large high speed computer can
be programmed to exhibit them also."
Newell and Simon (1958) "Heuristic Problem Solving: The
Next Advance in Operations Research. Operations Research
6 (Jan-feb)"
Preface to:
Judgement under uncertainty: Heuristics and Biases
Ed Kahneman, Slovic and Tversky (1982)
"The approach to the study of judgment that this book
represents had origins in three lines of research that
developed in the 1950s and 1960s: the comparison of
clinical and statistical prediction, initiated by Paul
Meehl; the study of subjective probability in the
Bayesian paradigm, introduced to psychology by Ward
Edwards; and the investigation of heuristics and
strategies of reasoning, for which Herbert Simon offered
a program and Jerome Bruner an example. Our collection
also represents the recent convergence of the study of
judgment with another strand of psychological research:
the study of causal attribution and lay psychological
interpretation, pioneered by Fritz Heider.
Meehl's classic book, published in 1954, summarized
evidence for the conclusion that simple linear
combinations of cues outdo the intuitive judgments of
experts in predicting significant behavioral criteria.
The lasting intellectual legacy of this work, and of the
furious controversy that followed it, was probably not
the demonstration that clinicians performed poorly in
tasks that, as Meehl noted, they should not have
undertaken. Rather, it was the demonstration of a
substantial discrepancy between the objective record of
people's success in prediction tasks and the sincere
beliefs of these people about the quality of their
performance. This conclusion was not restricted to
clinicians or to clinical prediction: People's
impressions of how they reason, and of how well they
reason, could not be taken at face value. Perhaps
because students of clinical judgment often used
themselves and their friends as subjects, the
interpretation of errors and biases tended to be
cognitive, rather than psychodynamic: Illusions, not
delusions, were the model.
With the introduction of Bayesian ideas into
psychological research by Edwards and his associates,
psychologists were offered for the first time a fully
articulated model of optimal performance under
uncertainty, with which human judgments could be
compared. The matching of human judgments to normative
models was to become one of the major paradigms of
research on judgment under uncertainty. Inevitably, it
led to concerns with the biases to which inductive
inferences are prone and the methods that could be used
to correct them. These concerns are reflected in most of
the selections in the present volume. However, much of
the early work used the normative model to explain human
performance and introduced separate processes to explain
departures from optimality. In contrast, research on
judgmental heuristics seeks to explain both correct and
erroneous judgments in terms of the same psychological
processes.
The emergence of the new paradigm of cognitive
psychology had a profound influence on judgment
research. Cognitive psychology is concerned with
internal processes, mental limitations, and the way in
which the processes are shaped by the limitations. Early
examples of conceptual and empirical work in this vein
were the study of strategies of thinking by Bruner and
his associates, and Simon's treatment of heuristics of
reasoning and of bounded rationality. Bruner and Simon
were both concerned with strategies of simplification
that reduce the complexity of judgment tasks, to make
them tractable for the kind of mind that people happen
to have. Much of the work that we have included in this
book was motivated by the same concerns.
Fragments of Behaviour: The Extensional Stance
(A context of Heuristics and biases)
Between 1979 and 1983, I worked on the putative neural
mechanisms underlying the physiological control of operant
behaviour. In my undergraduate years I had focused primarily on
the relationship between Husserlian phenomenology and
contemporary psychology - examples were the work of JJ Gibson in
perception, and what has come to be known as Attribution Theory,
originating with the work of Heider in the late 40s, and Kelley
in the 1950s. The work in neuroscience required a degree of
mastery in several technical areas of behavioural and
neuroscience in order to run the necessary experiments, and it
was during this time of running hundreds of animals through
thousands of operant sessions that the fundamental issues
outlined in "Fragments of Behaviour: The Extensional Stance"
http://www.longley.demon.co.uk/Frag.htm, were first drafted as
background for a PhD. For the purposes of this post, this can be
summarised quite briefly.
In setting up an operant experiment, one generally does some
preliminary handling of the subjects along with some "magazine
training" whereby the animals learn where the reinforcer is
delivered from (here food pellets). This reduces stress and
"exploration", allowing the experimenter to get on with the
business of establishing baseline behaviour.
