Thursday, January 13, 2011

Noë on Gibson on Affordances

Here is a kind of interesting passage from Noe's paper:
On a familiar reading of Gibson (1979), he is said to have held something like this. For example, Campbell (2002) characterizes Gibson as follows:
... suppose, for example, that an unfamiliar piece of apparatus appears on a workbench. I have no idea what this thing is for. I don't know if I can touch it - maybe I will be electrocuted, or the thing will blind me, if I do that. Or maybe it is simply the latest kind of television, or a paper weight. So I don't see it as affording anything in particular. In that case, by Gibson's theory, the thing should be simply invisible; I should be able to see it only when I am told what it is for. But that is not a persuasive conclusion; it seems perfectly obvious that we can see things without knowing what they can be used for.
(Campbell, 2002, p.143)
As I read Gibson he is not committed to such a conclusion, for it is not Gibson's view that we only see affordances, or that we can only see objects in so far as we can see their affordances. Gibson's point comes earlier: that we can see affordances. The significance of his thought is this: for Gibson, perceptual consciousness is not confined to so-called categorical properties of things, such as shape say, or size, or qualities like colour. Gibson is advancing 'the radical hypothesis' that 'the "values", and "meanings" of things in the environment can be perceived directly' (Gibson, 1979, p.127).  (Noë, 2010, p. 247).
Now, I had always thought that Gibson's view is that one always perceives affordances.  I'm not entirely sure why, but I thought that perception was always supposed to work something like this: Affordances structure light (for example) and humans pick up the information in the light.  Simple.

I wonder how many people read Gibson as Noë does.  Comments welcome.  Comments with references to Gibsonian pronouncements even more welcome.

But, suppose that Noë is right.  Is there some place where Gibson spells out when one perceives (or can perceive) affordances and when not.  In other words, when do we get affordance perception and when do we not?

91 comments:

  1. First of all, the object is going to have affordances. Just because you don't know if it's a TV, it will still have dispositions for grasping, lifting, sitting, etc etc. So Campbell really doesn't seem to understand what affordances are.

    Gibson also predicted that events are perceivable; there was a good argument about that in Ecological Psych vol 12(1) from a target article by Stoffregan.

    There's also the issue that 'to see' and 'to perceive' have technical meanings for Gibson, and Noë seems to be being a little careless there. You can detect all kinds of things, but not everything is information for affordances and thus isn't perceived (by Gibson's definition).

    I'd need to see the paper though, and I don't have the book. But there's numerous problems with Campbell's approach and Noë's point seems to (correctly) be that Gibson was adding affordances to the list of things available (then relying on them alone for the control of action).

    ReplyDelete
  2. Well, I don't see that Campbell is denying that things have affordances. Only that we "get at" them, whether by direct or indirect perception.

    ReplyDelete
  3. He seems to be saying that until you know what the object is for, you can't get to that affordance. That's backwards; you come to know what it's for by getting to the affordance.

    As I said, it's clear he doesn't know what affordances actually are.

    ReplyDelete
  4. Oh, I'm betting he's taking an affordances to be what an object is for.

    But, that's kind of an irrelevant wrinkle. What he is getting at is that you can (in the non-technical Gibsonian sense) perceive a thing without perceiving or knowing what it affords.

    ReplyDelete
  5. "I don't know if I can touch it - maybe I will be electrocuted, or the thing will blind me"

    This example seems wrong. I would describe those as possible consequences of acting on the object's affordances. To trivialize somewhat, presumably no one doubts that a horizontal disc supported by three verticals columns (AKA, a "stool") affords the opportunity to sit on it. The fact that it's a practical joke and the "legs" are not secure is a separate matter.

    Also, to the limited extent I understand Gibson and affordances, it strikes me that the consequence of Gibson's approach is that an object becomes defined not by its properties as we usually think of them but by the union of all its affordances. Yes?

    ReplyDelete
  6. There are many things you can do in a non Gibsonian sense that don't involve affordances. That's why they're non-Gibsonian :v

    But given that he's not correct about what affordances are it's not that interesting, surely.

    ReplyDelete
  7. But, it seems to me that neither of you are trying to see what Campbell is getting at. You want to nitpick with the formulation.

    I think that the stool example is a problem for EP.

    Take some physical object. Is it a grabbable thing (to use a TSRM example of an affordance). You can't tell just by looking at it and moving around it. It could be such that were you to touch it it would electrocute you or blow up or distingrate. That's what he seems to be getting at when he writes, "So I don't see it as affording anything in particular."

    ReplyDelete
  8. Hello Ken

    Some (rather uninformed) thoughts: Your last point seems epistemic. That is, since you can't tell if an object is grabbable just by looking, you don't see it as affording anything. But this seems implausible to me. Surely there's some plausibility to the thought that I can see an object as F even though, for some reason, I don't know that it is F and, further, know that I don't know that it is F (still, it looks F)?

    On Campbell's original point, and here's where my ignorance of Gibson really kicks in, isn't it implausible for him to suppose that just because I don't know what something is for, I don't see it as affording anything in particular. Doesn't it look bump-into-able or, for that matter, see-able? Suppose that Gibson thinks that to see an object one must see it as being some way or other (something that is certainly questionable - but I take it that Campbell's point is assuming it), then surely it is reasonably easy to construct some affordance or other out of whatever ways it appears? It's very difficult to think of a categorical property that doesn't support at least one affordance that the thing will be seen to have by all perceivers (well, I guess some might want to query that for animals and infants). So I'm not sure that I buy Campbell's line, even if Noe is wrong.

