Saturday, February 5, 2011

Chemero and the Exploding Boxes

In comments, Andrew has asked whether Chemero's, 2009, approach to information helps with the exploding box type cases.  I have not read the whole of the relevant chapters yet (Chapters 6 and 7, I think), but I think the answer is now.  Moreover, the problem rears its head right about here:
Imagine that there is a beer can on a table in a room that is brightly lit from an overhead source. Light from the source will reflect off the beer can (some directly from the overhead source, some that has already been reflected off other surfaces in the room). At any point in the room to which there is an uninterrupted path from the beer can, there will be light that has reflected off the beer can. Because of the natural laws governing the reflection of light off surfaces of particular textures, colors, and chemical makeup, the light at any such point will be structured in a very particular way by its having reflected off the beer can. In situation s1, the light at point p has structure a of type A. Given the laws just mentioned, there is a constraint connecting the situations with light-structure type A to the beer-can-present situations of type B. So the light structure at point p contains information about token beer-can-presence b (of type B). Notice too that, because of conventional constraints governing the relationship between cans and their contents, beer-can-presence b being of type B carries information about beer-presence c of type C. Furthermore, the light at some point in the room from which the beer can is visible will contain information about the beer can's affordances. Take some point p, which is at my eye height. The light structure available at this point will contain not just information about the beer can and the beer, but also about the distance the point is from the ground, the relationship between that distance and the distance the beer can is from the ground, and hence the reach ability of the beer can and drinkability of the beer for a person with eyes at that height. (Chemero, 2009, pp. 118-9)
Ok.  So, the stuff up until "Notice too that ..." is essentially the Gibsonian line (of Gibson, 1979, p.127) about surfaces structuring light.  That sound right.  The trick, then, is to somehow connect the structuring of light by surfaces to affordances.  Gibson, 1979, p. 127, tried to make the connection by way of the hypothesis that the composition and layout of surfaces constitutes affordances.  Chemero has a different idea: There are constraints between surfaces and their innards.  In the case of beer cans, society puts beer in beer cans.  That's the constraint.  From there, Chemero glides willy-nilly (i.e. without telling us how) to having constraints go all the way to drinkability.

But, this is not going to work.  Take two beer cans that are physically identical on the outside.  Have one with beer and the other with something non-potable, say, bleach.  Subjects will, presumably, have the same visual perceptions of the two cans, but the two cans afford different things, so visual perception is not of affordances.  I take it as evident how to run this with the exploding boxes, piles of white stuff, panes of glass, etc.

This goes back to the cartoonish formulation of the problems.  Visual perception is typically based on outers; affordances are typically based on innards.

Chemero could rightly point out that in this imagined scenario, we do not have the requisite conventions regarding the contents of beer cans.  Correct, but we still have perception, right?  Which suggests that social conventions regarding the physical configurations of things do not have that much to do with perception.

Maybe there is some trick Chemero has up his sleeve later, but we shall see ...

13 comments:

  1. So, there is indeed another trick. There is that idea that even though there is no beer in the poison-filled can, there is still beer information. So, in all these cases, Chemero can say that the two objects produce identical visual perceptions because there is identical information (not derived from the objects in current visual field).

    Now, it's not clear to me right away, what's the best way to deal with this. But, there is this. On Chemero's view, one does not necessarily perceive what is actually afforded. That seems to be what Chemero is willing to give up with the view that information is not necessarily veridical. Since the poison-filled can does not actually afford potability, one must not be perceiving potability. Maybe Chemero thinks one perceives Charles' dreaded "apparent affordances". But, then he needs to explicate what an apparent affordance is and (to satisfy other goals) do this without appeal to mental representations.

    ReplyDelete
  2. It's not clear to me that it's a good idea, either. But your description doesn't quite seem to capture what (I think) Tony's up to.

