Monday, February 7, 2011

Chemero Goes All Hegelian on TSM

I will call arguments like this Hegelian arguments.  Specifically, Hegelian arguments are arguments, based on little or no empirical evidence, to the conclusion that some scientific approach ... will fail. (Chemero, 2009, p. 7).
So, that's what a Hegelian argument is and here is what one looks like:
So, on the Turvey-Shaw-Mace view, either babies do not perceive their mothers (because the information for direct perception is unavailable) or they do not perceive them directly.  I take it that either alternative is unacceptable to radical embodied cognitive scientists.  (Chemero, 2009, pp. 112-3).
This looks to me like an argument based on little or no empirical evidence to the conclusion that some scientific approach will fail.  Now, one might say "But, Ken asking for empirical evidence that babies perceive their mothers is like asking for empirical evidence that dogs have tails."  (Cf., Chemero, 2009, p. 210, fn, 7)  But, maybe the thing to say is that one should be able to demand evidence for any empirical premise in an argument.

But, enough of ribbing Tony.  The more serious reply, I think, is that Chemero would seem to me to beg the question against TSM.  I thought their view is that one perceives affordances, rather than objects.  So, rather than seeing chairs, we see sit-on-ables.  So, by extension, it would seem that TSM would deny that one perceives Ken Aizawa; instead, one might perceive "argue-with-able" or "slap-in-the-faceable".

I happen to agree that we perceive objects and individuals, but I would want to give an empifical argument for that, given that TSM seem to disagree with me on that.


  1. Actually Tony's argument about the TSM approach is grounded in the idea that you can't have laws about individuals, and TSM insist on laws for information, so (Tony says) TSM can't have information about individuals.

    This worries him and motivates the situation semantics move, but actually I don't think it's a problem at all. Your face lawfully structures the light in unique ways, and I think an individual can be readily identified by their unique set of affordances. My hunch is that this concern goes away very easily once you allow yourself access to the sensible broader range of properties that do, in fact, differentiate you from someone else.

  2. I accept the idea that one can't have laws about individuals. I've even invoke that in a paper before. So, Chemero has this, but something else as well, that grounds his argument.

    What he doesn't get without "empirical evidence" is "Humans perceive individuals". And, it seems to me, you articulate the kind of line that TSM would take to resist Chemero's "unsupported" assumption that humans do perceive individuals: you don't perceive individuals, you perceive affordances, and these affordances "pin down" the individuals.

    So, it seems to me that we are both plumping for the same TSM reply to Chemero. (Only, I don't think the TSM reply works, since we almost never perceive affordances....)

  3. Well, empirically, I can indeed pick someone I know out of a lineup. But yes, my hunch is that 'perceiving an individual' is underpinned by other mechanisms.

  4. Yes, so here is Tony's situation. On the one hand, it does seem plausible that one can perceive individuals. On the other hand, he provides no "empirical evidence" for this (which is my jab about his Hegelian proclivities), but, more importantly, TSM don't think that humans perceive individuals by way of laws, but through other mechanisms that begin with the perception of affordances. But, if he's going to challenge TSM, he ought to bring some evidence.

  5. Face recognition is a good example of why it seems to me that the writers Ken is after have taken a good idea way too far.

    I agree that taking the general idea of affordances to an extreme, one might consider an entity as being defined by the intersection of all its affordances (and suggested that in a long-ago comment, except mistakenly saying "union"). But as I understand it, face recognition, for example, is executed at a very low level of brain structure - as improbable as it seems, maybe even at the single cell level. I suppose in principle recognition processing could amount to intersecting affordances, but that seems unlikely.

    Anecdote: I recently picked my wife up at the airport. We have a designated stretch of curb where she waits. As I approached, in that stretch was an object-like surface offering drive-up-to-able and comfortably-huggable (ie, about the right dimensions). The object was also recognizably clad, but I don't immediately see how to convert that into an affordance. As I approached the object, it started making moves that suggested pick-up-able. Etc, etc. But I really don't think I looked closely enough to recognize the object as my wife until she entered the car.

    Ie, affordances and complementary actions did almost all the work. So what if the final step involved some other process?

  6. But, if he's going to challenge TSM, he ought to bring some evidence.
    True. But you philosophers always complain when I tell you to go run the study, so... :)

    But as I understand it, face recognition, for example, is executed at a very low level of brain structure - as improbable as it seems, maybe even at the single cell level.
    There are two camps in the face processing literature. I know because I accidentally ran into this fight without knowing it was there with a simple study we'd run.

    The two camps are 'face expertise is a special case of perceptual learning' (something we're very very good at but underpinned by the same basic mechanisms as other perceptual learning), and 'faces are magic', where faces are an entirely different thing altogether. So at one level, the argument is still being had.

    It's a struggle to see an obvious affordance based route for this; but 'hard' != 'impossible' and part of it is that we're still busy with simpler tasks. Hell, Geoff's throwing affordances stuff still hasn't revealed a candidate information variable, and the property to be specified is just a size-weight relation. It's not for lack of trying, either; just so far, none of the good ideas have turned out to be true.

    So faces are complicated, but the real fight's still to happen there, I think.

  7. But, unlike Tony, I don't think you have to run a study to see whether dogs have tails.

    And I'll try to make clear in a day or so why your requests for me to run certain studies are requests for such a study!