Thursday, July 12, 2012

A Reply to Fodor and Phylshyn - Part 2

Continuing the reply that I think should have been made to Fodor and Pylyshyn's 1981 attack on Ecological Psychology. In F&P's article, the key elements of which are summarized here. They assert a very traditional, dualistic view of perception - as a process requiring sensory information to be supplemented by other cognitive processes in order to create an representational mental model of the world. They then point out (rightly) that some of Gibson's insights can be integrated into the traditional view and further assert (wrongly) that Gibson is thus offering nothing new. In so doing, I want to avoid as much as possible taking any bait offered by F&P which risk reeling us into to covert dualistic assumptions. I suggest that the best way to avoid such missteps is to stay firmly rooted in the line of thinking descended from pragmatism. Part 1 of my reply covered the meaning of "perception", "specification", and "direct perception", and the importance of remembering that if two things have all the same consequences, then they are the same thing (one crucial way of avoiding false distinctions). In this part I will continue to explain Gibson's approach by elucidating problems in F&P's critique.

On Percepts, or the Lack Thereof

I have pointed out some of the confusion created by discussions of percepts and concepts before. F&P repeatedly assert that the function of perception is to lead to percepts - which they seem to believe is a basic form of mental representations. However, Gibson rejects this altogether. Not just in his anti-representational stance, but as a part of the general distrust the pragmatic lineage has for a clear distinction between percepts and concepts. For Gibson, the underlying function of perception is not to create "percepts", but to guide action. The perceptual system and the action system form a loop that, when functioning properly, results in the organism being able to behave successfully - attempting actions that are possible and (typically) beneficial, and working around obstacles that would impede such actions. There is no straight causal path in perception, no line that begins with sensation and ends with a snap-shot mental representation, which we can pause and interrogate indefinitely.

F&P, and other Gibson critics think that they can acknowledge a "perception-action loop" while still retaining the rest of the establishment's ideas. They assert, generally, that while the perceptual process might well be continuous and ongoing, so too we can stop the process at any moment and demand an explanation for the subject's view of the world in that instant. But Gibson is not trying to explain frozen instants because.... they never occur! Rather than try to explain things that only happen in psychologists' and philosophers' imaginations, Gibson is trying to explain the real process by which we continuously interact with our world. The change in focus from imagined static entities to actual processes was a hallmark of the thinking that grew out of pragmatism.  James's emphasis on the "stream of consciousness" was revolutionary primarily in rejecting all notions of static mental entities. One well known consequence of this shift in focus was the battle between "structuralism" and "functionalism". F&P want to play a structuralist game, but it is not a game worth playing. If there is not an end-point of the perceptual process, then it makes no sense to focus our efforts on better understanding the imagined endpoint. It is a better use of our time to understand the process; a process that typically leads to accurate behavior, because the structured energy of the world is specific to relevant properties of the organism's physical surroundings. 

This does not create a magic solution to the problems of perception, but is creates a fundamentally different method of empirical investigation into perceptual processes: 1) investigate how the properties of objects structure ambient energy arrays, 2) investigate how the movements of organism can provide access to those patterns, 3) investigate the extent to which organisms do access those patterns, and 4) study developmental change in how organisms interact with those patterns. There is more to ecological psychology than those 4 points, but they lay the foundation. No talk of representations needed, no study of other mental processes needed, and the focus is on the basis of accurate perception, with perceptual errors easily accommodated.

Theories of perception often focus on ambiguous situations, but they typically fail to make a distinction that is crucial in ecological psychology - situations in which the organism cannot disambiguate a situation that is, in fact, one way or the other, and situations in which the organism correctly perceives ambiguity. The former is not a problem for ecological psychology, so long as it is clear that further exploratory movements on the part of the organism would, in fact, disambiguate things. The latter is not a problem for ecological psychologists, because accurate perception of ambiguity is still accurate perception.
Temporary Ambiguity
Some situations are not ambiguous, but the organism does not have access to the patterns that would disambiguate. This happens, for example, in the classic situation of the Ames room. The "illusion" occurs because the organism cannot make the movements that would immediately make the true situation clear. This is analogous to the experiences that organism might have in the early stages of exploring many fairly common situations. One sees and object and starts to reach for it, as they reach their hand adjusts to the size of the object - meaning that their hand started out open too much (or too little) but was correct by the time it arrived.

Critics of ecological psychology like to cease upon such temporary discrepancies. If, at that instant, the subject has a view of the world is incorrect, then, the critics insist, the argument for accurate (veridical) perception fails. But, this is to misunderstand Gibson's claim. Gibson claims that perception is typically correct, and that it can be corrected when it is wrong. Let us assume for the moment that it is possible not only to move towards accurate perception, but to actually get there, a vast majority of the time. (It is fair to make that assumption here, because F&P seem to agree.) Then the only question that remains is what processes are at play between the time of inaccuracy and the time of accuracy. F&P believe that the processes at play are inferences and other manipulations of mental representations. Gibson asserts, in stark contrast, that the processes occurring between those two times are describable purely in terms of behavioral interaction with the world - the organism is continuing exploratory movements which reveal and / or create new structure in the ambient energy, which is specific to the true situation.

