Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.
Dismiss

Computer Science Education

16 views
Skip to first unread message

Ryan Holbrook

unread,
Dec 4, 2001, 12:30:54 PM12/4/01
to
I am a second-year computer science student at the University of Oklahoma.
This past summer I stumbled across Touretzky's text and then SICP. I feel I
learned more about what computer science is really about in that two-month
period than in the three semesters I've spent studying UML diagrams and the
Java API. What I discovered that summer I truly enjoyed. Unfortunately, the
courses offered seem to offer none of what I had found so fascinating.
(With one exception: discrete mathematics.)

Eventually I decided the computer science program was not for me and that I
should explore other majors. What I have decided upon is a "planned
program" offered by the College of Arts and Sciences which allows a student
to follow a course of study of their own design. I am calling it cognitive
science. It is supposed to be a study in the computational modeling of
higher-order cognitive functions (however limited). A preliminary proposal
is on my web page in postscript and pdf format:

http://students.ou.edu/H/Ryan.P.Holbrook-1/

Because of the nature of the subject area, implementing any significant
project in the languages and programming paradigms currently taught is
undesirable. I would like to be able to use the "best tool for the job," as
they say, which I believe to be Lisp.

Does anyone have any advice or experience in trying to use Lisp in an
educational environment which is mostly ignorant of it?

Any critiques of the proposal itself are appreciated, also. I have been
mostly flying blind on this and would very much like any additional input.

Ryan Holbrook
--
http://students.ou.edu/H/Ryan.P.Holbrook-1/

Kent M Pitman

unread,
Dec 5, 2001, 4:02:47 AM12/5/01
to
[Most of my reply I sent in private email to Ryan, but this part seemed
relevant to the group.]

Ryan Holbrook <rhol...@mmcable.com> writes:

> Does anyone have any advice or experience in trying to use Lisp in an
> educational environment which is mostly ignorant of it?

Focus on what you plan to accomplish and deemphasize the vehicle. The
mention of Lisp should be in the context of identifying areas that you
think are tractable and in showing that tools of similar kind have been
built in that area. Don't ever make a plan that says "I propose to do
X in Lisp (or C or any one thing)." Instead, make sure your proposal stands
as interesting regardless of the target medium. Use Lisp, if you find it
useful to do so, as a way of suggesting that it can be done, or just to say
"I plan, by the way, to do this in Lisp" somewhere in the text. But don't
make it criterial. This means that if someone tries to attack your choice,
you can defend yourself by saying "If it doesn't work out in Lisp, I can
always use another language." What stands in practice between you and
success is not going to be your choice of language, probably, but your
understanding of your project. But being able to disarm uninformed critics
easily is what you need to do, so making Lisp detachable from your project
will make it detachable from their criticisms, if you get my meaning.
I think you'll find Lisp a fine vehicle, but who knows, people are always
inventing new things and something might come along that you like better.
We don't make people sign blood oaths of allegiance.

Kenny Tilton

unread,
Dec 5, 2001, 2:54:10 PM12/5/01
to

Ryan Holbrook wrote:
> Does anyone have any advice or experience in trying to use Lisp in an
> educational environment which is mostly ignorant of it?

I guess for once the good news is that Lisp and AI are assumed to go
together, so you have that going for you. Also Lisp is known for rapid
prototyping, so just propose to prototype in Lisp and port to C++ later,
they'll love that, then just don't get around to the port, no one will
notice.

>
> Any critiques of the proposal itself are appreciated, also. I have been
> mostly flying blind on this and would very much like any additional input.

I looked at the proposal. My two cents: a lot less math, a little less
philosophy, much more neuroscience.

The first wave of AI researchers dismissed the brain. They did not care
how the biological system worked. Instead they tried to calculate mind
and came up empty. Since the bio system works so well it might be a good
place to start. Especially if you want to figure out consciousness.

kenny
clinisys

Kent M Pitman

unread,
Dec 5, 2001, 5:05:09 PM12/5/01
to
Kenny Tilton <kti...@nyc.rr.com> writes:

> I looked at the proposal. My two cents: a lot less math, a little less
> philosophy, much more neuroscience.