This tends to be done by what's called "the method of successive
approximation" or, in some cases "celloshaping". In the first,
traditional way, each time the animal shows some "fragment of
behaviour" which approximates what the experimenter ultimately
wants to shape up, is rewarded by pressing a remote control,
which triggers the food hopper to deliver a reward. Behaviours
approximating the required final class of behaviours are thereby
progressively "shaped" up by the experimenter, or, one could say,
the environment. (in "celloshaping", a few pellets are stuck on
the lever, and the animal left to pick up the contingency between
lever press and pellet delivery).
My first 6 months were spent testing out areas around the arcuate
nucleus of the hypothalamus to see if they would support
naloxone reversible intra-cranial self-stimulation (pleasure or
reward to some). (The arcuate nucleus has cells of origin for a
descending beta-endorphinergic pathway). Here the reward is not
food pellets, but a more immediate stimulation of parts of the
central nervous system which conventional reward was believed to
patch into.
Watching the animals acquire the lever press behaviour was quite
an education to me, as soon began to appreciate how much learning
goes on in what is usually just regarded as a preliminary stage
of shaping and baseline establishment. It also gave me the basis
for developing an alternative explanation to my chief rival's
theory of what lesions to central noradrenergic systems of the
Locus Coeruleus might be doing. He was proposing that the system
was involved in "attention" like processes, screening out
irrelevant associations. I proposed that what was happening on
the contrary, was that such animals were failing to configure
their behaviours into a class operant - this theory, a
configuring account of conditioning, lies at the heart of
"Fragments".
To cut a long story short, after about a year I began to run
experiments which focused on the early stages of familiarity or
habituation, examining the animal's unconditional responses to
novelty (neophobia). Apart from half a dozen more experiments on
noradrenaline and operant acquisition and extinction, I spent
most of my time refining experiments on naloxone and neophobia -
the thesis being that the endogneous opiates play a critical role
in establishing new fragments of behaviour or 'habits'.
In the NIMR annual report, the work was summarised as indicating
that endogenous opioid peptides are involved in the processing of
novel stimuli. Rather than regard naloxone as 'enhancing'
neophobia it is better to take the complement of that notion and
say that it retards the decline of neophobia. That decline is, I
have argued, co-extensive with learning, or habit formation. The
pre-requisite to learning, or if you prefer, the constraint on
learning, is a variant of the UCR of neophobia. Opioids must be
initially OFF (theoretical elaboration on that point takes one
back to a neo-Hullian model of learning/habit formation). It is
an attempt to get at the mechanics through one of the apposite
disciplines.
J.F.W. DEAKIN & D.C. LONGLEY*
(*introduced by T.J. Crow)
National Institute for Medical Research,
Mill Hill, London, NW7 1AA
Several studies report that naloxone, an opiate receptor
antagonist, reduces deprivation induced eating and drinking.
However, in the present study, naloxone (5mg/kg,i.p.) did not
reduce food intake of rats maintained on a 22 h deprivation - 2 h
feeding schedule. In contrast, naloxone (5 mg/kg,i.p.)
progressively reduced water intake in deprived animals to 46% of
saline treated controls. No effects of naloxone (1, 5 mg/kg) on
established bar pressing for food or water were observed with
either continuous or fixed ratio schedules of reinforcement.
However, naloxone (5mg/kg) accelerated extinction of responding
when food and water were no longer available.
Animals treated with naloxone (5mg/kg) during training of the
bar-pressing ate only 26% of the pellets delivered whereas
controls ate all pellets delivered. Since the animals had not
previously experienced the pellets or the operant apparatus, the
possibilities arose that naloxone effects were due to enhanced
neophobic effects of the novel food pellets or novel apparatus
cues, or were due to conditioned taste aversion. Therefore, food
novelty, apparatus novelty and timing of injections were
independently varied in different groups of 8-10 rats treated
with saline or naloxone. Rats were maintained at 85% body weight
with 12g lab chow per day. On experimental days 46 small pellets
(Cambden instruments) were placed on a small petri dish in the
home cage of some groups or released from a pellet dispenser in
an operant box for other groups. The dependent variable was the
number of pellets eaten over 15 minutes.
