Sunday, June 24, 2012

A Reply to Fodor and Phylshyn - Part 1


In a prior post, I hummed a few bars of “Ecological Psychology needs to be evaluated within the context of AmericanPhilosophy.” I then started wading into one of the pivotal debates in the history of Ecological Psychology, the 1981 debate that pitted Fodor and Pylyshyn against Turvey, Shaw, and Mace. F&P’s criticism was published in Cognition, shortly after Gibson’s death, and TSM’s reply established the new direction for the field. In the last post, I summarized F&P’s arguments, and interspersed brief notes about when they did, or did not, seem to be giving Gibson a fair shake. In this post, I want to try to avoid nit-picky details about where F&P went wrong. Instead, I want to outline a broader reply to F&P’s criticism. 

The overall problem, it seems to me, is that Gibson is playing an American Philosophy game, working within the intellectual lineage of Peirce, James, etc., while F&P want to play a Continental Philosophy game. I don’t want to go into too much details about the historic differences between the two approaches, or how they arose. My more meager goal is to defend ecological psychology in a way that stays true to its roots. American philosophy, in general, is concerned with earthly particulars, is suspicious about intellectual distinctions, and does not privilege a first-person point of view. While F&P want solutions to the traditional problems of perception to happen on an intellectual level, Gibson proposes that the supposed problems are typically solved in the grit of everyday interactions.


Perception
Let us begin by defining ‘perception’. The modern, western, philosophical tradition has it that questions about perception revolve around situations where someone can tell you that they perceive something. They privilege the first-person point of view, and if someone asserts they perceive something, their claim must generally be taken at face value. But outside of strange philosophical discussions, that is not what the term means. The most natural way to use the term is in the third person, to refer to certain types of matchings between other people’s behavior and their world. “Why did he jump?” you ask. “Because he saw a snake,” I reply. In so doing, I point to a relationship between the jump and an illuminated reptile. Why do I think this is the most natural way to use the term? Because when there is a dispute between the first person and third person, the third person typically wins.

“I saw a ghost,” you say.

“No, you saw the curtain swaying in the breeze,” I say.

“Oh,” you reply.

At that point, we might agree that you “imagined” a ghost, “thought” it was a ghost, or even “experienced” a ghost, but one glance back in the room and you will agree with me that you did not “see” a ghost. While we might accept as a perceptual question “Why did you react to the curtain that way?”, we clearly do not want to treat in the same manner the question “Why did you see a ghost?”. The latter is premised on a false assumption, because, as we both agreed, you did not see a ghost.

Similarly, if I do an experiment in which I place different plants in the hallway (spiky on one side, soft-leafed on the other), and you walk a foot further from the spiky plant no matter which side I put it on, but only in that part of the hallway, then I have evidence that you saw the plant. (Dick Coss has done manipulations like this using joggers in a arboretum.) If I stop you in the hall, and you claim never to have seen the plants, my third-person observations trump your first person perspective. If I go back and show you the video tapes, where the plants are switched back and forth, and your walking is altered, you will agree that you must have seen the plants, even if you were not aware of having seen them.

Questions about vision are questions about how behavior is altered by the presence of illuminated objects and events in a person’s surroundings. 

Specification
Ostensibly, much of F&P’s criticism regards Gibson’s notion of specification. Gibson’s claim is 1) that certain patterns of energy are specific to certain states of the world, 2) that for organisms with access to those patterns, the state of the world would be unambiguous, and 3) that organisms typically have access to such patterns. Below we will work through a case in which an animal is “wrong” about what it sees, but for now let us remember that organisms are typically not wrong. They are not wrong, Gibson asserts, because they have continuous access to patterns that disambiguate the world. F&P repeatedly assert that animals do not have access to such patterns. They go so far as to claim that Gibson’s system seems to require organisms to access “the distribution of the light across the entire inhabited universe” (p. 170). In this suggestion, they willful neglect two of Gibson’s important ideas A) higher-order invariants, B) redundancy in specificity. If Gibson is correct that there are higher-order invariants, constants in the changing flux of energy patterns, that are specific to the properties of object, then the animal need only have access to the invariant. Further, if Gibson is correct that there is tremendous redundancy in the energy, such that the same invariants can be found in several types of energy, and can be found in many arbitrary segments of that energy, then the animal’s task is easier still. 

