March 27, 2010

How Reductive Theories of Mental Representation Lead to Phenomenalism

It seems initially plausible to suppose that mental representation can be reduced to phenomenal character. That is, we all know that when we think about things we get into certain states of mind, and there is such a thing as what it's like to be in that state of mind. Now, when we think about things, we are representing the world as being in certain ways. It is tempting to suppose that this representing can be explained entirely in terms of the what-it's-like (phenomenal character).

According to naive forms of representative realism, this is because that phenomenal experience resembles the object. For instance, the experience of seeing a white paper has a certain characteristic - whiteness - which is like the whiteness of the paper. With the rise of modern philosophy and science, this view was rejected, and the primary/secondary qualities distinction was born. Empirical investigation showed that the whiteness of the paper was actually a disposition to reflect all visible wavelengths of light (approximately) equally. However, there is nothing in the subjective experience which reflects light. Thus color was classified as a 'secondary' quality. Secondary qualities are dispositions to cause ideas not resembling the quality itself. (Actually, exactly what secondary qualities are for, e.g., Locke is a difficult question, but his official definitions take this form - see EHU 2.8.10, 2.8.21.) Qualities like shape and size, however, are supposed to be real qualities in the things themselves which resemble our ideas.

This, however, won't work either - at least not as stated. For in what respect would my experience resemble the shape or size of the object? To resemble something is to share some property with it. However, no experience can share the shape or size of a physical object. For instance, my desk is about five feet long and rectangular. If the mind is physical it must be located in the brain. But nothing that resembled my desk in respect of being five feet long and rectangular could fit in my brain, which is much less than five feet long. If the mind is non-physical, then it is not extended at all, and so nothing in the mind has any shape or size. These sorts of considerations give rise to Berkeley's famous dictum "an idea can be like nothing but an idea."

The obvious response is to say that it's not the case that my idea of the desk is five feet long and rectangular - that's just goofy. Rather, my idea of the desk represents the desk as being five feet long and rectangular. But now how are we to succeed in our project of explaining representation in terms of phenomenal character? If there is nothing in the phenomenal character which is like an external object, how can it be in virtue of the phenomenal character of the experience that it is a representation of that object?

In the New Theory of Vision, Berkeley argued at length that visual ideas come to represent tangible objects by a learned association between the two: I learn that there are such and such connections between visible roundness and tangible roundness, and so whenever I see visible roundness I think of tangible roundness, automatically and by habit. In the Principles of Human Knowledge, Berkeley places tangible and other sensory ideas on the same level as visual ideas.

Suppose, then, that it is by our learning the connection between ideas and what they represent that our ideas come to represent things. This learning takes place by observing correlations. So infants start out with ideas that don't represent anything at all - all they have is the what-it's-like, the phenomenal character. Since their ideas are not representative, they can't think of or observe external objects, but only their own ideas. (It is only by means of ideas, we are supposing, that one can think of or observe anything.) So the connections they learn must be between one idea and another. It will follow that ideas can only represent other ideas. If this is so, then it is impossible to so much as think of, let alone perceive, mind-independent external objects. But we do in fact think of some sort of physical objects. So physical objects must not be mind-independent external objects but rather complexes of ideas. (If I am right in attributing this line of thought to Berkeley, it is not going to be easy to get the doctrine of notions off the ground, and if we can't get the doctrine of notions of the ground we won't be able to meaningfully affirm - let alone prove - the existence of God or other minds; I suspect that the solution to this problem is going to lie in the ideas of reflection, which I haven't discussed here.)

Now, there are basically two tactics for wriggling out of this argument: (1) denying the premise that mental representation is reducible to phenomenal character, or (2) developing an explanation of mental representation in terms of phenomenal character which does not depend on resemblance. (1) comes in two basic flavors: direct realism, and primtivism. According to direct realism, we perceive the objects themselves, not independent mental representations of them. This view has problems with misperceptions and it's also unclear what the best way of accounting for memory, imagination, etc. on this kind of view is. (But there is a vast literature on the subject, and I am not familiar with much of this literature; the solution may well be out there.) Primtivism takes representation as primitive. That is, that my idea of the desk represents the desk as five feet long and rectangular is a basic fact that cannot be given a more basic explanation. One thing to note about this view: if there is any sense to be made of the mind-body distinction (and there is), then mental representation, as the name suggests, falls on the mind side. So to take mental representation as primitive is to commit oneself to (at least) property dualism. Finally, (2) is exemplified by some forms of functionalism in the philosophy of mind: the mental state encodes certain data about the object and is used by the agent to interact with that object. It is in virtue of the total functioning of the system that the state counts as a representation of the object as being that way.