Naloxone (1,5 mg/kg i.p.) injected 5 or 20 min before test almost
completely suppressed pellet eating if the animals had not been
previously exposed to the pellets (p<0.01 't' test vs saline
groups). This occurred independently of whether tests were
carried out in the home cage or novel operant box. Naloxone
induced suppression of pellet eating was almost completely
abolished in either environment if animals had been exposed to
the pellets for the five preceding days in the same or different
environment. Naloxone (5mg/kg, i.p.) administered immediately
after pellet eating tests failed to suppress subsequent pellet
eating.
Thus, naloxone suppressed pellet eating if the pellets were novel
and if naloxone was administered before eating tests. The results
suggest naloxone enhances neophobic effects of novel foods and
that suppression of novel pellet eating is not due to enhanced
effects of novelty of apparatus cues or to conditioned taste
aversion.
Reference
FRENK, H & ROGERS G.H. (1979) The suppressant effects of naloxone
on food and water intake in the rat.
Behav. Neural. Biol, 26, 23-40.
PROCEEDINGS OF THE BRITISH PHARMACOLGICAL SOCIETY (BPS)
1-3 April 1981
(Also printed in British J Pharmacology 1981)
The rationale for all such research is ultimately to shed light
on the mechanisms of learning/behaviour. Note that in the shaping
procedure, behaviours are acquired even though they are not
causally instrumental in bringing about the reinforcer.
Similarly, when an animal a Conditioned Taste Aversion is
established by pairing a novel food or drink with malaise, the
malaise is not caused by the food or drink, it is just very
strongly associated with it because of the contingency and the
modality within which the stimuli are presented (in this case the
internal environment and taste). The mechanisms which subserve
these adaptive, defensive responses are in effect *heuristics*.
They are contingency sensitive up to a degree. From this, one can
see how when the heuristics get it right so to speak, it is, in
effect, mediated by the same adventitious/superstitious process.
I have covered the recent development context of "Fragments" in
the applied field of "corrections" in another post, and quite
comprehensively in "A System Specification For PROfiling
BEHaviour" (1994).
Here, I will just cite some further extracts from the work of
Tversky and Kahneman in the hope that it will make clear how
terms like "normative", "descriptive", "heuristic" and "bias" are
being used in the context of what I have been posting more
generally. The whole programme of work requires readers to look
at what is being said in the context it is being presented (Quine
1951).
o o o
Preface to:
Judgement under uncertainty: Heuristics and Biases
Ed Kahneman, Slovic and Tversky (1982)
"The approach to the study of judgment that this book
represents had origins in three lines of research that
developed in the 1950s and 1960s: the comparison of
clinical and statistical prediction, initiated by Paul
Meehl; the study of subjective probability in the
Bayesian paradigm, introduced to psychology by Ward
Edwards; and the investigation of heuristics and
strategies of reasoning, for which Herbert Simon offered
a program and Jerome Bruner an example. Our collection
also represents the recent convergence of the study of
judgment with another strand of psychological research:
the study of causal attribution and lay psychological
interpretation, pioneered by Fritz Heider.
Meehl's classic book, published in 1954, summarized
evidence for the conclusion that simple linear
combinations of cues outdo the intuitive judgments of
experts in predicting significant behavioral criteria.
The lasting intellectual legacy of this work, and of the
furious controversy that followed it, was probably not
the demonstration that clinicians performed poorly in
tasks that, as Meehl noted, they should not have
undertaken. Rather, it was the demonstration of a
substantial discrepancy between the objective record of
people's success in prediction tasks and the sincere
beliefs of these people about the quality of their
performance. This conclusion was not restricted to
clinicians or to clinical prediction: People's
impressions of how they reason, and of how well they
reason, could not be taken at face value. Perhaps
because students of clinical judgment often used
themselves and their friends as subjects, the
interpretation of errors and biases tended to be
cognitive, rather than psychodynamic: Illusions, not
delusions, were the model.
With the introduction of Bayesian ideas into
psychological research by Edwards and his associates,
psychologists were offered for the first time a fully
articulated model of optimal performance under
uncertainty, with which human judgments could be
compared. The matching of human judgments to normative
models was to become one of the major paradigms of
research on judgment under uncertainty. Inevitably, it
led to concerns with the biases to which inductive
inferences are prone and the methods that could be used
to correct them. These concerns are reflected in most of
the selections in the present volume. However, much of
the early work used the normative model to explain human
performance and introduced separate processes to explain
departures from optimality. In contrast, research on
judgmental heuristics seeks to explain both correct and
erroneous judgments in terms of the same psychological
processes.