I liked most of the philosophy courses he'd outlined, except the
symbolic logic. I'm skeptical of "symbolic logic" as a course to
take, and especially of two such, only because I took a real waste of
time course. I'm sure the discipline of this is fine, but it's easy
for courses to waste time. Mine did a day on the truth table for AND,
a day on the truth table for OR, etc. I computed how many truth tables
remained and went away from the class for several weeks. Sure enough,
I came back and they were doing the last truth table. Two days
remained in the class. They did a day on incompleteness and a day on
the halting problem. Sigh. Talk about badly distributing one's
resources! I kept trying to imagine how anyone who took that many weeks
to learn a set of logic truth tables was going to get either of those
last two topics each in a single day.

I agree it was too heavy on math. I said I thought cog sci people
don't do much calculus. A course or two may be good for the soul, but
four levels of Calc is wasted as prep for cog sci, IMO. I suggested
maybe topology if he wanted another math. Something that would speak
to what might ultimately become "web science". Even probs&stats, while
important, probably is overkill at two regular classes and a lab. I'd
get it to one or two to buy room for other things.

I agree suggested a chem course to help complement the bio stuff, and also
because I thought the experience in physical modeling was a good thing.

And I suggested picking up a few other computer languages even though
I think Lisp will be the best for his stated purpose. One can't be
too diversified in that area... I'm not going to recommend that a
student put blinders on too early even if it wins us a convert. His
choice of Lisp needs to be an informed one, and in the modern
heterogeneous environment, he needs multiple language skills.

> The first wave of AI researchers dismissed the brain. They did not care
> how the biological system worked. Instead they tried to calculate mind
> and came up empty. Since the bio system works so well it might be a good
> place to start. Especially if you want to figure out consciousness.

I'm not convinced that the gap is small enough that you can work up to
AI from what we know of the brain. Studying neuro may be important in
its own right, and there are certainly reasons to study it, but I
think one of them is not to understand about how people think. That's
like, as happened to me in college, and caused me to transfer from the
EE&CS dept to the philosophy department, insisting that understanding
electrical wave forms would give me insight into programming languages
and their design. I finally got some professors to admit they were making
me take signals and circuits as a major requirement not because I was going
to need it for high level programming, but because they knew how to grade
hard sciences and couldn't figure out how to grade computer science, and
so they wanted to make sure I'd taken at least some courses they could grade,
whether they were relevant or not. College is too expensive for that kind
of game. I wanted it to provide me with a targeted experience, not just a
pat on the back and a sense that I'd done something random of someone else's
choosing. The major I ended up with in the MIT Philosophy dept. is
remarkably similar to the one Ryan is designing, and I ended up there for
very similar motivations as well. So I think he's on the right track.

Kenny Tilton

unread,
Dec 5, 2001, 6:17:32 PM12/5/01
to

Kent M Pitman wrote:


>
> Kenny Tilton <kti...@nyc.rr.com> writes:
>
> > The first wave of AI researchers dismissed the brain. They did not care
> > how the biological system worked. Instead they tried to calculate mind
> > and came up empty. Since the bio system works so well it might be a good
> > place to start. Especially if you want to figure out consciousness.
>
> I'm not convinced that the gap is small enough that you can work up to
> AI from what we know of the brain. Studying neuro may be important in
> its own right, and there are certainly reasons to study it, but I
> think one of them is not to understand about how people think.

True. My interest in consciousness happens to be more about how we as
physical systems manage to recognize, perceive and even simply become
aware. Especially impressive is the nervous system's speed. So my
particular interests happen to be all about neuroanatomy; I suspect we
can learn a few tricks from Nature.

Bottom line, the mix of coursework depends on the student's particular
interest within the larger domain of cognitive science.

kenny
clinisys

Software Scavenger

unread,
Dec 5, 2001, 6:29:17 PM12/5/01
to
Kenny Tilton <kti...@nyc.rr.com> wrote in message news:<3C0E7C39...@nyc.rr.com>...

> place to start. Especially if you want to figure out consciousness.

What is consciousness, that it needs figuring out? Is it really a
mystery, or just a vague idea which needs to be defined more clearly?