Actual Ambiguity 
Another situation, which should not be confused with the first, involves the correct perception of an actually ambiguous situation.  I don't want to get too metaphysical here, by delving into discussions about whether the world is "probabilistic to the core", etc. Instead, I only wish to point out that there are 1) near threshold situations, 2) situations in which the organisms perceptual capacities cannot access the specifying information, and 3) situations in which there are "information" wars. The first type of situations is well illustrated by studying about people's abilities to fit through doorways. People are very good at making these judgements, but if you control the size of the opening carefully enough, you will reach the size that people successfully get through only half the time. If, as you approach threshold, people start saying they are not certain if they will fit through, that indicates accurate perception. The second type of situation is well illustrated by examples such as a person examining a gem with the naked eye, who might correctly determine that it is either a diamond or a cubic zirconian, with no ability to further differentiate. Surely it is simple to tell the difference under different circumstances, but the present situation is ambiguous. The third type is probably only present in social situations, in which one party intentionally hides information from another. For example, a poker player might provide place a bet unable to cover up certain signs that he is bluffing, while also intentionally providing signs that he has a good hand. An observer could pick up on some signs from each set, and thereby - i.e., in her response to those real occurrences -  see the situation as ambiguous.

What was the point again?... Oh, yeah... the fact that some situations are perceived as ambiguous, i.e., the participant in a study cannot give a confident answer or has conflicting behavioral tendencies, is not a problem for the ecological approach. No reference to non-perceptual mental processes is required to explain what is happening in these situations.

Belief and Knowledge
As F&P point out, Gibson and his followers often claim that their newfangled understanding of perception has implications for our broader understanding of "belief" and "knowledge". F&P claim that it does not. That is because F&P are still insistent that all Gibson is doing is adding a few tweaks to the first part of the establishment story about perception - one in which sensations get mentally processed and enhanced into percepts, which are further processed and enhanced into concepts, which are further processed and enhanced into beliefs and knowledge, which themselves are conceived as dualistically mental something-or-others. But that is a bad way of understanding the phenomenon referenced by the terms "belief" and "knowledge", and ecological psychologists can only support their position by re-entrenching in the pragmatist tradition. As mentioned in Part 1, in this tradition, we must try to ground our discussion in particulars. We cannot be sure where we stand when talking about belief or knowledge in general, so we need to delve into particular cases to get our bearings. How do you know, for example, when someone believes the bathroom is to the left? Answer: When they are searching for the bathroom, they turn left.

Belief is a propensity for action. In "The Will to Believe", James tells us that, in the extreme, "Belief" means a willingness to act irrevocably. Peirce could have similarly told us a few decades before, that "Knowledge" is just an acting 'as if' - to "know that" yellow and blue make green is nothing other than to mix yellow and blue when green is needed and none is readily available. Thus, if Gibson is providing an explanation of the process that leads organisms to act in certain ways in certain circumstances, then his work does have the types of implications that his supporters claim.

But What of Truth?!?

In the face of arguments from the pragmatist lineage, critics often get quite riled up about 'the truth', and F&P are no exceptions. The word "truth" comes into play in two related questions: 1) How can the organism know the true world, and 2) How can the organism know that its knowledge of the world is true. The first question we have already answered: We have already discussed how Gibson's system demonstrates that an organism can be responding to actual objects and events in the surrounding world. F&P's arguments otherwise fall flat. The second question is simply not an issue for Gibson - he makes no claims about it, so his system is neutral with regards to it. There is no expectation that the organism has self-awareness of what it does during the perceptual process, and certainly know expectation that the organism could use that self-awareness as an element in a larger, linguistic, justification system. William James also draws out the distinction between these questions in "The Will to Believe":
the faith that truth exists, and that our minds can find it, may be held in two ways. We may talk of the empiricist way and of the absolutist way of believing in truth. The absolutists in this matter say that we not only can attain to knowing truth, but we can know when we have attained to knowing it; while the empiricists think that although we may attain it, we cannot infallibly know when. To know is one thing, and to know for certain that we know is another. One may hold to the first being possible without the second; hence the empiricists and the absolutists, although neither of them is a sceptic in the usual philosophic sense of the term, show very different degrees of dogmatism in their lives.
Gibson and F&P are on opposite sides of this divide. Using James's terminology: F&P want, as absolutists, for proof that we can be sure when we are correct. Gibson only wants to prove that we can be correct. For Gibson, as for Peirce, the only way to be sure you are correct is to follow through on all possible consequences of your belief, and have none of them fail to achieve. It is trivial to point out that a finite organism rarely, if ever, engages any object or event of the world long enough to ensure that it is correct in all possible ways. It is sufficient for us, as third parties, to affirm the correctness and attempt to explain it. Such an explanation is offered by Gibson, and F&P can go jump off a cliff.


P.S. Conference travel is upon me, and then, probably, surgery for my wife, right before classes start anew. Expect a brief blackout in new posts. Subscribe to email notices or RSS feed if you want to be updated. I will be at Cheiron in Montreal, Quebec, chairing a session on the history of behaviorism, then I will be at APA in Orlando, Florida, talking on "William James, Edwin Bissell Holt, and the Once and Future Behaviorism." If anyone wants to catch me at one of those events, drop me a line.

No comments:

Post a Comment