Mild Direct Perception versus Strong Representationalism
Marina Rakova
questions, but I would really appreciate if Steven and everybody else
who has a stake in the debate (especially Professor Brook) could let me
know whether the summary of contrasts I am trying to present below is
correct with respect to their views.
Labelling the positions
For the present purposes,
MDP - mild direct perception (the view held by Andrew Brook and
arguably Fred Dretske and Michael Tie).
SR - strong representationsalim (the view held by Steven Lehar; hence,
the word 'representationalism' is used here with a different meaning
from what seems to be common in philosophy of mind).
Main commitments and constraints on theories of perception
Prerequisite: the expression 'real world' will be substituted with 'an
organism's environment' (will this be an acceptable move to a proponent
of MDP?)
MDP - organisms as a rule function successfully in their environments
(don't bump into chairs, find food, etc.).
SR - organisms can act (or interact with the environment) only out of
what they represent.
So far I personally see no contradiction between MDP and SR
Methodological commitments
MDP - the explanation of normal perception has priority and normal
perception itself is a mystery.
SR - a theory which cannot make sense of how error is possible has no
explanatory validity.
Both are needed I believe, and don't need to be contradictory.
Hidden commitments
MDP - we should stick with human beings; the way the world appears to
us can be presumed to be the real world with the real objects and
properties with which we interact; the reason for this presumption is
that unless we have a fixed framework of what is external to one's mind
we cannot even describe what's happening with other creatures (we need
unchangeable reference points to talk about perception; and it's not
easy for us to talk about perception in any other way than phrasing the
discussion in terms of objects and properties of our environment).
SR - what is in an organism's environment is determined by what this
organism is capable of representing, so the assumption of the real
world does no explanatory work (chairs are there in the world for
myself and my cat and one can see similarities in our modes of
interaction with them, but flies don't interact with them in a similar
way and paramecia wouldn't so explaining what flies perceive in terms
of chairs and desks would throw no light on the nature of perception).
But even here I don't see any real contradiction. MDP has the option of
saying that a fly perceives a chair as a '_' (description inserted),
and SR is committed to the interaction with the environment which can
be productively studies at different levels of description.
Representation
MDP - representation is always representation of. MDP says that
representation is representation of things outside organisms' minds.
(What does SR say on this reading of representation? That
representation is representation of sensory input on the retina?)
SR - representation is always representation full stop. It's an entity
within an organism's mind on which computations can be carried out. (As
long as MDP is committed to representations, it can drop the 'of').
Again, I see no contradiction.
Experience
MDP - experience is always subjective; and so for organisms in their
experience there is direct (unmediated) contact with the world - this
is the most theoretically interesting point.
SR - experience is constructed out of brain's representations - the
most interesting point about it is how the construction is done and how
it produces the feeling of direct contact.
To quote from Steven Lehar:
> In fact, the representationalist concept
> explains *HOW* the organism interacts with the environment so as to
> feel as
> if it were in direct perceptual contact with the world, it does so by
> building a world in its brain, the very part of the problem that
> remains
> deeply mysterious in the direct perception view.
>
Still, I see no contradiction as long as one doesn't equate here
experience and qualitative consciousness.
Perception
MDP: perception is always perception of something, but perception is
needed for action, we act in the external world, so perception must be
perception of the external world. Perception is direct in the sense of
being perception of the world.
SR: perception is always perception of something, but perception can
only be implemented in biological systems in a certain constrained way
via representations of increasing complexity, so perception must be
perception of our constructed representations of the world. Perception
is indirect because our access to the world is our access to our
representations of it.
But up to this point there were no real contradictions, why should they
arise at this stage? Or rather, why should the word 'direct' cause so
much trouble? Doesn't the whole debate rest on the conflation of the
issue of the function of perception or feature-detection in
organism-environment interaction and the way this function is
implemented in biological systems? And hence on 'perception' being used
in slightly different senses or with slightly different emphases (what
an organism perceives subjectively vs. what is the perceptual
experience out of which an organism acts as a complex physical entity).
But then it cannot an ontological or an epistemological debate but
rather the debate about explanatory hierarchies, which is very
important but which doesn't need any formulation in terms of 'direct'
vs 'indirect', can it not be both given different priorities in
explanation, so that when one ultimately wants to explain perception
one needs to keep in sight both perspectives? For how could the debate
about 'directness'-'indirectness' per se (unlike what I qualify as
hidden assumptions) have any empirical outcome? So, I would tentatively
say that we normally perceive the world, not our representations, but
in so far as what constitutes our world of objects and properties is
determined by what we are capable of representing. (This, though,
requires another qualification - the word 'normally' cannot be left
unclarified, as in totally novel situations we perceive our
representations which are not object-dependent - but MDP makes an
important point when it raises the issue of normal perception - why
should there be object-integration for an organism to successfully
interact with its environment?, - still MDP cannot be a complete theory
of perception because it cannot be implemented in a physical system).