    ReplyDelete
  9. No, I do see what he's getting at. It's just that he seems to be implicitly defining affordance in a way that doesn't make sense:

    A Gibsonian construct offers an affordance only if there are no unforeseeable consequences to acting on the apparent affordance it offers.

    OT: Ken - I have taken a cut at writing up in some detail the unvocalized-sentence view of thoughts here:

    http://secularlogic.blogspot.com/

    It's not much more in essence than what I said in my comments, but I spelled out the essential argument as completely as I can.

    ReplyDelete
  10. I see what he's getting at - I'm telling you he is incorrect. He clearly doesn't understand information, ie how affordance come to be specified in energy arrays. Without that he can never be correct in his characterisation of Gibson.

    Take some physical object. Is it a grabbable thing (to use a TSRM example of an affordance). You can't tell just by looking at it and moving around it.
    You probably can actually, but even if you can't, then you simply are unable to perceive that affordance. The object doesn't disappear, that's idiotic. Not all affordances are always available; that's hardly a problem, because EP recognises the role of information in all this - no information, no access to the affordance, even if (from a third person perspective) you might be able to define one.

    Charles, you're on the right track. But the issue to remember is not whether the affordance can be defined, but whether there is information for it. So in the stool example, there is only information that it affords sitting until you sit on it, at which point there is information that it didn't.

    ReplyDelete
  11. Andrew -

    My "definition" of affordance was a sarcastic interpretation of what appears to be Campbell's view. Mine is just as you describe.

    ReplyDelete
  12. @Charles and his "definition",
    Also, to the limited extent I understand Gibson and affordances, it strikes me that the consequence of Gibson's approach is that an object becomes defined not by its properties as we usually think of them but by the union of all its affordances. Yes?

    Well, in truth, EP seems to me equivocal on the metaphysics of affordances. Sometimes they are supposed to be objective dispositional properties of things, but they also have this observer relativity about them. So, I don't know if this is right. It is, however, I think orthogonal to the issue most commented on here, namely, the perceivability of affordances.

    ReplyDelete
  13. Hi, Joel,

    Thanks for stopping by and commenting.

    On the epistemic point:
    I think that you are right that, implicit in the Campbell argument is the idea that if I perceive something as X, then I will know that it is X. (And, I think that Turvey, Shaw, Mace, and Reed, seem to endorse this principle as a way of enabling perception to lead to knowledge.)

    With you, I wonder if the principle is true (although I have not commented on this.) But, this does seem plausible: If I perceive (read factively, as a "success term", right?) something as a tennis ball, then I know it's a tennis ball.

    But, do you have a counterexample in mind?

    ReplyDelete
  14. @Joel,
    Regarding your second point, let me rephrase it a bit to make sure I get it. That is, let me set aside the talk about what an object is for (since Andrew objects and it's not crucial, I think).

    So, take what non-Gibsonians would call a stool. It's going to have lots of different affordances, say, ten, right? So, you are entertaining conceding to Campbell that a human being might not be able to perceive, say, five of the affordances, but miss five. So, that Campbell would be wrong to say that simply that one does not perceive affordances, right?

    I think this is what you are driving at. And, I think that this is right. But, I think that Campbell's strategy is to pick some typical physical object and argue that the affordances (like my example of graspable) that EP says one should be able to perceive, one cannot know what it affords just by exploring it.

    And going back to your first point. Campbell can probably run his object without using the principle that, if you can perceive X as A, then you know that X is A. Instead, he could probably make do with something like, if you can perceive X as A, then this will have some behavioral or cognitive manifestation.

    ReplyDelete
  15. Well, in truth, EP seems to me equivocal on the metaphysics of affordances. Sometimes they are supposed to be objective dispositional properties of things, but they also have this observer relativity about them. So, I don't know if this is right. It is, however, I think orthogonal to the issue most commented on here, namely, the perceivability of affordances.
    It's not that people are equivocal; it's that there is genuine disagreement. Gibson is fairly clear about what he means, but not exactly precise; Turvey (1992) was an effort to formalise the ontology as dispositional (which I agree with these days). Others (notably Chemero and Stoffregan) think affordances must be relations, but I think this is an error.

    It's also not orthogonal to the issue of perceivability; the ontology matters because affordances must be things you could, in principle, perceive.

    But, I think that Campbell's strategy is to pick some typical physical object and argue that the affordances (like my example of graspable) that EP says one should be able to perceive, one cannot know what it affords just by exploring it.
    Surely it matters that this isn't true?

    ReplyDelete
  16. So, I do think that Campbell seems not to get the objection to the perception of affordances down right, e.g. the stuff about being invisible, but I think one can see the problem he is driving at: the affordances that EP says should be directly perceived are not directly perceived.

    ReplyDelete
  17. But, I think that Campbell's strategy is to pick some typical physical object and argue that the affordances (like my example of graspable) that EP says one should be able to perceive, one cannot know what it affords just by exploring it.
    Surely it matters that this isn't true?


    Yes, the truth value matters, but what you've been faulting him on is that he doesn't set it up in just the right Gibsonian way.

    At the start of Gibson, 1979, Chapter 8, you get this picture that there are surfaces, they constitute affordances, the affordances are specified by information (in light perhaps), and then (presumably) this information is picked up.

    What Campbell is aiming for is a counterexample What matters, as you say, is whether he has one.

    ReplyDelete
  18. And he doesn't, because he sets it up wrong. He thinks you have to know what the object is for in order to be able to match that knowledge to the perceived affordance. But perception of the affordance is how you come to know what the object is for/does, as Charles rightly said.