Sashank Varma

unread,
Dec 5, 2001, 7:00:28 PM12/5/01
to
In article <OoiP7.2009$c4.59...@ounews.ou.edu>, Ryan Holbrook
<rhol...@mmcable.com> wrote:

>Eventually I decided the computer science program was not for me and that I
>should explore other majors. What I have decided upon is a "planned
>program" offered by the College of Arts and Sciences which allows a student
>to follow a course of study of their own design. I am calling it cognitive
>science. It is supposed to be a study in the computational modeling of
>higher-order cognitive functions (however limited). A preliminary proposal
>is on my web page in postscript and pdf format:
>
>http://students.ou.edu/H/Ryan.P.Holbrook-1/

I took a similar path through college. Some advice:

First, there are established majors at other universities that
meet your needs. I am not saying transfer, but rather consult
the structure of their programs for hints on structuring your
own. None are perfect; you'd be surprised how academc politics
usually annoints one department the 'home' of an interdisciplinary
major, with predictabley parochial results. Nevertheless:

http://www.stanford.edu/dept/symbol/
http://hss.cmu.edu/HTML/departments/philosophy/BMB/logic_comp/index.html
http://web.mit.edu/bcs/
http://cogsci.ucsd.edu/cogsci/index.html

These are just representative majors; the first two slant to
the logical side, the latter two to the neuro/psychology side.

Computer science classes: If you find yourself liking 'language'
in its myriad forms -- computer languages, formal languages,
natural languages -- a class on compiler design may be a nice
complement to your existing selection. Also, you should
consider more AI courses. Artificial neural networks and
machine learning are 'hot' and useful topics.

Linguistics: Linguistics is a crapshoot. I enjoy it tremendously,
but many cognitive scientists could care less. Use the intro
class to discover in which of these two camps you fit. If you
like it, you may branch in several directions. You may cover
syntactic (generative grammar) theories in an elective. I find
Linguistic Semantics an island within cognitive science and would
avoid spending a precious undergrad class on this topic unless
you feel compelled. A class on computational linguistics is
almost certainly worthwhile; if you really love the material take
one on symbolic parsing techniques and one on the use of corpora
and statistical techniques -- this too is very 'hot' at the
moment. Consider classes on anthropological linguistics (how
culture and language interact) and historical linguistics (how
languages change over time), as these are fascinating topics.

Math: Unless you have an intrinsic interest in math, you may
be going overboard here. (I majored in math, so what follows
pains me slightly!) You could probably cut the third and fourth
semesters of calculus -- triple integrals and differential equations
are not commonly encountered in cognitive science. Abstract
algebra is also rarely seen in cognitive science -- the only
example I can think of off the top of my head is Piaget's theory
of 'groupings'. Depending on what topics you like in your discrete
math class, you may take a semester-long class on combinatorics or
graph theory instead.

Philosophy: From their names, the first two classes sound really
basic, redundant with your more advance classes, and thus skippable.
Consider courses on epistemology and philosophy of science. Also,
you may want to take a course on the 'softer' side of cognitive
science. For example, phenomenology is 'hot' in cognitive science
at the moment (e.g., the 'embodied mind' camp).

Psychology: Skip the intro class if at all possible. A useful
tactic may be to substitute a history of psychology instead. You
will learn nothing that sticks in intro psychology. The learning
and conditioning class is a relic of the past. These principles
are easily mastered on your own, and are rarely seen in the
cognitive science literature (outside of animal work). If the
'Research Methods in Statistics' class teaches you stats, it is
redundant with your mathematical variant above, so drop it. If
it teaches you how to design and run experiments, then you MUST
take it. This is essential knowledge that most armchair cognitive
scientists lack. I'm not sure about classes called 'Physiological
Psychology'. This might signify a dusty relic of the past or it
may be a neuroscience course lurking under a vestigal name. Make
sure it's the latter. If not, seek out (cognitive and developmental)
neuroscience courses. It is unlikely you'll ever work with 'the wet
stuff', but you might one day conduct research on (or consume
research about) patients with brain damage, or PET and especially
fMRI studies of healthy subjects. Try to take classes or gain
experiences in this direction.

DO NOT OVEREMPHASIZE NEURO!!! It is the hottest of everything
hot right now, but these things go in cycles. You need to be
an educated consumer of this work. You do not have to be a
producer, unless you feel called, so don't force the issue.
You will be better served taking classes on cognitive
development, perception and motor behavior, etc., before taking
that 3rd neuro class if your heart's not in it.

Zoology: This is not on the critical path of cognitive science.
Take it if it looks interesting, but don't feel compelled. A
course on animal behavior, though...

The single most important training experience you can have in
college is to work under a professor on new research. Find a
professorial mentor. Then infilitrate the local network of
graduate students. Get your hands dirty.