    The issue is whether you can have information without a law. TSRM say the presence of the beer can is not informative about the presence of beer, because the relationship 'beer can -> beer' is mere convention and is breakable. Barwise & Perry, on the other hand, say that the beer can can be informative about the presence of beer, if you also know the relevant constraint (beer cans typically have beer in them) because constraints just have to be sufficiently reliable.

    Tony wants to use situation semantics (so it's not willy nilly: I don't know much about this topic but I do know everyone at IU had nothing but respect for Barwise) to underwrite a notion of information that includes reliable conventions so as to dodge any need for, say, inference. It struck me that the motivation (wanting an account for how we come to know about the presence of beer given beer cans) is related to your worry about surfaces, etc.

    Just to be clear: I don't think this is the answer either. But the parallels to the current argument seem obvious, so I'm interested in your analysis.

    ReplyDelete
  3. Agreed that Tony is stalking a different horse here than what I'm getting at with the surfaces/affordances gap. He's not trying to solve this "gap" problem.

    Tony says that conventions get from beer-can to beer, but he says nothing about how to get from beer to affordances of beer (such as potability). It's this last step that is willy-nilly. Of course, maybe one could dig more out of Barwise and Perry ...

    Yeah, Barwise is great. I'm using the Barwise & Etchmendy logic text for my intermediate logic class (for which I am preparing now).

    One thing that Chemero is not explicit about, however, is whether it is merely the existence of the beer-can/beer convention or believing that there is a beer-can/beer convention that matters for making the link all the way to the beer. It sounds like the former. But, a challenge to Chemero's view will probably have to be worked out for both cases. (Sucks when an author is ambiguous, since then the challenge has to work by proof by cases.)

    But, Chemero's view is hard to get a handle on. In the Gibson and TSRM cases, one could spell out what the affordances are and what the light is like and then show how they come apart. For Chemero, it seems this won't work, because one does not necessarily perceive affordances. In the poisoned beer can case, Tony is (probably) going to say that one still perceives "potability", even if there is no actual potability. There is only what Charles might call "apparent potability".

    Notice that Tony's approach here seems to conflict with another Gibsonian idea: "Perceiving is an achievement of the individual, not an appearance in the theater of his consciousness. It is a keeping-in-touch with the world, an experiencing of things rather than a having of experiences." (Gibson, 1979, p. 239). In Tony's scheme, it does not sound like perception is "a keeping-in-touch with the world, an experiencing of things". It is instead a having of experience.

    ReplyDelete
  4. And, what the heck, Perry is an outstanding philosopher as well. So, Tony is drawing on good stuff, but that does not necessarily mean it's going to work.

    ReplyDelete
  5. "Perceiving is an achievement of the individual, not an appearance in the theater of his consciousness. It is a keeping-in-touch with the world, an experiencing of things rather than a having of experiences."

    This is a fascinating statement. I have been asking essentially the following question everywhere I encounter discussions of this sort of thing (translated here into Gibsonian):

    If perception allows us to "keep-in-touch with the world", in particular to get around in the world by executing the actions determined by extraction of the information in light from the environment, what is the advantage of "appearances in the theater of consciousness", AKA (I assume) "phenomenal experience" or "qualia"?

    Even from a more conventional perspective on visual processing it appears that we can do quite a lot without the need for phenomenal experience. It seems like it must be useful, but I haven't been able to identify a specific activity where it is obviously crucial.

    Does Gibson address that issue?

    ReplyDelete
  6. One thing that Chemero is not explicit about, however, is whether it is merely the existence of the beer-can/beer convention or believing that there is a beer-can/beer convention that matters for making the link all the way to the beer. It sounds like the former.
    Actually one thing I'm worried about here is that I think, in situation semantics, you need to have access to (however you want to cash that out) both the informative thing and the constraint. This seems like a problem for a direct perception story; in TSRM, laws do their work behind the scenes, structuring the informative thing so that it simply can be informative. You then only need access to the information, and not to the law.