    Detection of the information and perception of the affordance aren't instantly synonymous; you have to learn what the information means, via exploration. Babies can tell the difference between point light displays of an event (a ball being hit then rolling to a stop) played backwards and forwards (sensitivity to the information) but show no preference for the right one (insensitivity to the underlying dynamics; Wickelgren & Bingham 2001). The latter takes learning, but (for Gibson, anyway) doesn't require any additional, external knowledge of the dynamics (mostly because you can't ever have such a thing, independent of perception).

    Campbell can disagree with this all he likes, but I'm not going to take his objections seriously until he characterises the problem correctly. In that quote, he is clearly misrepresenting (:v) Gibson's approach.

    ReplyDelete
  19. Pursuing Joel's comment on the epistemics, I'd like to understand if in the Gibson scheme "perception" is just detecting that something is there and that the something offers certain affordances, or if it also includes capabilities like knowing "what kind of thing the something is" and thus perhaps knowing "what the something is for".

    As all of us were taught as kids, some things that afford sitting-on aren't meant to be sat-on because that's not what they are "for". But knowing the difference is epistemic, no?

    And re disposition vs relation. When I use the phrase "something offers affordances", I have in mind that the something creates in the perceiver a disposition (not necessarily obeyed) to take advantage of the offered affordances. But then the disposition is associated with the perceiver, not the something. And in any event, the scenario as I envision it seems necessarily to involve a relation. I'm clearly confused.

    ReplyDelete
  20. Ken,

    Some rambling thoughts.

    The epistemic point (going a little off topic, I'm afraid): Suppose I see a red ball as red (my experience has a character such that, all other things being equal, I would judge it to be red, etc.). However, a normally reliable source (in this case, falsely) tells me that the ball is, in fact, white but under a red light. I believe the source and, so, do not believe that he ball is red. On my view, I don't know that the ball is red. So, seeing that an object is F is not sufficient for knowing that it is F.

    Now, things may be different for seeing that the ball is red (seeing that, rather than seeing as). Quite plausibly, seeing that entials knowing. Of course, if one bought that claim, one would have to say that, in the above scenario, I don't see that the ball is red.

    On whether Campbell must assume any of these epistemic principles, I agree that he probably doesn't. Campbell says that you don't know what it affords, but really the point (isn't it?) is that you don't see it as affording anything. I take it that his talk of what you know is just supposed to be defeasible evidence (nothing more) for claims about what you see the object as.

    My other points was just that it is implausible that you could fail to see ANY of an object's affordances. For example, one can see it, then it is seen as see-able. Or if you don't like that, since seeing isn't an action, it is seen as attend-able. Perhaps this is to misuse Gibson's notion. I don't know. But I'm not sure why it would be - being attendable looks like a dispositional property of the object that enables certain types of behaviour in perceivers, etc. Plus, surely things DO look attend-able, see-able, visable (what would something have to look like to not look visible?!). So, my thought was just that even if Cambpell is right (which you all now seem to agree that he is not or, at least, his point is not well expressed) that if I see an object I must see it as having at least one affordance, his example doesn't show what he suggests (about the thing being visible or not).

    But what you say suggests something else (that I hadn't thought of). Perhaps Campbell's point is that (as Andrew & Charles have said) seeing that an object is graspable is supposed to be how one comes to know that it is graspable, but we can't know that it is graspable just by looking, so we can't see that it is grapable. This, I take it is valid, but I deny the second premise. I think hat you can know that it is graspable just by looking. The fact that it might electrocute you seems to play a role similar to the thought that a zebra might be a disguised mule (therefore, one cannot know that something is a zebra just by looking). Of course, one needs to do some work to deny that, but the premise certainly isn't obviously true.

    However, I must admit to not having read the Campbell, so I'm not sure how to interpret him. One thought is that his wider target is the idea that we ONLY see the dispositional properties of things (this is something that he has argued against before), but rather we have to be aware of the categorical bases of those dispositions. If that it his aim, and he is interpreting Gibson as claiming that we ONLY see an object's affordances, that would make sense of his attempted reductio via the invisibility comment.

    Anyway, enough already...

    ReplyDelete
  21. For Gibson, perception is not just detection; the knowledge of what the affordance means is part of the game.

    As all of us were taught as kids, some things that afford sitting-on aren't meant to be sat-on because that's not what they are "for". But knowing the difference is epistemic, no?
    As far as perception goes, these things still afford sitting. You presumably then have to learn about that specific object in order to also know not to sit on it.

    When I use the phrase "something offers affordances", I have in mind that the something creates in the perceiver a disposition (not necessarily obeyed) to take advantage of the offered affordances. But then the disposition is associated with the perceiver, not the something.
    Dispositions come in matching pairs; salt is soluble in water because water has the matching disposition needed. In Turvey's terms for affordances, the object in the world has a disposition (the affordance) and the perceiver has a complementary effectivity to effect that affordance. The symmetry remains, but the terms help keep things in order.

    So the object in the world does have dispositions; that's quite important. The whole thing is indeed relational, but the affordance itself is not a relation (unless you're Tony Chemero).

    See here, here, and here for the lay of the land.