Hope this helps. Feel free to email me p
ersonally should you feel
the need.

Frank A. Adrian

unread,
Dec 6, 2001, 12:38:28 AM12/6/01
to
Kent M Pitman wrote:
> I liked most of the philosophy courses he'd outlined, except the
> symbolic logic. I'm skeptical of "symbolic logic" as a course to
> take, and especially of two such, only because I took a real waste of
> time course. I'm sure the discipline of this is fine, but it's easy
> for courses to waste time. Mine did a day on the truth table for AND,
> a day on the truth table for OR, etc. I computed how many truth tables
> remained and went away from the class for several weeks. Sure enough,
> I came back and they were doing the last truth table. Two days
> remained in the class. They did a day on incompleteness and a day on
> the halting problem. Sigh. Talk about badly distributing one's
> resources! I kept trying to imagine how anyone who took that many weeks
> to learn a set of logic truth tables was going to get either of those
> last two topics each in a single day.

On the other hand, if you enjoyed discrete math, you might want to take at
least one symbolic logic course and one abstract algebra course to allow
you to understand non-standard logics (e.g., fuzzy stuff, multi-valued,
etc.). A good stats course is always helpful for understanding Baysian and
probablistic logic models, as well. You'd also want a good course in linear
algebra and perhaps differential equations to give you a good foundation
for the math used in neural networks, too.

The good news is that you might not need to take five or six courses for
that stuff. Since you're designing your own program, you might be able to
find a young mathematics professor who's interested in working with the CS
department in developing a course in these areas, using you for the guinea
pig and researcher for the work. Since you were able to tackle SICP and
the other Lisp text in two months, you should be able to do a quick survey
of "Mathematical Topics in Inexact Reasoning" in a couple of semesters. It
will also be good preparation and resume building for your PhD.

Because the other thing to remember is that if you want to be doing real
Cog Sci work right now, you *will* be slogging through to your PhD. As
much as I'd like to say differently, the applications in industry in pure
Cog Sci are few and far between and, usually, still in the research labs.
For those types of positions, you'll need those three letters on your
diploma. The good news is that by designing your own undergraduate
program, you will have shown yourself capable of independent study and your
ability to work with your professors. You should be able to get your pick
of about any graduate advisor you'd want to work under with those
credentials.

Good luck...

faa

Software Scavenger

unread,
Dec 6, 2001, 12:55:18 AM12/6/01
to
Kenny Tilton <kti...@nyc.rr.com> wrote in message news:<3C0EABBE...@nyc.rr.com>...

> aware. Especially impressive is the nervous system's speed. So my
> particular interests happen to be all about neuroanatomy; I suspect we
> can learn a few tricks from Nature.

The nervous system is massively parallel. When will we have computer
hardware with the same advantage to the same degree?

Using a programming language equivalent to Common Lisp on a TRS-80
from Radio Shack in 1978 would not have made much sense, regardless of
how powerful a language it is. Likewise, trying to do true AI on
present day hardware probably doesn't make much sense either. It was
probably better to program the TRS-80 in assembler language, learning
as much as possible about programming, gaining as much experience as
possible, while waiting for a CL-like-language to become feasible on
commonly available hardware. So instead of working on true AI now, we
should probably be working on the more primitive stuff appropriate for
the level of technology available in the early 21st century. There is
plenty of stuff needed which can be done with our present hardware.
Such as making it easier to program the computer by making the
computer do a larger share of the work of programming.

Ryan Holbrook

unread,
Dec 5, 2001, 1:11:31 PM12/5/01
to
Kent M Pitman wrote:
> I liked most of the philosophy courses he'd outlined, except the
> symbolic logic. I'm skeptical of "symbolic logic" as a course to
> take, and especially of two such, only because I took a real waste of
> time course. I'm sure the discipline of this is fine, but it's easy
> for courses to waste time. Mine did a day on the truth table for AND,
> a day on the truth table for OR, etc. I computed how many truth tables
> remained and went away from the class for several weeks. Sure enough,
> I came back and they were doing the last truth table. Two days
> remained in the class. They did a day on incompleteness and a day on
> the halting problem. Sigh. Talk about badly distributing one's
> resources! I kept trying to imagine how anyone who took that many weeks
> to learn a set of logic truth tables was going to get either of those
> last two topics each in a single day.
>

I considered axeing them, but I think I might keep the first one after all.
I looked at the course's homepage and it seems like it could be worthwhile.
This may be something of a political decision too, as the head of the
curriculum board is also the head of the philosophy department. I am going
to ask around some before I make any final decisions, though.