The argument is thus not inescapable, but it is certainly at least interesting.

Posted by Kenny at March 27, 2010 12:37 PM
Trackbacks
TrackBack URL for this entry: http://blog.kennypearce.net/admin/mt-tb.cgi/554

Comments

Is there more to be said on the "ideas can't resemble qualities" point than simply to point out the difference in ontological category?

Here's why I ask: Start with streetmaps. A streetmap represents, through resemblance, the layout of streets in a city. Some parts of the representational schema are Lagadonian (intersection on the map represents intersection in real life). Given a scale, certain spatial relations among representative elements on the map represent different spatial relations among entities in the world (if the representatives of x avenue and y boulevard are one inch apart on the map, this represents x ave. as being one mile from y blvd., etc.) Some elements are completely omitted (many maps offer no indication of elevation) or represented by using something entirely non-spatial to represent something spatial (many elevation maps use color variation to signify height variation).

I can see how one might choose to explain the first two things entirely via a Lagadonian representational scheme (on which the things represented, strictly speaking, are intersection (represented by intersections) and relative spatial location (represented by relative spatial location) etc., and this would be fine if it were not common to use completely something a-spatial (color) to represent something spatial (height/elevation). One might think that what color-coded elevation maps show is that a very very weak form of resemblance (including say, a purely structural isomorphism) is sufficient. And then it isn't clear that ideas can't exhibit such structural isomorphisms with qualities and objects.

Posted by: Lewis Powell at March 29, 2010 4:31 PM

So, it seems that maps and portraits, for instance, can represent in some way that is not merely conventional, because they resemble the object in respect of certain properties. For instance, the elements of the map resemble the objects represented in respect of (a) shape, and (b) relative size. (The resemblance will not be perfect, even in these respects.) It is also true that we can encode information about one type of thing with a completely different type of thing, for instance, by representing elevations by colors. But in these cases, it seems that we must establish a convention, and it seems that in order to establish a convention we must already have a way of thinking about the things in question, so this can't be where mental representation comes from.

Incidentally, Winkler (Berkeley: An Interpretation, 146) finds an argument for Berkeley's Likeness Principle in Philosophical Commentaries (AKA Commonplace Book) 378, which he thinks can be read between the lines of Principles 8 and Dialogues 206. According to this argument, in order for two things to be alike, they must be compared and found to be alike (to be is to be perceived). But in order for a mind to compare two things they must both be held before the mind at once. Only ideas can be held before the mind, so only ideas can be compared, so only ideas can be like anything. (Again, we've got to figure out how to get God and other minds out of this, but that's a question for another day.) It might be thought that this begs the question in favor of Berkeley's idealism, but some commentators believe that a similar theory of relations is to be found in Locke. Even if you think that's a bad interpretation of Locke, it seems like the fact that it's out there shows that one could come to have the mental act view of relations without already being an idealist.

Posted by: Kenny at March 29, 2010 4:57 PM

Reid has an interesting argument (not very thoroughly fleshed out) that natural language (in the sense of natural/god-given signifying relations) must precede artificial language (in the sense of English or other human languages). So you are in company with Reid on the thought that the conventions involved can't be the foundational story about representation.

I would be very surprised if someone could make a case (or make a case that is compelling to me, I should say) to the effect that Locke has that view of relations, and, I think such interpretations would only have the consequence you indicate if: a) the view they attribute isn't to be understood as (really) a form of nominalism about relations, or b) could be maintained by Locke, in conjunction with an indirect realist picture.

Posted by: Lewis Powell at March 29, 2010 5:51 PM

Well, the argument we're discussing is supposed to show that the view can't be held in conjunction with an indirect realist picture, so I'm in agreement there. I take it that the view is also supposed to be a form of nominalism. If it is a nominalist picture and Locke is inconsistent, the argument will still go through, won't it? Locke's indirect realism is based on resemblance relations between ideas and objects. If Locke can't claim that the sentence 'my idea of my desk resembles my desk' is true under his nominalist theory of relations (however exactly that theory works), then Berkeleian argument under discussion will refute Locke's version of indirect realism.

Posted by: Kenny at March 29, 2010 6:46 PM

Post a comment





Return to blog.kennypearce.net