The emergence of the new paradigm of cognitive
psychology had a profound influence on judgment
research. Cognitive psychology is concerned with
internal processes, mental limitations, and the way in
which the processes are shaped by the limitations. Early
examples of conceptual and empirical work in this vein
were the study of strategies of thinking by Bruner and
his associates, and Simon's treatment of heuristics of
reasoning and of bounded rationality. Bruner and Simon
were both concerned with strategies of simplification
that reduce the complexity of judgment tasks, to make
them tractable for the kind of mind that people happen
to have. Much of the work that we have included in this
book was motivated by the same concerns.
In recent years, a large body of research has been
devoted to uncovering judgmental heuristics and
exploring their effects. The present volume provides a
comprehensive sample of this approach. It assembles new
reviews, written especially for this collection, and
previously published articles on judgment and inference.
Although the boundary between judgment and decision
making is not always clear, we have focused here on
judgment rather than on choice. The topic of decision
making is important enough to be the subject of a
separate volume.
The book is organized in ten parts. The first part
contains an early review of heuristics and biases of
intuitive judgments. Part II deals specifically with the
representativeness heuristic, which is extended, in Part
III, to problems of causal attribution. Part IV
describes the availability heuristic and its role in
social judgment. Part V covers the perception and
learning of covariation and illustrates the presence of
illusory correlations in the judgments of lay people and
experts. Part VI discusses the calibration of
probability assessors and documents the prevalent
phenomenon of overconfidence in prediction and
explanation. Biases associated with multistage inference
are covered in Part VII. Part VIII reviews formal and
informal procedures for correcting and improving
intuitive judgments. Part IX summarizes work on the
effects of judgmental biases in a specific area of
concern, the perception of risk. The final part includes
some current thoughts on several conceptual and
methodological issues that pertain to the study of
heuristics and biases.
Daniel Kahneman
Paul Slovic
Amos Tversky
o o o
From "Judgments Under Uncertainty: Heuristics and Biases"
Intro to above volume and in:
Science 1974, 185, 1124-1131
"Discussion
This article has been concerned with cognitive biases
that stem from the reliance on judgmental heuristics.
These biases are not attributable to motivational
effects such as wishful thinking or the distortion of
judgments by payoffs and penalties. Indeed, several of
the severe errors of judgment reported earlier occurred
despite the fact that subjects were encouraged to be
accurate and were rewarded for the correct answers
(Kahneman & Tversky, 1972b, 3; Tversky & Kahneman,
1973,11). The reliance on heuristics and the prevalence
of biases are not restricted to laymen. Experienced
researchers are also prone to the same biases - when
they think intuitively. For example, the tendency to
predict the outcome that best represents the data, with
insufficient regard for prior probability, has been
observed in the intuitive judgments of individuals who
have had extensive training in statistics (Kahneman &
Tversky, 1973, 1974; Tversky & Kahneman, 1971, 2).
Although the statistically sophisticated avoid
elementary errors, such as the gambler's fallacy, their
intuitive judgments are liable to similar fallacies in
more intricate and less transparent problems.
It is not surprising that useful heuristics such as
representativeness and availability are retained, even
though they occasionally lead to errors in prediction or
estimation. What is perhaps surprising is the failure of
people to infer from lifelong experience such
fundamental statistical rules as regression toward the
mean, or the effect of sample size on sampling
variability. Although everyone is exposed, in the normal
course of life, to numerous examples from which these
rules could have been induced, very few people discover
the principles of sampling and regression on their own.
Statistical principles are not learned from everyday
experience because the relevant instances are not coded
appropriately. For example people do not discover that
successive lines in a text differ more in average word
length than do successive pages, because they simply do
not attend to the average word length of individual
lines or pages. Thus, people do not learn the relation
between sample size and sampling variability, although
the data for such learning are abundant.
The lack of an appropriate code also explains why
people usually do not detect the biases in their
judgments of probability. A person could conceivably
learn whether his judgments are externally calibrated by
keeping a tally of the proportion of events that
actually occur among those to which he assigns the same
probability. However, it is not natural to group events
by their judged probability. In the absence of such
grouping it is impossible for an individual to discover,
for example, that only 50 percent of the predictions to
which he has assigned a probability of .9 or higher
actually come true.