    ReplyDelete
  22. Joel -

    You may already know this, but in Sellars' Empiricism & Phil of Mind he addresses precisely your issue of the "red/white" ball (his example is a "blue/green" suit). His discussion is long and (at least for me) rather difficult, but IMO well worth the effort. He distinguishes three cases which, as you note, differ in degree of confidence (he calls it "endorsement"). If you are sure the ball is red, you might say "the ball is red". If you are concerned that it may not be, you might say "the ball appears red". If you are not even sure about the ball itself, you might say "there appears to be a red ball over there".

    That's a trivialization of his discussion, but in case you haven't read the essay perhaps it will whet your appetite.

    Andrew -

    Thanks. I think I'm getting it and am drafting some comments on Sabrina's Chap 6 summary (actually, mostly on Bingham). Hope the thought of that doesn't ruin your weekend!

    ReplyDelete
  23. We look forward to it (and Geoff's stuff is always good source material, he was my PhD advisor and is much smarter than me :)

    Joel:
    I think hat you can know that it is graspable just by looking. The fact that it might electrocute you seems to play a role similar to the thought that a zebra might be a disguised mule (therefore, one cannot know that something is a zebra just by looking).
    The intuition here, which Gibson formalised, is that there is information (structure in light, for example) for grasp-ability and (probably) none for whether it will electrocute you. Information and affordances go hand in hand, for Gibson, but most people only know of the latter. Campbell is clearly someone who knows a definition of affordances but nothing of information.

    ReplyDelete
  24. Hi, Joel,

    The ball example is right on. And, now that I think about it, you could probably get up a Gettier-like example as well.

    And, I think we agree that the issue is where an object X might have the dispositional property of affording A without you being able to perceive this.

    I now see why you were on about even though one might not perceive, say, affordance A1, one might still be able to perceive, say, affordance A2, so that the object would not be invisible. I evidently read right past what Campbell wrote, because I had the basic idea of what he was getting at and imposed my own interpretation.

    I guess I also agree with you that, if you see an object, then it has the affordance of being see-able. That is kind of a logical truth, right? But, there still remains another kind of problem for EP that we have not touched on. It could be that X affords A and one knows/believes that X affords A, but that one does not perceive that X affords A. Maybe this would happen when the moon affords being landed upon by a lunar module launched by a Saturn V rocket. Maybe this would happen when a pile of salt affords dissolution in water. So, an object might have the affordance of being-seeable or graspable, but I might perceive that the object is graspable, but only believe that it is graspable.

    Let me think a bit more about zebra cases...

    ReplyDelete
  25. So, Joel, let's switch from zebras to barn facades. I'm assuming you're a philosopher, so I'll go old school with a thought experiment.

    So, Barney is looking at a barn in normal country, not barn facade country. He's having a particular perceptual experience that we should construe non-factively. Now, teleport in all around him barn facades all over the place except for the one he is looking at. He still non-factively perceives a barn, but he does not know a barn. The perceptual experience does not change with the teleportation of the barns, even though the epistemic situation does, right?

    Now do the exercise using, say, a stool. Stan is looking at a normal stool and having a particular perceptual experience. This is not going to change with the teleportation in of fake stools.

    So, what I trying to get at is that you cannot dismiss Charles stool kinds of examples just be remembering some of the parade cases in epistemology. Does this sound right?

    Now, I don't know what EPists are going to say about perceptual experiences, but that's another matter.

    ReplyDelete
  26. typo: "So, an object might have the affordance of being-seeable or graspable, but I might perceive that the object is graspable, but only believe that it is graspable."

    Should be
    "So, an object might have the affordance of being-seeable or graspable, but I might not perceive that the object is graspable, but only believe that it is graspable."

    ReplyDelete
  27. Here's maybe another way to tackle the perception of affordance issue. Take two objects that are physically the same on the outside, but differ in their affordances. Are they perceived to be different?

    Take two boxes that are physically identical on the outside, but one is wired to explode or electrocute you, if touched. One is graspable, but the other is not. Are they visually perceived to be different?

    ReplyDelete
  28. But, there still remains another kind of problem for EP that we have not touched on. It could be that X affords A and one knows/believes that X affords A, but that one does not perceive that X affords A.
    ...
    So, an object might have the affordance of being-seeable or graspable, but I might not perceive that the object is graspable, but only believe that it is graspable.

    If you cannot perceive that it is graspable (ie there is no information for the affordance), on what basis can you come to believe that it is graspable?

    Take two boxes that are physically identical on the outside, but one is wired to explode or electrocute you, if touched. One is graspable, but the other is not. Are they visually perceived to be different?
    Not if the difference has no informational consequences. Also note that both boxes afford grasping to the same extent because they are identical on the outside (ie on the surface where you would come into contact).

    This is why I like change blindness studies. The trick to getting these to work is that you have to somehow mask the visual transient of the abrupt change (ie remove the information that something abruptly changed). No information, no change (same with the case of friction and locomotion I was posting about this week).

    ReplyDelete
  29. Andrew will correct me if I'm wrong, but I take it that his emphasis on the tie between information and affordance is that "you can't have one without the other". So, from a specific would-be perceiver's POV, if the information isn't adequate for that would-be perceiver to detect an affordance, the affordance can't be actualized by that would-be perceiver. Ie, the relationship between affordance and perceiver is cooperative.

    This is a bit tricky because as I understand it, one of the requirements of the affordance concept is that affordances must be available even in the absence of any would-be perceiver - a requirement satisfied by the dispositional approach. But that does not mean that an arbitrary would-be perceiver can perceive the offered affordance. And presumably, that's where the epistemics come in. The affordances that a specific would-be perceiver is capable of actualizing depend on the would-be perceiver's epistemic capabilities with respect to the information in the received light.