And another worry: does not the issue of perception being direct or
indirect arise only because when we are conscious of perceiving
something we are certain we are perceiving real world objects (could my
cat get the point of the debate?)? But is there nothing interesting to
say about perception when we subtract consciousness? Does not the
debate rest on the implicit distinction between understanding
perception as conscious perception (in our familiar feeling of
consciousness) and perception unqualified as conscious or unconscious?
(Consciousness would also create additional problems of distinguishing
between the perception of objects and the phenomenology of perception).
But if so, no agreement and no argument is possible, for one isn't even
talking about the same thing.
Please let me know if I go badly wrong somewhere, I would really
appreciate that.
Thank you,
Marina Rakova
Andrew Brook
I can't comment on everything in your long and interesting message but a few
points:
Marina Rakova wrote:
> Main commitments and constraints on theories of perception
> Prerequisite: the expression 'real world' will be substituted with 'an
> organism's environment' (will this be an acceptable move to a proponent
> of MDP?)
Yes, what we are directly aware of is usually our immediate environment.
> MDP - organisms as a rule function successfully in their environments
> (don't bump into chairs, find food, etc.).
>
> SR - organisms can act (or interact with the environment) only out of
> what they represent.
>
> So far I personally see no contradiction between MDP and SR
Nor do I
> Methodological commitments
>
> MDP - the explanation of normal perception has priority and normal
> perception itself is a mystery.
>
> SR - a theory which cannot make sense of how error is possible has no
> explanatory validity.
>
> Both are needed I believe, and don't need to be contradictory.
Again, I agree.
> Hidden commitments
>
> MDP - we should stick with human beings; the way the world appears to
> us can be presumed to be the real world with the real objects and
> properties with which we interact; the reason for this presumption is
> that unless we have a fixed framework of what is external to one's mind
> we cannot even describe what's happening with other creatures (we need
> unchangeable reference points to talk about perception; and it's not
> easy for us to talk about perception in any other way than phrasing the
> discussion in terms of objects and properties of our environment).
>
> SR - what is in an organism's environment is determined by what this
> organism is capable of representing, so the assumption of the real
> world does no explanatory work (chairs are there in the world for
> myself and my cat and one can see similarities in our modes of
> interaction with them, but flies don't interact with them in a similar
> way and paramecia wouldn't so explaining what flies perceive in terms
> of chairs and desks would throw no light on the nature of perception).
>
> But even here I don't see any real contradiction. MDP has the option of
> saying that a fly perceives a chair as a '_' (description inserted),
> and SR is committed to the interaction with the environment which can
> be productively studies at different levels of description.
Not sure what the issue is but I don't see any obvious problems.
>
> Representation
>
> MDP - representation is always representation of. MDP says that
> representation is representation of things outside organisms' minds.
> (What does SR say on this reading of representation? That
> representation is representation of sensory input on the retina?)
>
> SR - representation is always representation full stop. It's an entity
> within an organism's mind on which computations can be carried out. (As
> long as MDP is committed to representations, it can drop the 'of').
>
> Again, I see no contradiction.
I agree that Steve has so far failed to make a distinction between the vehicle
or representation (which is in the brain) and the object of representation
(which if often something outside the head), so asking him what he thinks
representation is of is a very good question to ask.
> Experience
>
> MDP - experience is always subjective; and so for organisms in their
> experience there is direct (unmediated) contact with the world - this
> is the most theoretically interesting point.
I agree.
> SR - experience is constructed out of brain's representations - the
> most interesting point about it is how the construction is done and how
> it produces the feeling of direct contact.
>
> To quote from Steven Lehar:
>
>> In fact, the representationalist concept
>> explains *HOW* the organism interacts with the environment so as to
>> feel as
>> if it were in direct perceptual contact with the world, it does so by
>> building a world in its brain, the very part of the problem that
>> remains
>> deeply mysterious in the direct perception view.
>>
> Still, I see no contradiction as long as one doesn't equate here
> experience and qualitative consciousness.
Again, I think your position should be the right one. However, Steve, as I
understand him, thinks that if representations are in the brain, if we have a
role in constructing them, etc., then, for some reason that he has never been
able to articulate, representations just *cannot* be of, cannot represent,
cannot make us directly aware of, anything outside the brain. If he dropped this
idea, then he and I would agree on most everything.
> Perception
>
> MDP: perception is always perception of something, but perception is
> needed for action, we act in the external world, so perception must be
> perception of the external world. Perception is direct in the sense of
> being perception of the world.
>
> SR: perception is always perception of something, but perception can
> only be implemented in biological systems in a certain constrained way
> via representations of increasing complexity, so perception must be
> perception of our constructed representations of the world. Perception
> is indirect because our access to the world is our access to our
> representations of it.
Yes, this is Steve's position as I understand it (see last comment). Against it,
I would say two things. First, we cannot literally *perceive" anything in our
head because there is no light there. But this is fairly trivial because Steve
could just substitute some other kind of (direct) awareness. The second point is
the important one. It is that our access to the world is via our representations
but is not limited to *access to* them. The vehicles are in us but what they
give us access to is the whole great world outside us.
>
> But up to this point there were no real contradictions, ...