My discrete mathematics professor recommended the two symbolic logic
courses. He seemed to think I could test out of the first one. The discrete
math course was somewhat non-traditional from what I can gather: the first
half of the semester was devoted to symbolic logic, while the second half
focused on proving a software system's correctness using Haskell. We also
touched on Godel's Theorem and the halting problem. Not the normal fare, I
think, but enjoyable none the less.

> I agree it was too heavy on math. I said I thought cog sci people
> don't do much calculus. A course or two may be good for the soul, but
> four levels of Calc is wasted as prep for cog sci, IMO. I suggested
> maybe topology if he wanted another math. Something that would speak
> to what might ultimately become "web science". Even probs&stats, while
> important, probably is overkill at two regular classes and a lab. I'd
> get it to one or two to buy room for other things.
>

Calculus III is a prerequisite to Linear Algebra, which is required for
many upper-division CS coures. I replaced the two statistics coures with a
combined Applied Stastical Methods and added Foundations of Analysis which
covers topology and is a requirement for other topology and graph theory
courses in case I decide to continue in that area.

> I agree suggested a chem course to help complement the bio stuff, and also
> because I thought the experience in physical modeling was a good thing.
>

I took Chemistry I as a freshman and had a similar experience to yours. The
chemistry I had in high school was more in-depth. Needless to say, the nine
hours a week I spent in lecture or lab were not pleasent.



> And I suggested picking up a few other computer languages even though
> I think Lisp will be the best for his stated purpose. One can't be
> too diversified in that area... I'm not going to recommend that a
> student put blinders on too early even if it wins us a convert. His
> choice of Lisp needs to be an informed one, and in the modern
> heterogeneous environment, he needs multiple language skills.
>

That sounds like good advice. Do you recommend any languages in particular?

Thank you for your comments. You have been very helpful.

Ryan Holbrook

unread,
Dec 5, 2001, 1:55:45 PM12/5/01
to
Sashank Varma wrote:

> First, there are established majors at other universities that
> meet your needs. I am not saying transfer, but rather consult
> the structure of their programs for hints on structuring your
> own. None are perfect; you'd be surprised how academc politics
> usually annoints one department the 'home' of an interdisciplinary
> major, with predictabley parochial results. Nevertheless:

I am supposed to find a faculty member to be my advisor and oversee my
senior project. I am trying to make the program as complete as possible
before approaching anyone, though, as I am afraid of the very phenomenon
you mention. While I want to work under someone who has similar interests
as mine, I realize that they will not be identical and I don't want to be
coerced (even unintentionally) into a course of study that I would not
otherwise be happy with. This way, I can get a variety of opinions before
feeling committed to anything.

> Computer science classes: If you find yourself liking 'language'
> in its myriad forms -- computer languages, formal languages,
> natural languages -- a class on compiler design may be a nice
> complement to your existing selection. Also, you should
> consider more AI courses. Artificial neural networks and
> machine learning are 'hot' and useful topics.
>

All of the advanced classes in AI are graduate level, so I may have some
trouble getting in to them. I'm not sure how strict the department is on
their policies.

> Linguistics: Linguistics is a crapshoot. I enjoy it tremendously,
> but many cognitive scientists could care less. Use the intro
> class to discover in which of these two camps you fit. If you
> like it, you may branch in several directions. You may cover
> syntactic (generative grammar) theories in an elective. I find
> Linguistic Semantics an island within cognitive science and would
> avoid spending a precious undergrad class on this topic unless
> you feel compelled. A class on computational linguistics is
> almost certainly worthwhile; if you really love the material take
> one on symbolic parsing techniques and one on the use of corpora
> and statistical techniques -- this too is very 'hot' at the
> moment. Consider classes on anthropological linguistics (how
> culture and language interact) and historical linguistics (how
> languages change over time), as these are fascinating topics.
>

That sounds like good advice. I've selected three linguistic classes for
breadth and I'll reserve the others for electives.

I don't believe OU offers computational linguistics, though that is
certainly something I could pick up through supervised independant study.