    If I have that more-or-less right, I think the response to various statements/questions would be:

    "an object X might have the dispositional property of affording A without you being able to perceive this"

    "one might not perceive, say, affordance A1, one might still be able to perceive, say, affordance A2"

    Yes to both.

    "if you see an object, then it has the affordance of being see-able."

    My inclination is to avoid the word "see" in this context because it's meaning is ambiguous. An affordance - in essence a "call to action" to the would-be perceiver - becomes actualizable only after being perceived. I can't quite parse the quote, but it seems inconsistent with that order of events.

    "It could be that X affords A and one knows/believes that X affords A, but that one does not perceive that X affords A."

    As I understand it, the perception of affordance comes first. This quote seems to assume the ability to detect and identify an object and then to make an epistemic analysis of the affordances that such objects offer, some of which are detectable. I don't think that's the way it works (in EP). That's why I asked above if an "object" is defined by the union of all it's perceived affordances.

    Re consequences as opposed to supposedly undetectable affordances. Below is part of a comment I'm composing for Andrew and Sabrina's blog, but I'm going to cross post it here since it seems directly relevant to the present discussion.

    ===================================
    As I understand it so far, an affordance offered by a fly ball is the opportunity to catch it. Whether an attempt is made, and if so, whether the attempt is successful, seem separate issues. In fact, although success will necessarily produce an out, the consequence of failure is contingent - it may be a home run, but it also may be any of several other possibilities, including an out. Does it really make sense to consider that a fly ball affords the perceiver opportunities for every possible consequence of the affordance's actualization? ... Taken to an extreme, one could argue that successfully catching the fly ball results in an out which ends the game - the last in the series, so the fielder's team wins the series; as a consequence he receives a bonus, buys a hot car, and is killed while driving it too fast on a slippery road. Shall we call all of these possible consequences affordances offered by the fly ball?
    ===========================

    ReplyDelete
  30. "on what basis can you come to believe that it is graspable? "

    You are schizophrenic and you hear a voice that tells you that the box is graspable.

    Also note that both boxes afford grasping to the same extent because they are identical on the outside (ie on the surface where you would come into contact).

    I guess there is room to fiddle around with what one means by graspable. So, how about pick-up-able? Are they equally pick-up-able?

    ReplyDelete
  31. How about take two piles of white powder made of crystals of the same size. One is edible, the other is poisonous. Do you visually perceive them differently?

    ReplyDelete
  32. How about two samples of clear liquids, one water and one a poison. Can you visually perceive that one is potable, but the other not?

    ReplyDelete
  33. Charles, up to the fly ball bit you are basically bang on with the eco-psych analysis.

    As I understand it, the perception of affordance comes first. This quote seems to assume the ability to detect and identify an object and then to make an epistemic analysis of the affordances that such objects offer, some of which are detectable. I don't think that's the way it works (in EP).
    This is a fair summary of the difference between a direct perception account of how we use affordances vs the indirect, more cognitive account. Affordances could (in principle) be used either way, but they are only of use to a direct perception account if they are specified in information, ie perceivable. Hence the second part of the analysis must always be about information.

    You are schizophrenic and you hear a voice that tells you that the box is graspable.
    On what basis did 'graspable' become the kind of thing a voice had to say? (I'm not just being pedantic - I'm saying it doesn't make sense to claim you can get the belief 'for free' as it were).

    Pick-up-able: any affordance composed of the properties the objects share will be shared between the objects. Prehension affordances are mostly about object size, etc, things which are apparently the same.

    Your next two examples have exactly the same answer; if, and only if, there is information about the difference, then that difference is potentially discriminable and the property perceivable.

    Things that are not seen to be different are not treated differently, and the issue is not 'is there poison?' but 'is there information about the poison?'. This is just the shark all over again.

    ReplyDelete
  34. The schizophrenia is tongue in check. There are abundant ways of coming to have a belief that something is graspable other than via perception. You can have a computer run a test, then tell you. Your friend can tell you. You can see other things that are graspable the infer by analogy. To use the Sellarsian language, there could be inferential beliefs.

    Here is Gibson, 1979, p. 127:
    "And if there is information in light for the perception of surfaces, is there information for the perception of what they afford? Perhaps the composition and layout of surfaces constitutes what they afford. If so, to perceive them is to perceive what they afford."

    There is, in my examples, information in the light for the perception of surfaces. Right? Then Gibson seems to be saying that that is enough for perceiving affordances.

    Or, maybe put it this way. Contrary to what Gibson appears to suggest, the information in the light for the perception of surfaces does not seem to provide information for affordance.

    ReplyDelete
  35. TSRM appear to do things differently than does Gibson regarding how one light gets structured for perception. (Cf. pp. 264f).

    So, TSRM suggest that dispositional properties have anchoring properties that make the dispositional properties what they are. So, chemical features of salts anchor their solubility. Climability of X is anchored by such things as orientation and opacity of X. These anchoring properties are, then, supposed to structure light in such a way as to enable perception of the affordance. That's my read.

    So, here's the problem. The anchoring properties that make a thing edible do not appear to structure light in such a way as to enable the perceiving of edibility. The anchoring properties of potability do not structure light in such a way as as to enable the perceiving of potability. The anchoring properties that make a thing capable of exploding do not structure light in such a way as to enable the perceiving of its explodability.