The empirical analysis of cognitive biases has
implications for the theoretical and applied role of
judged probabilities. Modern decision theory (de
Finetti, 1968; Savage, 1954) regards subjective
probability as the quantified opinion of an idealized
person. Specifically, the subjective probability of a
given event is defined by the set of bets about this
event that such a person is willing to accept. An
internally consistent, or coherent, subjective
probability measure can be derived for an individual if
his choices among bets satisfy certain principles, that
is, the axioms of the theory. The derived probability is
subjective in the sense that different individuals are
allowed to have different probabilities for the same
event. The major contribution of this approach is that
it provides a rigorous subjective interpretation of
probability that is applicable to unique events and is
embedded in a general theory of rational decision.
It should perhaps be noted that, while subjective
probabilities can sometimes be inferred from preferences
among bets, they are normally not formed in this
fashion. A person bets on team A rather than on team B
because he believes that team A is more likely to win;
he does not infer this belief from his betting
preferences. Thus, in reality, subjective proba-
bilities determine preferences among bets and are not
derived from them, as in the axiomatic theory of
rational decision (Savage, 1954).
The inherently subjective nature of probability has
led many students to the belief that coherence, or
internal consistency, is the only valid criterion by
which judged probabilities should be evaluated. From the
standpoint of the formal theory of subjective
probability, any set of internally consistent
probability judgments is as good as any other. This
criterion is not entirely satisfactory, because an
internally consistent set of subjective probabilities
can be incompatible with other beliefs held by the
individual. Consider a person whose subjective
probabilities for all possible outcomes of a coin-
tossing game reflect the gambler's fallacy. That is, his
estimate of the probability of tails on a particular
toss increases with the number of consecutive heads that
preceded that toss. The judgments of such a person could
be internally consistent and therefore acceptable as
adequate subjective probabilities according to the
criterion of the formal theory. Thcse probabilities,
however, are incompatible with the generally held belief
that a coin has no memory and is therefore incapable of
generating sequential dependencies. For judged
probabilities to be considered adequate, or rational,
internal consistency is not enough. The judg- ments must
be compatible with the entire web of beliefs held by the
individual. Unfortunately, there can be no simple formal
procedure for assessing the compatibility of a set of
probability judgments with the judge's total system of
beliefs. The rational judge will nevertheless strive for
compatibility, even though internal consistency is more
easily achieved and assessed. In particular, he will
attempt to make his probability judgments compatible
with his knowledge about the subject matter, the laws of
probability, and his own judgmental heuristics and
biases.
Summary
This article described three heuristics that are
employed in making judgments under uncertainty: (i)
representativeness, which is usually employed when
people are asked to judge the probability that an object
or event A belongs to class or process B; (ii)
availability of instances or scenarios, which is often
employed when people are asked to assess the frequency
of a class or the plausibility of a particular
development; and (iii) adjustment from an anchor, which
is usually employed in numerical prediction when a
relevant value is available. These heuristics are highly
economical and usually effective, but they lead to
systematic and predict- able errors. A better
understanding of these heuristics and of the biases to
which they lead could improve judgments and decisions in
situations of uncertainty."
o o o
On the study of statistical intuitions
Daniel Kahneman and Amos Tversky
Chapter 34 of above book, and in:
Cognition, 1982, 1, 123-141.
"Much of the recent literature on judgment and inductive
reasoning has been concerned with errors, biases, and
fallacies in a variety of mental tasks (see, e.g.,
Einhorn & Hogarth, 1981; Hammond, McClelland, &
Mumpower, 1980; Nisbett & Ross, 1980; Shweder, 1980;
Slovic, Fischhoff, & Lichtenstein, 1977; Tversky &
Kahneman, 1974,1). The emphasis on the study of errors
is characteristic of research in human judgment, but it
is not unique to this domain: We use illusions to
understand the principles of normal perception and we
learn about memory by studying forgetting. Errors of
reasoning, however, are unique among cognitive failures
in two significant respects: They are somewhat
embarrassing and they appear avoidable. We are not
troubled by our susceptibility to the vertical-
horizontal illusion or by our inability to remember a
list of more than eight digits. In contrast, errors of
reasoning are often disconcerting - either because the
solution that we failed to find appears quite obvious in
retrospect or because the error that we made remains
attractive although we know it to be an error. Many
current studies of judgment are concerned with problems
that have one or the other of these characteristics.