But there is. See my last two comments.
why should they
> arise at this stage? Or rather, why should the word 'direct' cause so
> much trouble? Doesn't the whole debate rest on the conflation of the
> issue of the function of perception or feature-detection in
> organism-environment interaction and the way this function is
> implemented in biological systems? And hence on 'perception' being used
> in slightly different senses or with slightly different emphases (what
> an organism perceives subjectively vs. what is the perceptual
> experience out of which an organism acts as a complex physical entity).
> But then it cannot an ontological or an epistemological debate but
> rather the debate about explanatory hierarchies, which is very
> important but which doesn't need any formulation in terms of 'direct'
> vs 'indirect', can it not be both given different priorities in
> explanation, so that when one ultimately wants to explain perception
> one needs to keep in sight both perspectives? For how could the debate
> about 'directness'-'indirectness' per se (unlike what I qualify as
> hidden assumptions) have any empirical outcome? So, I would tentatively
> say that we normally perceive the world, not our representations, but
> in so far as what constitutes our world of objects and properties is
> determined by what we are capable of representing.
No problem with this!
(This, though,
> requires another qualification - the word 'normally' cannot be left
> unclarified, as in totally novel situations we perceive our
> representations which are not object-dependent - but MDP makes an
> important point when it raises the issue of normal perception - why
> should there be object-integration for an organism to successfully
> interact with its environment?, - still MDP cannot be a complete theory
> of perception because it cannot be implemented in a physical system).
It clearly can because it is implemented in us and we are a physical system (he
said dogmatically). I think a major issue here is that what consciousness makes
us conscious of and how the brain manages to carry this out are two very
different questions. The first is to be settled by commonplace observations and
some conceptual analysis. The second is to be settled by science. Steve is sure
that the science shows the commonplace observations to be false. I claim that
science does not touch them; nor could it if my conceptual analysis is right.
About the science itself, I disagree with Steve about nothing.
> And another worry: does not the issue of perception being direct or
> indirect arise only because when we are conscious of perceiving
> something we are certain we are perceiving real world objects (could my
> cat get the point of the debate?)? But is there nothing interesting to
> say about perception when we subtract consciousness? Does not the
> debate rest on the implicit distinction between understanding
> perception as conscious perception (in our familiar feeling of
> consciousness) and perception unqualified as conscious or unconscious?
> (Consciousness would also create additional problems of distinguishing
> between the perception of objects and the phenomenology of perception).
> But if so, no agreement and no argument is possible, for one isn't even
> talking about the same thing.
Don't follow you here. It might help a bit if I tell you that I would say that
nonconscious representations often directly represent the world around us, too.
All the best,
Andrew
--
Andrew Brook, Professor of Philosophy
Director, Institute of Cognitive Science
Member, Canadian Psychoanalytic Society
2217 Dunton Tower, Carleton University
Ottawa ON, Canada K1S 5B6
Ph: 613 520-3597
Fax: 613 520-3985
Web: www.carleton.ca/~abrook
Steven Lehar
Rakova >
why should the word 'direct' cause so much trouble? Doesn't the whole
debate rest on the conflation of the issue of the function of perception or
feature-detection in organism-environment interaction and the way this
function is implemented in biological systems? And hence on 'perception'
being used in slightly different senses or with slightly different
emphases ...
But then it cannot an ontological or an epistemological debate
< Rakova
No, it is much deeper than that, it is indeed an ontological AND an
epistemological debate.
In computational terms the deep question is whether perception detects
regularities out in the world, or whether it recreates the external world
in an internal replica, and detects those regularities in the replica. It
is a question of whether perception is principle *detection*, or is it
*construction*. These are two completely different strategies.
The issue is epistemological in the sense that awareness, which is a form
of knowledge, of the outside world can only come to us through the vehicle
of internal representations of it, and therefore any form of direct
perception that posits features in experience that are not explicitly
represented in the brain, has a profound epistemological hole in it.
The issue is ontological, because it identifies the vivid spatial structure
of visual experience with either actual objects and surfaces in the world,
or with internal representations in the brain. That is quite a profound
difference.
In fact there can hardly be a more profound difference in philosophies than
to assume you can experience the world directly, as opposed to that your
entire world of experience is inside your head. That is an epistemological
and ontological inversion of the highest order.
It is an issue which absolutely must be gotten straight if we are ever to
make any real progress in understanding the brain.
Steve
Jeff Dalton
> why should the word 'direct' cause so much trouble?
Why should the word "indirect" cause so much trouble?
This whole discussion is going nowhere, slowly.
In the hope of at least clarifying some things, I'd like to ask
the two sides a couple of questions.
A while back GILBERTO GOMES wrote:
If 'directly' means 'noninferentially', and if 'noninferentially'
means 'not through a conscious inference', one can hardly disagree
with direct realism.
To Steven Lehar and other representationalists:
If "direct perception" were defined (tendentiously?) as above --
not through a conscious inference -- would you agree with
"direct perception" so defined?
To Andrew Brook and other direct perceptionists:
Is that what you mean by "direct": that it's not through a
*conscious* inference?