    ReplyDelete
  36. So, here's the problem. The anchoring properties that make a thing edible do not appear to structure light in such a way as to enable the perceiving of edibility. The anchoring properties of potability do not structure light in such a way as as to enable the perceiving of potability. The anchoring properties that make a thing capable of exploding do not structure light in such a way as to enable the perceiving of its explodability.
    But this isn't a problem. One, edibility is the kind of thing that might indeed structure energy appropriately (the shark example is an in principle example.

    But more importantly, if the properties do not structure light in the right way, then there won't be information and you won't be able to perceive that affordance. So it's not a problem for ecological psychology because no one ever claimed that every property you can think of is specified! This is why information is so critical: affordances are of no use to perception if not specified. (You should read my post on friction.)

    So with respect to the Gibson quote:
    There is, in my examples, information in the light for the perception of surfaces. Right? Then Gibson seems to be saying that that is enough for perceiving affordances.
    So you will be able to perceive affordances for prehension, etc. But that's not to say you get everything!

    An affordance can only be perceived if it creates information and an organism detects that information. If it doesn't create information (eg if the two clear liquids are not visibly different as a result of the presence or absence of poison) then you have no basis to perceive the presence or absence of poison.

    This isn't a complicated claim and it's certainly not posing any problems for eco-psych. And the scientific attack on this is simple: you first look to see whether people are able to shape their behaviour differently under the two conditions - if they can then there will be a property driving the difference and information for that property. You then analyse the task to identify candidate properties, figure out the potential information for each, and then experimentally manipulate things until you find the information. This is just the ecological research programme; upcoming posts on coordination experiments will deal with this.

    ReplyDelete
  37. There are (at least) three different accounts one might have of what structures light:
    1. Surfaces structure light.
    2. Anchoring properties structure light.
    3. Affordances structure light.

    Gibson, 1979, chapter 8 seems to take option 1.
    TSRM, 1981, circa pp. 264 seem to take option 2.
    You seem to take option 3.

    The example of edible is not a case of an anchoring property structuring light (or the ambient energy). Edible is an affordance; it is a disposition. Anchoring properties are those that make affordances what they are.

    Now, it could be that the properties that make a fish edible (edibility's anchoring properties) are also the ones that make it generate an electric field. That's an empirical question.

    So you will be able to perceive affordances for prehension, etc. But that's not to say you get everything But, Gibson doesn't say that that light from surfaces enables you to perceive affordances for prehension; he implies that the light from surfaces enables you to perceive affordances. (Check out the first page of Gibson, 1979, chapter 8). So, look at it this. It seems to be common ground here that surfaces do not structure light in such a way as to specify all affordances. Ok. So, the more focused question that you advancing, then, is whether surfaces structures light in such a way as to specify all (?) prehension affordances. But, the answer to that also appears to be no. It is common ground that the surfaces of the two boxes do not structure the light in such a way as to specify pick-up-ability (or not).

    ReplyDelete
  38. And the scientific attack on this is simple: you first look to see whether people are able to shape their behaviour differently under the two conditions - if they can then there will be a property driving the difference and information for that property. You then analyse the task to identify candidate properties, figure out the potential information for each, and then experimentally manipulate things until you find the information. This is just the ecological research programme;

    Sure. Give subjects a task and see what they respond to. Maybe it's affordances, maybe it is physical properties of light. That's an empirical question. EPists say it's affordances and the boxes, the powder, and the transparent liquid are all directed at this idea that subjects are responding to affordances. They do not seem to be; they appear to be responding to physical properties. It's not the experimental set up that trips up EP; it appears to be (what we take to be) the results of experiments.

    ReplyDelete
  39. There are (at least) three different accounts one might have of what structures light:
    1. Surfaces structure light.
    2. Anchoring properties structure light.
    3. Affordances structure light.

    The anchoring properties of the affordance lead to a surface which structures light. It seems like these are simply labels for various levels of analysis.

    But, Gibson doesn't say that that light from surfaces enables you to perceive affordances for prehension; he implies that the light from surfaces enables you to perceive affordances
    Sure. And which affordances are available are the ones which have structured light. He's still not saying 'any old dispositional property of the object can be perceived'. This is why he spends so much time laying out that which properties are relevant and which of those are specified is an empirical matter.

    But, the answer to that also appears to be no. It is common ground that the surfaces of the two boxes do not structure the light in such a way as to specify pick-up-ability (or not).
    Hardly common ground. You're making an error in thinking that the box that will electrocute you doesn't afford prehension. If the surfaces of the box are such that my hand can grasp it, then it affords prehension. The fact that I will die if I pick up the wrong one is neither here nor there, unless there is information about that. This is why things you can grasp that electrocute you tend to come with signs.

    ReplyDelete
  40. Sure. Give subjects a task and see what they respond to. Maybe it's affordances, maybe it is physical properties of light.
    And it's our job to run the study that enables us to tell the difference. That's why this is an empirical research programme.

    ReplyDelete
  41. 1. Surfaces structure light.
    2. Anchoring properties structure light.
    3. Affordances structure light.
    The anchoring properties of the affordance lead to a surface which structures light.
    It seems like these are simply labels for various levels of analysis.


    They might to you, but they did not to TSRM. They gave an argument that they believed connected the information generated by anchoring properties to the information generated by dispositions. It's in the second paragraph on p. 266.

    It's graspable, even if you can't grasp it.
    It's pick-up-able, even if you can't pick it up.
    Got it.

    ReplyDelete
  42. It's graspable, even if you can't grasp it.
    It's pick-up-able, even if you can't pick it up.
    Got it.

    I'm not sure that you have, based on this. Could you be a little more specific?

    ReplyDelete
  43. That was sarcastic, but here is how the example goes.