I don’t want to go into detailed response here, except to point out that F&P dismiss these possibilities out of hand. That is crazy and short-sighted. At the time, there were only a handful of examples available, but still it was an open question if many invariants could be identified, if they could be shown to be highly redundant, and if it could be shown that organisms take advantage of them. For example, the feedback provided by a wielded rod is specific to its length, one need only move the rod a small amount to have access to the needed feedback, and people are quite good at moving the rod in the correct ways when they are placed in a length-determination task. 

Specification is important to this discussion because, if the world before you is unambiguous, then it is unclear why you would need to “infer” about it: Today I have walked through several doorways. In each case, I had access to patterns specific to the properties of the doorway that allow my successful passage. That is, my movements permitted access to patterns of energy that made the size, shape, etc. of the doorway unambiguous. Or at least that is Gibson’s assertion, and F&P never face the assertion directly. Thus, they avoid the most forceful of Ecological Psychology’s assertions: If this entire scenario is in place, to say that I “inferred” that I could walk through the doorway is patently misleading. I perceived that I could walk through the doorway. One might even say that I knew I could walk through the doorway. 

Direct Perception
Arguments for “direct” perception are arguments that certain types of intermediaries are not involved in the perceptual process. While many people have challenged Gibson with petty irrelevancies, F&P are spot on to focus their challenge on “inference.” Gibson does not deny that inference happens, but he is adamant that inference is not required as a component of the perceptual process: Do people sometimes infer Y based on perceiving X? Yes. Do we need inference even to perceive X? No. That is, there is no psychological process properly labeled “inference” which happens in the middle of the psychological process of “perception.”

F&P have many arguments for why inference must happen in the course of perception. They are clearly convinced it must be so, and will not be dissuaded. Rather than address their concerns individually, we might ask why they are so certain of their position. The best explanation for their conviction is that they have fallen prey to the Psychologists’ Fallacy. As a reminder, the Psychologists’ Fallacy happens when you mistake the end of an intellectual inquiry for the beginning.

To be as generous as possible to F&P, let us focus here on a perceptual example in which an organism behaves "incorrectly":

To take a classic example from ethology, a hen with her brood will panic at the sight of a cross flying overhead, short-end first. The hen panics exactly as she would if a hawk flew overhead. There are a lot of interesting implications of the details of that research (see Heft, 2011), but they are not terribly germane here. The only facts we need are 1) The hen is acting as if a predator was present, and 2) there is not a predator present.* In the face of that evidence, F&P would certainly insist that the hen has inferred a predator is present. They would insist that some dualistically mental process acts as an intermediary between whatever it is that the hen can see and the hen’s behavior, because, they would assert, the perception-of-the-moving-shape cannot possibly explain the hen’s panic on its own. But this is the Psychologists’ Fallacy! They, F&P, know that the moving cross is not a predator, and therefore they assume that the panicking hen has made an inference. But that is hogwash. Having studied and thought about the situation before them, F&P have inferred about the relationship between the stick and the panic. Then, mistaking what they do for what the chicken does, they assert that the hen has similarly inferred.** 

The hen does not have an internal state, from which the hen makes an inference, and from the result of that inference conclude that it needs to panic. The hen is panicking (doing 'panic'), and this was caused by (among other things) the illuminated object moving overhead; no other mental processes required. That is what it means for perception to be direct, and F&P’s failure to think clearly about the problem does not dissuade us.

When Many Things are One Thing
In footnote 11, F&P make clear that, in their opinion, it is one thing to respond selectively to equilateral triangles, and another thing to respond selectively to equiangular triangles. From a pragmatic perspective, I am not sure there is a difference, as F&P readily admit that there is not a situation in which the two response-tendencies would lead to different outcomes. F&P's similarly attempt to assert that perceiving the morning star is clearly distinct from seeing the evening star, while admitting both to be the planet Venus. I suspect that a surprising amount hinges upon this point. Here we might well apply one of Pierce’s maxims for explaining pragmatism: If two things have all the same consequences, then they are the same thing.