If not, what is your criteria for deciding whether something
done in our brains, nervous system, or sense organs counts as
"inference"?
In some cases, a conscious inference *is* necessary. I might see
something in the distance in poor light, for example, and not be
able to tell whether it's an animal or a log. However, by paying
close attention and thinking about it, I might be able to determine
that it is an animal.
(Note BTW that poor conditions (distance, low light) are not
necessary features of such cases. For example, some animals
are difficult to detect against certain backgrounds even when
the light is good and they are not obscured by other objects.)
To Andrew Brook and other direct perceptionists:
Would you consider such cases indirect, so that we do sometimes
perceive indirectly, or are they still somehow "direct" despite
involving an inference (indeed, a conscious inference).
If perception is sometimes clearly indirect, why doesn't that
show that out perceptual systems implement indirect perception?
-- Jeff
Neil W Rickert
JD>A while back GILBERTO GOMES wrote:
GG> If 'directly' means 'noninferentially', and if 'noninferentially'
GG> means 'not through a conscious inference', one can hardly disagree
GG> with direct realism.
JD>To Andrew Brook and other direct perceptionists:
JD> Is that what you mean by "direct": that it's not through a
JD> *conscious* inference?
That is certainly an important part of it.
JD>In some cases, a conscious inference *is* necessary. I might see
JD>something in the distance in poor light, for example, and not be
JD>able to tell whether it's an animal or a log. However, by paying
JD>close attention and thinking about it, I might be able to determine
JD>that it is an animal.
This does not make a good example. For, while paying close
attention, you are also gathering additional information. There may
indeed be inference involved. But there is also additional
perception. I would consider the inference to be outside of, and
additional to the perception.
The decision that it was an animal may have involved inference, but
that inference was not a part of perception. This example does not
contradict directness of perception (in my opinion).
Let's try a different example. When first learning to drive a car,
it was difficult to judge whether I was properly in lane. I would
maybe look across the hood, to see where my line of vision struck the
road. That involves additional perception (getting additional
information) and inference. But now, as an experienced driver, I no
longer need to do that. I can immediately recognize where I am
positioned in the lane.
Representationalists tend to take the view that the same inference is
going on, but it has been somehow moved to an unconscious inference.
My view, by contrast, is that the neural system has rewired itself to
directly extract that information, so that no inference is required.
For comparison, I also consider that no inference or computation is
being done by a traditional mercury thermometer -- rather it directly
reports the temperature information.
-NWR
Andrew Brook
inference from anything else of which one is conscious, slightly
different. Seeing the words unfold on the screen in front of me is a
clear case. Seeing an image on a TV and inferring from it it that
someone is crossing the yard outside the building is a clear case of
indirect. Seeing a distant object in a bad light and making educated
guesses about what it is would be an inbetween case about which I don't
have any particular view.
That said, my definition, though the best I have ever been able to come
up with, is not as clear an anchor for the concept as clear examples are
('paradigm cases' in an old jargon). I see the leaf on the plant in
front of me waving in the breeze. It is the leaf of which I am aware,
nothing else (certainly nothing at the inner end of the process of
perceiving it). We know what direct perception is much more by
experiencing it than by definition. It is one of those concepts that is
very difficult to define discursively, but easy to define by example
(ostensively). Time is another such concept and some people think that
consciousness is a third, though about the latter I disagree.
About such examples/paradigm cases, I am inclined to say this: that is
what direct perception *is* -- however this result is achieved, this
situation is what we mean by the concept. Why? Because it is impossible
to define any notion of 'direct' according to which this comes out to be
indirect. This is as direct not only as it gets but as we can conceive.
That's why I keeping asking Steve Lehar, when he insists that even this
is indirect, to tell us what he means by that little word 'direct' -- an
invitation that he has never taken up.
Jeff Dalton
> Jeff has got it close to right. My definition of 'direct' is: not by
> inference from anything else of which one is conscious, slightly
> different.
Do you mean only conscious inference, or would some unconscious
processes also count as inference?
> Seeing the words unfold on the screen in front of me is a
> clear case. Seeing an image on a TV and inferring from it it that
> someone is crossing the yard outside the building is a clear case of
> indirect. Seeing a distant object in a bad light and making educated
> guesses about what it is would be an inbetween case about which I don't
> have any particular view.
I'm not sure that you should count seeing the image on a tv
as indirect. In the "traditional" debate about sense-data,
I think that perceiving an intermediate *image* would make
perception indirect, but your focus is on inference instead.
Sure, you described it as seeing an image and inferring, but
is an explicit, conscious inference normally involved? Suppose
a security guard is watching a tv monitor and sees someone enter
the building. Isn't it natural to say they saw someone enter
the building?
Of course, often an image on a tv screen isn't of something that
is actually happening in the world; but we can usually distinguish
those cases, just as we can usually distinguish hallucination.
In any case, direct perception shouldn't require that what's
perceived be happening now, because with vision, at least,
there's always a time delay.
So I think it's still not quite clear where you (should?) draw
the line between direct and indirect.