> Math: Unless you have an intrinsic interest in math, you may
> be going overboard here. (I majored in math, so what follows
> pains me slightly!) You could probably cut the third and fourth
> semesters of calculus -- triple integrals and differential equations
> are not commonly encountered in cognitive science. Abstract
> algebra is also rarely seen in cognitive science -- the only
> example I can think of off the top of my head is Piaget's theory
> of 'groupings'. Depending on what topics you like in your discrete
> math class, you may take a semester-long class on combinatorics or
> graph theory instead.
>

Calculus III is a requirement of many of the CS classes but I did decide to
drop calc IV. I've gotten varied suggestions about group theory. Some like
it, some don't, so I'll have to ask around a little more.

> Philosophy: From their names, the first two classes sound really
> basic, redundant with your more advance classes, and thus skippable.
> Consider courses on epistemology and philosophy of science. Also,
> you may want to take a course on the 'softer' side of cognitive
> science. For example, phenomenology is 'hot' in cognitive science
> at the moment (e.g., the 'embodied mind' camp).
>

>[...]

Many of the intro classes I selected, such as Psych I and Zoology, are
either required for other classes or fulfill a general education
requirement. Otherwise I would drop them as all the intro classes I've
encountered so far have been pretty forgetable.

>
> The single most important training experience you can have in
> college is to work under a professor on new research. Find a
> professorial mentor. Then infilitrate the local network of
> graduate students. Get your hands dirty.
>
> Hope this helps. Feel free to email me p
> ersonally should you feel
> the need.
>

Thank you very much for your comments. You have been very helpful. I
haven't had much to go on so far. Mostly I've been trying to adapt other
university's programs to classes offered here without knowing what it
meant. I have more confidence in what I am trying to do now.

Sashank Varma

unread,
Dec 6, 2001, 1:50:31 AM12/6/01
to
In article <%4EP7.2015$c4.60...@ounews.ou.edu>, Ryan Holbrook
<rhol...@mmcable.com> wrote:

>Calculus III is a prerequisite to Linear Algebra, which is required for
>many upper-division CS coures. I replaced the two statistics coures with a
>combined Applied Stastical Methods and added Foundations of Analysis which
>covers topology and is a requirement for other topology and graph theory
>courses in case I decide to continue in that area.

Think carefully before replacing separate probability and statistics
classes with a single one. I've not seen the two subjects combined
profitably in a single semester. And given the pervasiveness of
probabilitistic reasoning in current AI and the statistical basis
of empirical work (i.e., experiments), you'll want to make sure
you've got a grip on both before leaving college.

Although I applaud you for taking analysis, which is just as
important as abstract algebra if you want to keep your mathematica
doors open, which it appears you do.

Have you spoken with sympathetic professors in the various
departments in which you plan to take multiple classes? The
might help you draw a better balance given the vagaries of the
University of Oklahoma.

Erik Naggum

unread,
Dec 6, 2001, 2:54:00 AM12/6/01
to
* cubic...@mailandnews.com (Software Scavenger)

| The nervous system is massively parallel. When will we have computer
| hardware with the same advantage to the same degree?

When humans are willing to accept that a computer may be smarter than
they are. We have a situation today where the opposite willingness
manifests itself in how people respond to a computer that has "insulted"
them by pointing out that they are factually wrong about something or
have made illogical arguments or committed fallacies. A simple statement
of fact that would make nobody defensive in the real world, tends to make
some people massively irritated when their computer is the messenger, and
even simple error messages or failure to do what the user wants, however
stupid or irrational, does not infrequently lead to the untimely demise
of the hardware. If the computer were ever to _volunteer_ criticism to
your average user, even in the form of actually helpful suggestions, you
can be certain that it would be among the very last things it performs.

| Such as making it easier to program the computer by making the computer
| do a larger share of the work of programming.

When I first believed that it would be feasible to write an Emacs on top
of a commercial Common Lisp system (and hoping the non-commercial ones
would catch on as necessary), part of the rationale for wanting to
undertake this herculian task was that it would afford intimate ties
between the Common Lisp world and the source code in progress. Doing
this today, even with Allegro CL's Emacs-Lisp Interface, would drown in
interprocess communication, reading and writing Lisp objects.

///
--
The past is not more important than the future, despite what your culture
has taught you. Your future observations, conclusions, and beliefs are
more important to you than those in your past ever will be. The world is
changing so fast the balance between the past and the future has shifted.