    So, you have normal box, as for a wedding engagement ring. So, Gibson might think that the surface of the box structures light in such a way as to enable a perceived to perceive the box as pick-up-able (a prehensible affordance). Not unreasonable.

    But, then you have a second box that is physically just like the first on the outside, but is wired with explosives so that if you touch it it will explode. You can't pick it up; it would explode. It's not pick-up-able, because you can't pick it up.

    Now, if the first box were able to structure light in such a way as to enable an observer to perceive pick-up-ability, then the second one would be able to structure light in such a way as to enable an observer to perceive pick-up-ability.

    But, the second box doesn't enable an observer to perceive pick-up-ability, because the observer cannot pick it up. It will explode.

    So, it looks like we shouldn't say in the first case that the observer can perceive pick-up-ability.

    One move you were making is to stop talking about pick-up-ability and graspability and start talking about prehensibility. This, however, is just an obfuscation.

    ReplyDelete
  44. After reviewing the recent flurry of comments, here are two major disconnects I see:

    1. Ken seems to be interpreting "affordance" as an inherent property of an entity in the environment, a property that may or may not be perceivable. Ie, it is extant whether perceived or not. Hence, statements like "Edible is an affordance".

    As I understand it, Andrew (and, I assume, Gibson) define "affordance" as an offer to a specific perceiver of an opportunity for action, an opportunity that is derived by that perceiver from the information in light structured by its reflection off of the entity. Ie, no perceiver or no perception, no affordance.

    Thus, from the perspective of Gibson's scheme, I'm not quite sure how to parse "the information in the light for the perception of surfaces does not seem to provide information for affordance". Whether the information in received light offers affordances depends on the perceiver. If the perceiver can sit and the information is such as to suggest a sit-on-able surface appropriate to the perceiver, then there emerges an affordance. Otherwise, not.

    But then Andrew says re edibility:

    "if the properties do not structure light in the right way, then there won't be information and you won't be able to perceive that affordance."

    which is confusing since it suggests that there can be an unperceived affordance. If I have the concept straight, it seems that this should be something like:

    "if the structure of the light doesn't provide the information necessary for the perceiver to derive the affordance "edible" from it - ie, an offered opportunity to eat the source of the light - then there is no affordance."

    2. Ken is still taking consequences of actualization of an affordance as having existential significance for the affordance - if you act on an apparent affordance and as a direct consequence are killed, it wasn't really an affordance. This is not consistent with Andrew's view. Until this gets resolved it appears no convergence is possible.

    ReplyDelete
  45. On reflection, I see that I remain confused about the relationship between an affordance and perception. It seems that some additions to the vocabulary might help.

    An affordance seems to have three possible "states":

    - the "persistent" state prior to being perceived by a specific perceiver

    - the "pre-actualized" state (with respect to a specific perceiver) in which it is perceived but not actualized

    - the "actualized" state, which may be a version of the actualized affordance (eg, a park bench the sit-on-ability of which is altered once actualized by one sitter), a new affordance, or non-existence (eg, a match which affords striking no longer affords striking after actualization)

    If this is right, then in my last comment instead of saying things like "no perceiver or no perception, no affordance" I should have said things like "no perceiver or no perception, no pre-actualization of any affordances".

    And Ken's statement becomes something like "the information in the light for the perception of surfaces does not seem to provide the information necessary for pre-actualizing all affordances" - which I can parse.

    But that raises the question: In Gibson's scheme are Gibsonian entities like surfaces perceivable per se or they only perceivable as pre-actualized affordances? My guess is the latter. Andrew?

    And a description of the exploding box scenario seems to become something like:

    One box has the persistent affordance "explodability", but in the scenario as described, the information necessary for pre-actualization is not available to a human perceiver from only reflected light.

    ReplyDelete
  46. But, the second box doesn't enable an observer to perceive pick-up-ability, because the observer cannot pick it up. It will explode.
    As Charles notes, this is where the error happens. Both boxes afford being picked up in the same way, because such affordances are composed of the properties that the boxes share. The fact that one then kills you doesn't mean the box doesn't look like you can pick it up; and as you've set it out, there is no information for the property 'can be picked up but will then kill you' because the two boxes are identical. If that property was specified, and the observer could detect it, then the boxes would not be identical.

    I was just using prehension as a more general label for graspability and pick-up-ability because it's shorter to type. I wasn't changing what affordances we were talking about.

    Charles:
    You can have an unperceived affordance. This is what learning is for, coming to be able to perceive an affordance you were previously unable to perceive.

    Affordances are real dispositional properties of the environment; they pre-date the observer and (I think) have to in order for evolution to have found them. They can do this because they are composed of suitably anchored properties. They are perceived relationally (in terms of the observer's effectivities; that distance is reachable by me) but are not themselves a relation. Affordances and information are not the same thing; the analysis must keep 'world' variables and 'information' variables separate, and not confuse the two.

    ReplyDelete
  47. Hi, Charles,

    I suppose that affordances are organism relative, but I've been surpressing that detail. I think you can read the foregoing as relativized to humans. So, edible-for-a-human is an affordance. Make other changes mutatis mutandis. TSRM take this liberty of abbreviation (cf., e.g., p. 261), so I do too.

    TSRM treat affordances as a type of disposition, like fragility. So, a chair affords sit-on-ability for a human even when there is no human around to sit on it. Similarly, a Waterford crystal wine glass is fragile, even when there is no human (or anything else) around to break it.

    Maybe Gibson thinks affordances are dispositions; I don't recall off the top of my head.