> That said, my definition, though the best I have ever been able to come
> up with, is not as clear an anchor for the concept as clear examples are
> ('paradigm cases' in an old jargon). I see the leaf on the plant in
> front of me waving in the breeze. It is the leaf of which I am aware,
> nothing else (certainly nothing at the inner end of the process of
> perceiving it).
But how do you know you're not aware of something at the inner
end of a process? It seems to me that something at the inner
end is exactly what I'm aware of.
A problem for direct perception theories is always going to be:
if perception is direct, why isn't naive realism true? How can
my visual system present so much that is so different from what
is actually the case if I'm seeing directly, rather than
experiencing the end result that my visual system provides?
> About such examples/paradigm cases, I am inclined to say this: that is
> what direct perception *is* -- however this result is achieved, this
> situation is what we mean by the concept. Why? Because it is impossible
> to define any notion of 'direct' according to which this comes out to be
> indirect. This is as direct not only as it gets but as we can conceive.
But to say what you mean by the concept, you simply have to explain
what you *mean*. You don't have to show that your meaning is in some
sense the only meaning that it's reasonable or possible to have.
So the "impossible to define" argument is either redundant or
else it's serving some other purpose than saying what you mean
by the concept.
Indeed, you seem to be arguing that we ought to use the word
"direct" here because there's no more legitimate use. But that's
not how things normally go. For example, if libertarian free
will is impossible (or incoherent), that doesn't mean that we
should stop using the term "free will" in that case. If substance
dualism is impossible (or incoherent), that doesn't mean we should
call some other view "substance dualism".
> That's why I keeping asking Steve Lehar, when he insists that even this
> is indirect, to tell us what he means by that little word 'direct' -- an
> invitation that he has never taken up.
I suspect that, like me, he regards direct perception as impossible
(unless there's some explanation that shows it is possible after all).
In my view, direct perception would be what we'd have if all the
things that make perception indirect were absent. So direct visual
perception would be perception that was not via (or not affected
in any significant way by being via) light, the eye, the optic
nerve, etc. That is: that's what ought to be called "direct".
Using "direct" for anything else is misleading, IMO.
Really, two issues ought to be separated: (1) are certain theories
of perception correct?, and (2) should we use the word "direct"?
In this discussion, you seem relatively uninterested in (1) ("however
this result is achieved"), and to be focusing on (2). I find
that puzzling. Why does it matter so much that we use the word
"direct"?
-- Jeff
Marina Rakova
disagreement. It seems to me that Steve has not given *direct* answers
to them. It would be nice, if he could. (Unlike Jeff I believe the
debate might be leading somewhere as long as one uses the terms with
the same meaning and tackles the points the other side raises instead
of drifting off somewhere else).
A change in terminology:
direct representationalism = Andrew's position
indirect representationalism = Steve's position
1) Direct representationalism is supported by viewing representation as
'representation of' (as a relational rather than an intrinsic
property), the view arrived at by conceptual analysis.
Question(s) to Steve: can you agree that representation is
representation of? do you accept the distinction between contents and
vehicles of representations? If so, what are representations
representations of?
(In fact Steve comments on these issues in his response to Andrew, but
I think his comments there swing the pendulum towards direct
representationalism.)
2) What does science say?
Andrew: science is indifferent on the issue of direct
representationalism versus indirect representationalism, because the
issue is conceptual, not empirical.
Steve: science undermines direct representationalism.
I derive this from the following in Andrew's message:
> I think a major issue here is that what consciousness makes
> us conscious of and how the brain manages to carry this out are two
> very
> different questions. The first is to be settled by commonplace
> observations and
> some conceptual analysis. The second is to be settled by science.
> Steve is sure
> that the science shows the commonplace observations to be false. I
> claim that
> science does not touch them; nor could it if my conceptual analysis is
> right.
> About the science itself, I disagree with Steve about nothing.
>
But doesn't science operate on the assumption that we perceive external
objects? How else could one detect abnormalities of perception if
something weren't taken to be the normal case? (To illustrate what I
mean with an analogy: in the discussion of placebo effects there is an
implicit assumption - something maybe that is used as a shorthand for
something more scientific but nonetheless well within the province of
the naive world view - that people's thoughts are causally efficacious.)
A question to all: What does mainstream neuroscience really say on the
issue, if anything?
Marina
Jeff Dalton
> Thank you to Andrew for bringing into focus the major points of
> disagreement. It seems to me that Steve has not given *direct* answers
> to them. It would be nice, if he could. (Unlike Jeff I believe the
> debate might be leading somewhere as long as one uses the terms with
> the same meaning and tackles the points the other side raises instead
> of drifting off somewhere else).
After I had about a week away, when I started reading PSHYCE-D
again, the discussion looked the same as before.
However, it may be possible at least to clarify what the positions
are. That's still needed, because it's still unclear exactly what
Andrew Brook's position is or why he doesn't think its been refuted.
For example, you now highlight as one of the major points of disagreement
the question of whether the representations are representations *of*
the things in the world. But surely indirect perception can allow
that indirect perception of things is still perception of things,
and that representations are representations of things in the world.