The presence of an error of judgment is demonstrated
by comparing people's responses either with an
established fact (e.g., that the two lines are equal in
length) or with an accepted rule of arithmetic, logic,
or statistics. However, not every response that appears
to contradict an established fact or an accepted rule is
a judgmental error. The contradiction could also arise
from the subject's misunderstanding of the question or
from the investigator's misinterpretation of the answer.
The description of a particular response as an error of
judgment therefore involves assumptions about the
communication between the experimenter and the subject.
(We shall return to this issue later in the chapter.)
The student of judgment should avoid overly strict
interpretationS. which treat reasonable answers as
errors, as well as overly charitable interpretations
which attempt to rationalize every response.
Although errors of judgment are but a method by
which some cognitive processes are studied, the method
has become a significant part of the message. The
accumulation of demonstrations in which intelligent
people violate elementary rules of logic or statistics
has raised doubts about the descriptive adequacy of
rational models of judgment and decision making. In the
two decades following World War II, several descriptive
treatments of actual behavior were based on normative
models: subjective expected utility theory in analyses
of risky choice, the Bayesian calculus in investigations
of changes of belief, and signal-detection theory in
studies of psychophysical tasks. The theoretical
analyses of these situations, and to a much lesser
degree the experimental results, suggested an image of
people as efficient, nearly optimal decision makers. On
this background, observations of elementary violations
of logical or statistical reasoning appeared surprising,
and the surprise may have encouraged a view of the human
intellect that some authors have criticized as unfairly
negative (L. J. Cohen, 1979,1981; W. Edwards, 1975;
Einhorn & Hogarth, 1981).
There are three related reasons for the focus on
systematic errors and inferential biases in the study of
reasoning. First, they expose some of our intellectual
limitations and suggest ways of improving the quality of
our thinking. Second, errors and biases often reveal the
psychological processes and the heuristic procedures
that govern judgment and inference. Third, mistakes and
fallacies help the mapping of human intuitions by
indicating which principles of statistics or logic are
non-intuitive or counter-intuitive.
The terms intuition and intuitive are used in three
different senses. First, a judgment is called intuitive
if it is reached by an informal and unstructured mode of
reasoning, without the use of analytic methods or
deliberate calculation. For example, most psychologists
follow an intuitive procedure in deciding the size of
their samples but adopt analytic procedures to test the
statistical significance of their results. Second, a
formal rule or a fact of nature is called intuitive if
it is compatible with our lay model of the world. Thus,
it is intuitively obvious that the probability of
winning a lottery prize decreases with the number of
tickets, but it is counterintuitive that there is a
better than even chance that a group of 23 people will
include a pair of individuals with the same birthday.
Third, a rule or a procedure is said to be part of our
repertoire of intuitions when we apply the rule or
follow the procedure in our normal conduct. The rules of
grammar, for example, are part of the intuitions of a
native speaker, anal' some (though not all) of the rules
of plane geometry are incorporated into' our spatial
reasoning.
The present chapter addresses several methodological
and conceptual| problems that arise in attempts to map
people's intuitions about chance| and uncertainty. We
begin by discussing different tests of statistical
intuitions; we then turn to a critique of the question-
answering paradigm in judgment research; and we conclude
with a discussion of the non-intuitive character of some
statistical laws."
o o o
The programme of work outlined in "Fragments of Behaviour: The
Extensional Stance" formed part of the Home Office/Birkbeck
College University of London MSc in Applied Criminological
Psychology, 1988-1995. Whilst the Tversky and Kahneman work was a
standard reference handled out to all course participants from
its inception, even within the field for which it was designed,
the implications were rarely fully appreciated. A point
recognized by Dawes, Faust and Meehl 1989:
'Research on clinical versus statistical judgement has
had little impact on everyday decision making, particularly
within its field of origin, clinical psychology......The
interview remains the sine qua non of entrance into mental
health training programs and is required in most states to
obtain a license to practice. Despite the studies that show
that clinical interpretation of interviews may have little or
no predictive utility, actuarial interpretation of interviews
is rarely if ever used, although it is of demonstrated value.