Kenny Tilton

unread,
Dec 5, 2001, 11:41:58 PM12/5/01
to

The above is a good example of how nasty it is to understand
consciousness: a meta-question about whether we are dealing with
missing answers or a bad question. Certainly it would be easier to
define away the question than to answer it. Many a writer on mind has
attempted that bold stroke. The problem is doing so in a way persuasive
to a consciousness.

Anyway, I /thought/ I was simply repeating Ryan's avowed interest, but I
misread. au contraire:

Ryan wrote:
"It is supposed to be a study in the computational modeling of
higher-order cognitive functions (however limited)."

Oops. My bad.

kenny
clinisys

Kenny Tilton

unread,
Dec 6, 2001, 2:00:57 AM12/6/01
to

Software Scavenger wrote:
>
> Kenny Tilton <kti...@nyc.rr.com> wrote in message news:<3C0EABBE...@nyc.rr.com>...
> The nervous system is massively parallel. When will we have computer
> hardware with the same advantage to the same degree?

True (tho who was the guy who put a retina on a chip?) But dividing and
conquering is faster serially or in parallel, so it is worthwhile
attending to the algorithm, hence (IMHO) to the implementation of
organic consciousness. Agreed we need faster hardware, but also faster
algortihms. Especially ones amenable to parallelization.

>
> There is
> plenty of stuff needed which can be done with our present hardware.
> Such as making it easier to program the computer by making the
> computer do a larger share of the work of programming.

That's the justification I use for the runtime cost of the dataflow
engine I have been talking up lately: if this hack makes me much more
productive and the software much more reliable, then it is worth a few
percent extra cycles at run time.

kenny
clinisys

Thomas F. Burdick

unread,
Dec 6, 2001, 6:25:20 AM12/6/01
to
Erik Naggum <er...@naggum.net> writes:

> * cubic...@mailandnews.com (Software Scavenger)
> | The nervous system is massively parallel. When will we have computer
> | hardware with the same advantage to the same degree?
>
> When humans are willing to accept that a computer may be smarter
> than they are. We have a situation today where the opposite
> willingness manifests itself in how people respond to a computer
> that has "insulted" them by pointing out that they are factually
> wrong about something or have made illogical arguments or
> committed fallacies. A simple statement of fact that would make
> nobody defensive in the real world, tends to make some people
> massively irritated when their computer is the messenger

One thing to watch out for when you're talking about "people" is which
people you're talking about. Certainly in the US, there is a large
amount of technophobia, but it's not evenly distributed in the
population -- it's very focused in the petty bourgeoisie (or
white-collar middle-class, or whatever you want to call that group).
In addition to more extreme forms like "Earth Liberation" types, there
is also the *much* more mild technophobia that causes some bourgeois
who own computers to resent, distrust, or dislike them, nonetheless.
I think it's this mild technophobia that's the cause of the irritation
you cite. You don't really see this among poor and working people.
You just don't see them complaining that computers (or TV, or cars,
or...) are somehow, indescribably, but inherently bad. Of course
these are generalizations, and there are certainly exceptions to them,
but I think this is a fair approximation of technophobia in the US, at
a social level. I'll avoid going off topic into a several page essay
theorizing *why* this is the case, and just leave this at the level of
observation.

> | Such as making it easier to program the computer by making the computer
> | do a larger share of the work of programming.
>
> When I first believed that it would be feasible to write an Emacs on top
> of a commercial Common Lisp system (and hoping the non-commercial ones
> would catch on as necessary), part of the rationale for wanting to
> undertake this herculian task was that it would afford intimate ties
> between the Common Lisp world and the source code in progress. Doing
> this today, even with Allegro CL's Emacs-Lisp Interface, would drown in
> interprocess communication, reading and writing Lisp objects.

I'm not quite sure what you mean by "intimate ties between the CL
world and the source code in progress." Do you mean the ability to do
things like structure editing, browsing, and inspection from within
Emacs? Or, what?

--
/|_ .-----------------------.
,' .\ / | No to Imperialist war |
,--' _,' | Wage class war! |
/ / `-----------------------'
( -. |
| ) |
(`-. '--.)
`. )----'

Sashank Varma

unread,
Dec 6, 2001, 12:04:31 PM12/6/01
to
In article <3C0F185B...@nyc.rr.com>, Kenny Tilton
<kti...@nyc.rr.com> wrote:

>True (tho who was the guy who put a retina on a chip?)