Moreover, the issues had seemed to be different ones.
In a recent message, Andrew said he had
1. ... defined the position. Direct perception is being aware
of objects in the world, nothing short of them, and not by
inference from anything else of which I am aware. (We can be
directly aware of other things, too, including our own
representations, ourselves, and our bodily states. But direct
awareness of the world around us is the crucial case.)
The main issues there seemed to be whether there was an intermediate
object of perception and whether inference was involved, with nothing
about whether representations were 'representation of'.
Note moreover that direct perception required that three things
*all* be true. If, say, there's awareness of something "short
of them", direct perception cannot be saved by showing that
inference isn't involved. That is the move that is usually
made, but it cannot work unless the definition is abandoned.
Another problem is that Andrew has also said he has
2. ... given examples. If Steve is reading this, the state he
is in a a perfect example of direct awareness. He is aware of
the words on the screen directly, not by inference from anything
else of which he is aware. And he is aware of the words on the
screen, not any intermediary. He won't accept this as an example
but has given no reason for this refusal.
But that's building Andrew's direct realism into the description.
An advocate of indirect perception would deny that one is directly
aware of the words in the very same case.
Yet Andrew says the directness is independent of how the result
is achieved! For example, here:
About such examples/paradigm cases, I am inclined to say this:
that is what direct perception *is* -- however this result is
achieved, this situation is what we mean by the concept.
How can that be? If, for example, inference turns out to actually
have a certain role in the paradigm case, then by Andrew's own
definition, the paradigm case is a case of *indirect* perception.
And that, in turn, creates a problem for one of Andrew's other
points. The above paragraph continues:
Why? Because it is impossible to define any notion of 'direct'
according to which this comes out to be indirect. This is as
direct not only as it gets but as we can conceive.
But Andrew's definition gave us another way to establish that
perception is indirect. If, for example, perception in ordinary
or paradigm cases turns out to be awareness of objects in the
world "by inference from anything else of which I am aware",
then it's indirect regardless of what notions we can define
as "direct".
A similar point apples if we are merely aware of something
"short of" objects in the world (regardless of whether inference
is involved).
Andrew sometimes expains his view by saying one is aware of
objects in the world and not of something "at the inner end
of the process of perceiving [an object]". On the other hand,
he acknowledges that we can be aware of our own representations.
The problem here is that colour, for example, might be something
added by our visual system rather than a property of objects out
there in the world. Then we'd be aware of something at the
inner end most of the time.
That this is a real problem for Andrew's view was brought out
by a message from Michael Baggot who asked:
Just what is it that you see when you look through a
representation [Andrew's direct case] as opposed to what
you see when you look at a representation?
Andrew replied:
When you are aware of the representation, you are aware of seeing
something clearly or obscurely, in focus or out of focus, vividly
or dimly, and so on. Whe you are aware of what the representation
represents or presents, you are aware of the little red chair.
But I, at least, am often aware of both. For instance, right now
while looking at the screen I am also aware of a telephone off to
the left *and* of how fuzzy my perception of it is.
So I'm aware of the representation and only via that of the
object. That's enough to make my perception indirect regardless
of whether I am strictly speaking using inference to get to
the object. To avoid this refutation, Andrew's position has
to change, and indeed it does as the same message continues:
My point against indirect representationalists is that here
it is the chair that you are aware of (under some description,
only certain properties, etc., etc.), not some intermediary,
for example an image or data-structure in the brain, from
which you infer that a chair is out there somewhere ...
Before, three things needed to be true: to be direct
perception must be awareness [1] "of objects in the world,
[2] nothing short of them, and [3] not by inference from
anything else of which I am aware).
Now, an "intermediary" means indirect only if it is an
an intermediary "from which you infer ...".
So inference is the key. Or is it?
-- Jeff
Steven Lehar
Dalton >
To Steven Lehar and other representationalists:
If "direct perception" were defined (tendentiously?) as above --
not through a conscious inference -- would you agree with
"direct perception" so defined?
< Dalton
Negatory.
It has nothing to do with conscious inference or not. When we see a table
in front of us, its identity as a table may be an inference, but its
spatial structure, spatial extent, and spatial form, appear to us
immediately and in parallel *as if* we were seeing it directly out there,
with no inference involved. Nevertheless, what we think we are seeing "out
there" is actually an image "in here" inside our brain.
Steve
Steven Lehar
Rakova >
It seems to me that Steve has not given *direct* answers
to them. It would be nice, if he could.
< Rakova
Ok, all right! I answered what I thought was the *key* issue at the center
of the debate,
( http://listserv.uh.edu/cgi-bin/wa?A2=ind0505&L=psyche-d&D=0&P=7601 )
I am not sure whether these other peripheral issues are so important, but I
will go through the list if you insist. (Lurkers be warned: this might get
tedious!)
Rakova >
SR - strong representationsalim (the view held by Steven Lehar; hence,
the word 'representationalism' is used here with a different meaning
from what seems to be common in philosophy of mind).