Ultimately, then, clinicians must choose between
their own observations or impressions and the scientific
evidence on the relative efficacy of the clinical and
actuarial methods. The factors that create difficulty in
self-appraisal of judgmental accuracy are exactly those that
scientific procedures, such as unbiased sampling,
experimental manipulation of variables, and blind assessment
of outcome, are designed to counter. Failure to accept a
large and consistent body of scientific evidence over
unvalidated personal observation may be described as a normal
human failing or, in the case of professionals who identify
themselves as scientific, plainly irrational.
Finally, actuarial methods - at least within the domains
discussed in this article - reveal the upper bounds in our
current capacities to predict human behavior. An awareness of
the modest results that are often achieved by even the best
available methods can help to counter unrealistic faith in
our predictive powers and our understanding of human
behavior. It may well be worth exchanging inflated beliefs
for an unsettling sobriety, if the result is an openness to
new approaches and variables that ultimately increase our
explanatory and predictive powers.
What is needed is the is the development of actuarial
methods and a measurement assurance program that maintains
control over both judgement strategies so that their
operating characteristics in the field are known and an
informed choice of procedure is possible. Dismissing the
scientific evidence or lamenting the lack of available
methods will prove much less productive than taking on the
needed work.'
Dawes, Faust & Meehl (1989)
Science v243, p1668-1674
Clinical Versus Actuarial Judgement
It is worth repeating one sentence from the above:
to accept a large and consistent body of scientific evidence
over unvalidated personal observation may be described as a
normal human failing or, in the case of professionals who
identify themselves as scientific, plainly irrational'.
o o o
For those interested, I have made available the most recent
version of "Fragments" at:
http://www.longley.demon.co.uk
along with other supporting papers. Given the difficulties which
this type of work generally encounters, my efforts are now being
invested primarily in trying to bring the implications of the
research and development work to the attention of policy makers.
>
> |> As I say in "Fragments" -
>
> I bet Nostradamus said it too. Probably better.
David Longley
> What is irritating about Longley, is that after several years
> involvement with c.a.p. he still has not discovered that it is not a
> research journal, and that citation is never required and often
> inappropriate.
>
It's time people did. It's time a little evidence and reality was
introduced in place of rhetorical statements to one another of
differential states of scientific ignorance, presented as
"beliefs".
What's I've posted is designed to help change the status quo.
Call it therapy.
Bill Taylor
Da...@longley.demon.co.uk (David Longley) writes:
|> There are good reasons for using quotation rather than "thought"
|> - as I explain in "Fragments".
No-one here is interested in your Nostradamus-like tomes.
|> like several others in this group, you just won't be told anything.
We keep waiting for you to tell us something, but you never do.
And it's *everyone* here, not "several others". Why don't you get the
message and bugger off?
Jim Balter
David Longley wrote:
>
> In article <6nheu8$l...@ux.cs.niu.edu> ric...@cs.niu.edu "Neil Rickert" writes:
>
> > What is irritating about Longley, is that after several years
> > involvement with c.a.p. he still has not discovered that it is not a
> > research journal, and that citation is never required and often
> > inappropriate.
> >
>
> It's time people did. It's time a little evidence and reality was
> introduced in place of rhetorical statements to one another of
> differential states of scientific ignorance, presented as
> "beliefs".
That sounds a bit rhetorical.
> What's I've posted is designed to help change the status quo.
> Call it therapy.
It doesn't seem to be an "effective procedure".
--
<J Q B>
David Longley
In article <6ni7lg$q41$5...@cantuc.canterbury.ac.nz>
mat...@math.canterbury.ac.nz "Bill Taylor" writes:
>
> And it's *everyone* here, not "several others". Why don't you get the
> message and bugger off?
>
Do you have any evidence for this? or like Rickert, doesn't
evidence count in your closed and idiosyncratic world?
I've been answering the title of this thread:
"What has Longley got against common sense?"
I don't see you contributing anything but abuse.
David Longley
In article <359CAC3F...@sandpiper.net>
j...@sandpiper.net "Jim Balter" writes:
DL
> >
> > It's time people did. It's time a little evidence and reality was
> > introduced in place of rhetorical statements to one another of
> > differential states of scientific ignorance, presented as
> > "beliefs".