We must hold our ground by challenging false divisions of the world, especially those on arbitrary linguistic grounds. For example, one person could say “I see it is morning” and another say “I see that the sun is now rising in the east.” Because different words were used, we are sorely tempted to say that different things were seen by each person, and F&P are convinced that a theory of perception must here explain two distinct happenings. But are the perceptions as distinct as the language makes it seem? No, there is only one thing to explain here: When I see the sun rising in the east, I see that it is morning, because morning is nothing other than the time during which the sun is rising in the east. I might or might not be able to tell you that it is “morning”, depending on my past and current experiences, but to see one is to see the other; there is no additional process that goes into seeing that it is morning. 

Where does this logic take us if, as Gibson claimed, the properties of objects or events and the patterns found in ambient energy are so related that one cannot be found without the other. If property X and energy pattern Y are so linked, then to respond to pattern Y is to respond to property X and vice versa. An organism that does one will always do the other. It does no good to hand wave about possible worlds, nor to give intellectual arguments about how the two are distinct. If “the organism responded to property X” and “the organism responded to pattern Y” yield all the same consequences, then they are the same thing.

And here it is absolutely crucial, as stated in my last point, that one cannot construct the energy pattern specific to some property of the world, except by constructing that property. One can go about all day creating, in a variety of manners, patterns probabilistically related to properties, but if the pattern is specific to the property, then you are stuck. If there is a pattern of energy specific to a hole in the wall that I can fit through, then creating that pattern requires making a hole that I can fit through. 

Initial Wrap Up
This post is getting long, and there is still more to do. Hopefully, however, I have started to give a flavor for how I think Ecological Psychology should have been defended against F&P’s attack.  

Saved for the next post: “On Percepts and Concepts”, “Ambiguity”, and “Belief and Knowledge”.


---------------

*We will skip the cheap shot of asking how we can know that the hen is wrong, as F&P’s principles would seem to require that our perception of the world be just as suspect as the hen’s. 

**At this point the reader might wonder if my own logic here is at all consistent with my logic from the prior section. That is, why do I not privilege F&P’s third person perspective? A very fair challenge. To avoid larger discussion, I will merely point out that the descriptive use of the term “inference”, which is what the third-person perspective refers to, bears little resemblance the internal, mental, dualistic, explanatory process F&P are trying to assert by using the term. Thus, if we were to apply the same logic as in the first section, we would admit that the organism “infers”, in some sense of the word, but absolutely none of F&P’s subsequent arguments would follow from that.

3 comments:

  1. A couple of wonderings outloud at having skimmed this while writing an exam for my Lifespan class....

    1) "The modern, western, philosophical tradition has it that questions about perception revolve around situations where someone can tell you that they perceive something"--I had always thought that the Western tradition treated questions about perception revolving around the private experience of, as JJG liked to say, having a percept.

    2) "Questions about vision are questions about how behavior is altered by the presence of illuminated objects and events in a person’s surroundings" seems dangerously S-R Psychology. EC, I grok the point you're trying to make, but this wording concerns me. Gibson/EP is most surely about taking up the charge that psychology studies behavior, but Gibson/EP ardently rejects the study of behavior as an S-R enterprise.

    3) I always thought that the foundation of a charge against F & P has to begin with challenges to the causal theory of perception. Recognizing that was my "a-ha" moment for getting Eco Psych.

    -Jim

    ReplyDelete
  2. Jim,
    Thanks for the comment.

    1) The "having a percept" issue is important. Hopefully I will get to that when I next have a moment to breath, and can do the next post. Here I was trying to point out the hyper-intellectualization of the Cartesian system. In that system, if I tell you that I did or did not see something, I always win.... because I know myself perfectly and you don't. F&P seem to want an explanation for the internal world I seem to describe, in poem, when inspired by a sunset... rather than an explanation for how I walk through the world when stumbling for my morning tea.

    2) I think is EC is very close to radical behaviorism, though there are many forms of behaviorism, with S-R as one of the lower forms. Surely there is a world we are acting within, and we have little obvious need to use the brain as a way-station in our explanations. Interestingly, TSM, as I understand it, presents an S-R behaviorism (how else can we possibly understand the 'disposition' talk?!?).

    3) The ...tion-preception-action-perception-action-perc... loop is really crucial. You are right that there probably should have been a place to emphasize that more.

    ReplyDelete