<...later...>
indirect representationalism = Steve's position
< Rakova
I will accept this for the purposes of the present discussion, but although
I think Andrew's position is accurately described as "direct
representationalism", my own position is just plain *representationalism*,
pure and simple, unadulterated, as defined in Websters, ignoring Tye and
Dretsky's corruption of the term to mean something *other* than
representationalism.
Rakova >
MDP - organisms as a rule function successfully in their environments
(don't bump into chairs, find food, etc.).
SR - organisms can act (or interact with the environment) only out of
what they represent.
So far I personally see no contradiction between MDP and SR
< Rakova
Agreed!
Rakova >
Methodological Committments:
MDP - the explanation of normal perception has priority and normal
perception itself is a mystery.
SR - a theory which cannot make sense of how error is possible has no
explanatory validity.
Both are needed I believe, and don't need to be contradictory.
< Rakova
Its more serious than that. Direct perception ignores illusion and
hallucination, or they attempt to redefine them out of existence by denying
that illusions can have a spatial structure that is explicitly experienced.
Although the proponents deny it, visual illusions and hallucinations are a
fatal blow to direct perception.
Rakova >
Hidden commitments:
MDP - we should stick with human beings;
But even here I don't see any real contradiction.
< Rakova
Yes, this is key! They refuse to discuss artificial vision at any length or
detail because robots make a mockery of the concept of direct perception by
demonstrating that it is *either* impossible in principle, *or* it is
indistinguishable from representationalism.
Rakova >
Representation:
MDP - representation is always representation of.
SR - representation is always representation full stop.
Again, I see no contradiction.
< Rakova
As long as we stick to veridical perception there is no contradiction. In
the case of illusions and hallucinations however, there is nothing for the
representations to be *of* except of themselves, so MDP collapses
catastrophically.
Rakova >
Experience:
MDP - experience is always subjective; and so for organisms in their
experience there is direct (unmediated) contact with the world - this
is the most theoretically interesting point.
SR - experience is constructed out of brain's representations - the
most interesting point about it is how the construction is done and how
it produces the feeling of direct contact.
Still, I see no contradiction as long as one doesn't equate here
experience and qualitative consciousness.
< Rakova
There is a *profound* contradiction here, that requires us to deny
that "experience" is the same as "qualitative consciousness". MDP requires
that we adopt the *category error* that the world of experience is
ontologically equal to the world that it represents. You can't get more
profound than that with ontological errors. Big contradiction!
Rakova >
Perception:
MDP: perception is always perception of something, but perception is
needed for action, we act in the external world, so perception must be
perception of the external world.
SR: perception is always perception of something, but ... Perception
is indirect because our access to the world is our access to our
representations of it.
But up to this point there were no real contradictions,
< Rakova
**No real contradictions?** "perception must be perception of the external
world" and "perception is indirect [through] our representations of it"
There could not be two more profoundly contradictory statements!
Steve
Steven Lehar
Rakova >
1) Direct representationalism is supported by viewing representation as
'representation of'
Question(s) to Steve: can you agree that representation is
representation of? do you accept the distinction between contents and
vehicles of representations? If so, what are representations
representations of?
< Rakova
When perception is veridical (non-illusory) then the experience of
perception represents the object you believe to be seeing, it is an
experience *of* that object. But, as when watching TV, that experience is
brought to you courtesy of your representational machinery, the thing that
allows you to see colored volume embedded in a spatial void. It is the
modulations of that mechanism that give you experience *of* objects.
There is indeed a distinction between the vehicle and the contents. On the
phenomenal side the vehicles are the dimensions of conscious experience,
that finite expanse of three-dimensional space any point of which can
appear in any (hue, brightness, saturation) color. The content of
experience is the actual pattern of colored objects currently represented
in the spatial void of experienced space.
This correlates exactly with information theory, where information is
carried by modulations of some carrier. The carrier is the vehicle, and the
signal it carries is the contents.
Rakova >
2) What does science say?
...
A question to all: What does mainstream neuroscience really say on the
issue, if anything?
< Rakova
I think it has become clear in this exchange that science favors
representationalism, because that is the only theory that has ever been
formulated with enough precision as to make testable predictions of future
experiments, or to construct an actual model of the concept. Direct
perception is more a belief than a science.
As to what *mainstream* neuroscience says, thats another issue. For the
longest time this question was simply ignored, it was off limits for
scientific debate. But it is encouraging to see that this attitude is
changing, although not without a lot of resistance.
In my article to the Behavioral & Brain Sciences
http://cns-alumni.bu.edu/~slehar/webstuff/bubw3/bubw3.html
there were 11 open peer commentaries opposed to the representationalist
thesis, 6 in support of it, and 5 who were neither explicitly opposed nor
supportive.