>
> That sounds a bit rhetorical.
>
No, I mean it quite literally. What people should do is
consult the material which has been quoted and referenced.
They'll probably find that this changes what they know. My
experience over the past 10 years suggests that people confuse
their own restricted beliefs with what is actually known.
> > What's I've posted is designed to help change the status quo.
> > Call it therapy.
>
> It doesn't seem to be an "effective procedure".
>
What I have posted refers to a system of effective procedures.
Here, I'm making references to it, along with providing pointers
to the relevant literature.
One of the points made in the material is that just talking about
it is rarely enough to change behaviour - that's why what I have
to say is so critical of the current investment in "cognitive
skills" programmes. But the truth or falsehood, and ultimately
the merits of what I have to say should not be judged by its
*popularity* and especially not its immediate familiarity.
There are innovative proposals being made - and they are
proposals that will work, and have been demonstrated to work.
To appreciate that, you have to do it, not just think about it. I
assume that most people will not understand what I am talking
about - that's why I've written what I have, in the way that I
have.
Phil Roberts, Jr.
David Longley wrote:
>
>
> No, I mean it quite literally. What people should do is
> consult the material which has been quoted and referenced.
> They'll probably find that this changes what they know. My
> experience over the past 10 years suggests that people confuse
> their own restricted beliefs with what is actually known.
>
The wonderful thing about the net is that it makes it possible to advertise
ideas that in the old days would have never been allowed to see the light
of day. However, as with all advertising, you have to give the public
a _reason_ to act on your information.
I really mean it (as in not intending to be flippant or in any way
demeaning) when I say I have been reading your posts for some
two years now and, not only do I not understand what your
position is (e.g., the extensional stance has never
been clearly explained or justified in simple English), but the
only argument I have ever been given
for acting on your advertisement which _is_ simple and clear
has been the argument from authority (Trust me, I'm a professional).
Since I
have spent most of my life with what is perhaps an almost irrational
_disdain_ for authority, and since I presume it to be a fairly commonplace
attitude out here in cyberland, I believe it is reasonable to
conclude that the problem is not so much with me, Mr. John Q. Public,
(assuming I am typical of your intended
audience -- a patient reasonably intelligent human being
constantly in search of interesting new ideas) as with your
"advertisements" themselves.
> One of the points made in the material is that just talking about
> it is rarely enough to change behaviour - that's why what I have
> to say is so critical of the current investment in "cognitive
> skills" programmes. But the truth or falsehood, and ultimately
> the merits of what I have to say should not be judged by its
> *popularity* and especially not its immediate familiarity.
>
> There are innovative proposals being made - and they are
> proposals that will work, and have been demonstrated to work.
>
> To appreciate that, you have to do it, not just think about it. I
> assume that most people will not understand what I am talking
> about - that's why I've written what I have, in the way that I
> have.
>
I have a proposal for remedying _your_ communication problem
if you are interested. As in advertising, I would simply
like you to show me how your "truths" relate to things which
are near and dear to my own heart. I know that I am only one
person, but I can't imagine a better endorsement for your
product than to have all the viewers out there in TV land be
able to witness a conversion right here in front of their eyes.
(Cool. Our own little version of the Truman Show).
I won't delude you here. I am profoundly skeptical that
you can do this, and I also have my own axe to grind. What
I do quarantee you though, is that I will at least _try_
to understand your point of view, I will at least _try_
to balance your objectives with my own. Perhaps most
importantly, I will at
least _try_ not to be flippant and demeaning, and if you
catch me doing it, by all means call it to my attention.
All I ask is that you remember that you are going to
have to relate your "truths" to _my_ "truths" and show
me _personally_ how your "truths" should alter my
perspective on them.
A final note: Maybe it isn't quite as democratic as the net
is supposed to be, but I am concerned that your own
views have been so much under assault as of late that
it really isn't fair for you to have to fend off, not
only me, but
fifteen others guys all at the same time. To the
extent possible, I think we should try to keep this
a dialogue in the literal sense of the term and to
ask the viewers out there in TV land to honor our
desire to restrict the thread to a conversation
between just the two of
us. Where others do intrude, we ourselves should
either not respond at all, or move our response
to a different thread.
Are you game?