Carver Mead?

Nice overview here: http://web.mit.edu/invent/www/Mead.html
Lab web page here: http://www.pcmp.caltech.edu/

Erik Naggum

unread,
Dec 6, 2001, 3:26:50 PM12/6/01
to
* Thomas F. Burdick

| I'm not quite sure what you mean by "intimate ties between the CL world
| and the source code in progress." Do you mean the ability to do things
| like structure editing, browsing, and inspection from within Emacs?

Basically that instead of the editor working with textual representations
of some objects, it would work with the internal representation and the
textual representation would only be for the user.

Sashank Varma

unread,
Dec 7, 2001, 1:56:34 PM12/7/01
to
In article <tKEP7.2017$c4.60...@ounews.ou.edu>, Ryan Holbrook
<rhol...@mmcable.com> wrote:

>I am supposed to find a faculty member to be my advisor and oversee my
>senior project. I am trying to make the program as complete as possible
>before approaching anyone, though, as I am afraid of the very phenomenon
>you mention. While I want to work under someone who has similar interests
>as mine, I realize that they will not be identical and I don't want to be
>coerced (even unintentionally) into a course of study that I would not
>otherwise be happy with. This way, I can get a variety of opinions before
>feeling committed to anything.

This is a good philosophy to begin with. But in all likelihood
you will at some point have to bend your education to the will
of others. So when the time comes, be flexible. Better this
than going it alone, re-inventing a lot of wheels, and never
getting feedback from a particular discipline. You want
competency in a number of trades, but you should also strive
for mastery in at least one. This will also help if you decide
to apply for grad school.

>All of the advanced classes in AI are graduate level, so I may have some
>trouble getting in to them. I'm not sure how strict the department is on
>their policies.

This is worth investigating. Some departments are flexible,
others not. This would not have been an option where I went
as an undergrad, but is fine where I attend grad school.

>I don't believe OU offers computational linguistics, though that is
>certainly something I could pick up through supervised independant study.

If you do, you'll learn something of current syntactic theories,
parsing, and probability in one setting. Allen's "Natural
Language Understanding, 2nd Edition" is okay, but Dan Jurafsky's
new book is much better (and of course more current). Parts of
it may be digestable as part of an independent study. See Amazon
for reviews from, among others, Peter Norvig! (Search Books for
"Jurafsky".)

>Calculus III is a requirement of many of the CS classes but I did decide to
>drop calc IV. I've gotten varied suggestions about group theory. Some like
>it, some don't, so I'll have to ask around a little more.

Abstract algebra and analysis are the capstones of an undergrad
math major. (Perhaps topology as well.) The question is whether
they are on your critical path. Worry about their inclusion later,
after you've precisely defined this path. If you wind up leaning
towards logic and computation, they may be. If you wind up leaning
towards cognition and neursocience, they may not be.

>Many of the intro classes I selected, such as Psych I and Zoology, are
>either required for other classes or fulfill a general education
>requirement. Otherwise I would drop them as all the intro classes I've
>encountered so far have been pretty forgetable.

As an undergrad, I found it possible to talk departments into
substituting advanced classes for intro classes when counting
core requirements. Give it a shot; you've got nothing to lose
and precious free class hours to gain! Wouldn't you prefer
an extra class on cognition or neuroscience to Psych 101, or
an upper-division class on ethology to a survey of zoology?

Jacek Generowicz

unread,
Dec 10, 2001, 4:29:52 AM12/10/01
to
Ryan Holbrook <rhol...@mmcable.com> writes:

> A preliminary proposal is on my web page in postscript and pdf
> format:
>
> http://students.ou.edu/H/Ryan.P.Holbrook-1/

Small error in your bibliography: Dennett is NOT and autor of Godel,
Esher, Bach, but he IS an editor of The Mind's I.

Ryan Holbrook

unread,
Dec 9, 2001, 8:29:16 PM12/9/01
to
Jacek Generowicz wrote:

Oops. Thanks for pointing that out. (I should hire an editor!)

Fetter Heiko

unread,
Dec 10, 2001, 6:19:52 PM12/10/01
to
?

Fetter.Häiko

unread,
Dec 10, 2001, 8:52:13 PM12/10/01
to
:o)
0 new messages