I think the average consensus view is 1: most people have not given it a
lot of thought, even people in psychology, philosophy, and neuroscience
(!!!) 2: If they had to choose, they favor a mild form of direct
representationalism, that fits in with their knowledge of physiology
without rudely violating their naive realist assumptions. I thought
commentator Nial McLaughlin represented accurately the consensus view in
neuroscience to the representationalist thesis:
http://cns-alumni.bu.edu/~slehar/webstuff/bubw3/commentaries.html#McLoughlin
to which my response was:
http://cns-alumni.bu.edu/~slehar/webstuff/bubw3/response.html#McLoughlin
But there's nothing in there that you haven't heard right here on PSYCHE-D!
Steve
Marina Rakova
Comment on two messages by Steve Lehar:
> In computational terms the deep question is whether perception detects
> regularities out in the world, or whether it recreates the external
> world
> in an internal replica, and detects those regularities in the replica.
> It
> is a question of whether perception is principle *detection*, or is it
> *construction*. These are two completely different strategies.
This might well be the hidden assumption. Can one say that:
- direct representationalism may be supported by the view of perception
as feature detection
- indirect representationalism denies perception as feature detection,
but views perception as feature construction.
However, you say that perception detects features in a replica. But if
perception is still basically feature detection, then what is the point
of the construction stage?
What is the mainstream view on this in science circles?
> The issue is epistemological in the sense that awareness, which is a
> form
> of knowledge, of the outside world can only come to us through the
> vehicle
> of internal representations of it, and therefore any form of direct
> perception that posits features in experience that are not explicitly
> represented in the brain, has a profound epistemological hole in it.
I don't believe direct representationalism says this.
> The issue is ontological, because it identifies the vivid spatial
> structure
> of visual experience with either actual objects and surfaces in the
> world,
> or with internal representations in the brain. That is quite a profound
> difference.
I have a feeling that there might be some substitution in terms here.
Direct representationalism says that
- experience is in the head
- what it is experience of isn't
This is not the view you ascribe to direct representationalism.
> It is an issue which absolutely must be gotten straight if we are ever
> to
> make any real progress in understanding the brain.
Andrew has emphasized this point several times. Understanding the brain
requires understanding a) why the hell it does what it does? and b) how
the hell does it do it ? (and this is apart from the more basic
question of 'what does it do?'). These questions are different and
separable.
From your second reply to me:
Steven Lehar:
But by saying so, they assume that the
structure that they experiece is the world itself, so they are actually
saying that their experience is outside their head, although they deny
that
it is an experience, and thus avoid having to account for its existence
or
properties.
Me:
I think this point already appears above. How can we define the
'structure' here? Presumably it's objects, properties of objects,
spatial relations between objects, spatial positions of objects in the
egocentric frame, et cetera. But the inference from experiencing these
properties to experience being outside the head doesn't hold.
Steven Lehar:
...The internal spatial model gives us spatial consciousness, we become
aware of the surrounding world as a space full of objects...
Me:
However, you don't say that we become aware of our representations of
the surrounding world.
Marina
Steven Lehar
Rakova >
you say that perception detects features in a replica. But if
perception is still basically feature detection, then what is the point
of the construction stage?
< Rakova
See for yourself. Close your eyes and the world of visual experience, the
vivid spatial structure constructed by your visual brain, turns into a
murky brownish fog, because your brain has no input from the eyes. Isn't it
a lot more difficult to get around in the world?
Rakova >
What is the mainstream view on this in science circles?
< Rakova
I don't think there really is one just yet. For a long time the question
was thought to be a "pseudo problem" and was simply ignored, so the
literature is very sparse. But now with the development of 1: robotics and
2: virtual reality and 3: war games on the internet, the issue is just
beginning to bubble to the surface. But most people are naive realists by
instinct, some people terminally so, so it will take a while for the idea
to catch on.
Rakova >
How can we define the 'structure' [of experience] here? Presumably it's
objects, properties of objects, spatial relations between objects, spatial
positions of objects in the egocentric frame, et cetera. But the inference
from experiencing these properties to experience being outside the head
doesn't hold.
< Rakova
Structure is easy to define--just look at the experience of vision. You see
a three-dimensional volumetric space around you with a certain resolution.
It has certain bizarre nonlinear scale properties that we see as
perspective. And our awareness of objects takes the form of a vivid
experience of those objects as volumetric forms occupying a specific volume
of perceived space, that is a fully quantifiable model of the contents of
conscious experience. Whether or not one believes perception to be direct,
the spatial structure of experience is its most salient property.
How can *that* kind of experience arise without some kind of spatial
imaging mechanism in the brain? Experience has an information content, and
information cannot exist without a physical substrate to store or register
that information.
Lehar >>
...The internal spatial model gives us spatial consciousness, we become
aware of the surrounding world as a space full of objects...
<< Lehar
Rakova >
However, you don't say that we become aware of our representations of
the surrounding world.
< Rakova
No. Visual experience exists as a spatial structure, whether or not there
is a "viewer" there to "view" it. We know for a fact that experience
exists, and that it is spatially structured. We also know that experience
is the result of physical processes in the physical mechanism of the brain.
The reasonable conclusion is that physical processes can under certain
conditions become conscious. This does not address the question of *why* we
have consciousness, it just makes sense of experience as a physical
process, with no magic involved.