    I'm not quite sure what you mean by it conflicting with that Gibson quote. How is Tony's story 'having an experience' vs 'an experiencing of things?'

    ReplyDelete
  7. @Andrew,
    So, let's say that one does need both the constraint and "access" to the constraint. It is, I think, different than the TSRM account, which just has the laws working behind the scenes, as you say. (That's why the problems for TSRM are not essentially psychological, which is a point I will return to.)

    That does seems problematic to me. Maybe for not exactly the reason you think it is. I am not certain exactly what Gibson, et al., mean by direct perception, but I think it includes more than that it not be mediated by representations (which is all that Tony apparently wants).

    I would think that "access" to the constraint might be problematic, since it would seem to be a kind of use of information or use of memory, which Gibson abhors. But, Tony doesn't say anything about this kind of issue, at least in the sections that I have read.

    Re the Gibson passage, I take it that an experience of a thing involves that thing. Perceiving X involves the presences of an X. But, for Tony, perceiving beer does not require the presence of beer. One might perceive beer, in Tony's sense, even if there is just an empty can. That, however, sounds like having an experience of beer, rather than experiencing beer. Gibson is not all that clear here, but that seems to me to be what is going on.

    ReplyDelete
  8. @Charles,

    I don't really know much about the topic you are asking about, but have you looked at the "dual systems" hypothesis literature, the stuff on dorsal and ventral visual streams?

    ReplyDelete
  9. The way I interpreted "'having an experience' vs 'an experiencing of things" - which motivated my question above - is that the former is the phenomenal experience, ie, the mental visual image we "see", "have before the mind", etc. Unfortunately, I'm not aware of any standard vocabulary for it other than (I think) "qualia".

    The latter is what I take Gibson's perceive-process-react feedback system to be doing. It merely extracts an information stream from the light, runs it through some kind of processing analogous to an electronic signal processor for comm signals, which causes some action(s). In that model, there appears to be no role for the mental visual image. Ie, there is no "Cartesian theater" in which the perceived environment is "shown like a movie" (to whom?).

    Then my question is "why do we have the experience of the mental visual image if the perceptual system can function just fine without it?"

    My totally unsubstantiated hunch is that the image is - in some totally undefined sense - an illusion. In particular, the only use unique to it that I've identified is that it allows us to describe a representation of our visual input to others, which it appears we could not do without it. Precise, eh?

    I've read some about the dual processing streams, but I'll poke around and see if the issue comes up that literature.

    Any chance I have a comment stuck in a filter? I composed - and thought I had submitted - an earlier comment in this thread that hasn't appeared. I have a copy, so if not I can resubmit it (if I decide it might still be relevant).

    ReplyDelete
  10. I agree with your sense (in your first two paragraphs above) of what Gibson is rejecting.

    But, I don't know answers have been given for why we have mental images.

    I don't have any comments in the spam filter, so I'm not sure what might have happened. I've never deleted any comments, though there are some about which I have been sceptical.

    ReplyDelete
  11. Mental images are an odd topic; people vary greatly in whether they experience visual images, from feeling very detailed to having none at all. Sabrina posted about some mental rotation studies on a man with no mental imagery here, and a blog post come through my Twitter feed the other day about where people think their imagery is.

    Then my question is "why do we have the experience of the mental visual image if the perceptual system can function just fine without it?"
    It's been considered to either reflect or actually be the basis for mental operations (eg mental rotation, a la Roger Shepard) but it's a really mixed bag. I'm with Gibson; I tend to think of imagery as an epiphenomenon, a side effect of the neural implementation of our visual systems, rather than a causal link in the chain.

    The two-visual-systems stuff (Goodale & Milner) is usually studied looking at dissociations between visual experience and visually guided action and getting overly excited by it. I'm not a fan, for various reasons, not least of which is the fact there are about 10 different 'two-visual system' theories and the dorsal/ventral systems get to do all kinds of things.

    ReplyDelete
  12. I just read the Conclusion section of Gibson's book (Amazon's "Look Inside" feature) and now have a little better idea of his views. In particular, it seems clear that Gibson distinguishes between merely extracting enough information about the environment from received light to support execution of actions and additionally having a mental visual image of the environment. I assume that the former is "experiencing the environment", the latter "having an experience of the environment". When I first read Sabrina's summary of Chapter 5, I couldn't understand why Gibson made such a fuss about the idea of an "image on the retina", but I now assume that he was trying to dissuade readers from thinking in terms of any kind of image being involved in his concept. (Since I never think of the retina's function in terms of images, I found the fuss merely confusing.)

    I have been assuming that Chimero's "conventional constraints governing the relationship between cans and their contents" had something to do with extracting information from the mental visual image (an example of what I assume you guys mean by a "representation"), eg, reading the label (how else would a beer can be distinguishable from a soda can, soup can, etc?). In which case the process would involve "having the experience of" beer-can-present. Now I suspect that's wrong. But if there's no mental visual image, how is beer-can-present perceived?

    Well, there is that "something-blindness" (like change- or attention-, but neither of those) where a patient claims not to see any difference between two pictures of a house despite one having flames shooting out of its windows, but when asked which is preferred, reliably chooses the one not on fire. That suggests - yet again - that the mental image may not be necessary for making certain decisions, even those that we would normally describe as "recognizing" or "inferring from".

    So, maybe Chimero is assuming that beer-can-present can be unconsciously "recognized" via information extracted from the light and beer-present can be unconsciously "inferred" from the constraint but that no mental visual image is involved. In which case his scheme is an "experiencing".

    (I don't see why calling "beer-can-present -> beer-present" a law avoids inference, but if one insists that it does, presumably "inferred" can be replaced by some other word.)

    In any event, I find it confusing that these later Gibsonites insist on trying to view troublesome low-level features of the environment - beer-present, potable, perhaps even drinkable - as being affordances. I infer from Gibson's Conclusion section that his affordances were much higher-level features. What motivates extending the concept beyond what seem to be its natural bounds?

    ReplyDelete
  13. A couple of things, Charles.

    I couldn't understand why Gibson made such a fuss about the idea of an "image on the retina", but I now assume that he was trying to dissuade readers from thinking in terms of any kind of image being involved in his concept.
    This was important at the time of writing (and still some today). The idea that the retina forms an image and that that image is the basis for visual perception is a fairly standard view. It motivated most of the original representational accounts of vision, because images are rubbish.

    Chemero's not trying to extract information from a mental image. Quite the contrary - he wants direct perception (via a suitable informational coupling) of things like 'beer-presence', by virtue of the constraint 'beer cans contain beer' and the situation 'perception of beer can'. It is potentially direct (non representational) because the constraint allows the presence of the beer can to be informative about the presence of beer because of the constraint - you don't need to infer anything (the point of situation semantics was precisely to achieve this, as I understand what Tony says of it).

    In any event, I find it confusing that these later Gibsonites insist on trying to view troublesome low-level features of the environment - beer-present, potable, perhaps even drinkable - as being affordances. I infer from Gibson's Conclusion section that his affordances were much higher-level features. What motivates extending the concept beyond what seem to be its natural bounds?
    Good question. I agree this is all kinds of trouble. My guess is that part of the motivation is to try and address one critique of Gibson, namely that he can't handle things we can clearly do (like 'know' about beer in beer cans). Tony's trying to find an informational basis for this sort of thing to preserve the directness, and he doesn't think laws gets you there. I actually think he's wrong, or at least, that's my bet - I think the laws concept has plenty of life left in it and I think the actual problem is that these tasks have been mis-described.

    I've got a post on all this exploding box stuff I want to try and get out today if I can finish my lecture, and I'm about to start posting on Tony's book in detail, so these things will all come up soon.

    ReplyDelete