    And, I think you can run the examples in the vocabulary you have introduced.

    But, I've tried not to tie my argument too closely to the idea of affordances as dispositions or whatever other gloss one might want to place on them.

    ReplyDelete
  48. The fact that one then kills you doesn't mean the box doesn't look like you can pick it up;

    I can concede, if only for the sake of argument, that both boxes look like you can pick them up.

    But, only one box affords pick-up-ability (for a human) in the sense that only can be picked up (by a human).

    The first one affords pick-up-ability (for a human), because it can be picked up (by a human).

    The second does not afford pick-up-ability (for a human), because it cannot be picked up (by a human).

    Charles will be all over the distinction between looking pick-up-able and being pick-up-able.

    ReplyDelete
  49. It does afford pick-up-ability; it will just also kill you.

    Also if you are going to talk about affordances you really should bone up on the dispositional and relational accounts, because otherwise it's not clear what you'll be talking about.

    ReplyDelete
  50. It does afford pick-up-ability; it will just also kill you.

    So, it's true: It's pick-up-able, but you can't pick it up.

    ReplyDelete
  51. Not to intrude on the conversation, but this:

    "So, it's true: It's pick-up-able, but you can't pick it up."

    seems like an equivocation. If we reformulate it as "You can pick-it-up, but you can't pick-it-up", the equivocation lies in the "can". The first "can" implies the possibility of picking the thing up; the second "can" is prudential: you *shouldn't* pick it up, otherwise you will die. The second can depends on the former: if you literally couldn't pick the thing up, then there would be no worry about activating the explosives, as they are activated precisely by picking the box up!

    ReplyDelete
  52. Ok. I read those. No help.

    You say the box is pick-up-able. But, you just can't bring yourself to admit that one cannot pick up the box. You are dancing around this. Yes, it looks like you can pick it up. Yes, it will kill you. But, most importantly, a human can't pick it up. A human can touch it, in which case it will blow up. But, a human cannot pick it up.

    ReplyDelete
  53. Hi, Daniel,

    Thanks for your comment.

    I certainly don't mean to equivocate here. After all, part of what makes the view look so absurd is the assumption of a non-equivocal reading. But, I think I have set up the example in a way that allows a non-equivocal reading.

    I would take the modality in both cases to be nomological possibility. (Now, here I admit I could be mistaken, because I don't know that much about the metaphysics of disposition manifestation, for example.)

    So, I did not mean for the second modality to be merely prudential. It is not merely that it would be unwise for you to pick up the box; it is nomologically impossible for you to pick up the box. Why is it nomologically impossible to pick up the box? Well, it is nomologically impossible to pick up the box, since merely touching it will cause it to explode and it is nomologically impossible for one to pick up the box without touching it. (One might have to add a few whistles and bells to this, but I am trying to keep things simple for the blog.)

    Now, it might be true that the "prudential can" depends on the "nomological can" (that you literally can't pick it up). (That sounds like it could be a little strong. Maybe the prudential claim is only that you shouldn't try to pick that up, unless you want to die, or something like that.) But, that dependency does not show that one cannot read the second modality as nomological possibility.

    So, I'm sceptical that your particular version of this line of argument. On the other hand, I don't know enough to be able to forestall certain other variations one might think of. I just don't know that much about the metaphysics of modality. But, that's probably worth thinking about.

    Part of why I'm not thinking of writing any of this up any time is that there is a sizeable philosophical literature on dispositions and I know essentially none of it. I don't know when I could get up to speed on that stuff. Maybe next year.

    ReplyDelete
  54. Aha, finally. The disconnect is that none of us read Ken's scenario carefully. He's right that the exploding box doesn't afford pick-up-ability. But that's because (in my terminology) it doesn't offer the persistent affordance "pick-up-able". (I'm assuming that implicit in the definition of affordance that one must be actualizable by at least some perceiver.)

    So I think the comprehensive statement would be something like:

    The perceiver perceives the exploding box as having the pre-actualized affordance "pick-up-able", but it actually doesn't. The box doesn't offer that apparent affordance because it can't be actualized by any perceiver.

    Keep in mind that the core issue here is perception. Presumably no one is arguing that one can perceive the imperceptible no matter what concept of the perceptual process one assumes.

    ReplyDelete
  55. Hi, Charles,
    Yes, I think that the exploding box doesn't afford pick-up-ability, the normal box does afford pick-up-ability, but one perceives them in the same way. That's one way of running the problem.

    Andrew seems to be getting caught up in another way. He says that the exploding box is pick-up-able, but is desparately trying to avoid admitting also that it cannot be picked up (because it would explode first). And, Daniel senses that Andrew is on the verge of an absurdity, so it trying the very reasonable strategy of looking for a possible equivocation on my part.

    [N.B.: You guys are blog animals! I almost never read this far down into comments. Thank you for your readership.]

    ReplyDelete
  56. I bet Andrew was making the same mistake I was. We missed the touching bit and assumed it was picking-up that triggered the explosion, not touching. In which case, the box would offer the pre-actualized affordance pick-up-able, but as soon as the affordance was actualized, the box would explode.

    But in the "touching" scenario that you actually described, I agree with the first paragraph of your last comment.

    ReplyDelete
  57. I guess we shall see tomorrow. =)

    ReplyDelete
  58. I had missed the touch bit. But it still doesn't matter because this is about information. You pitched this as some kind of problem for eco-psych, but we don't have to even break a sweat on it. A person perceives both boxes to be pick-up-able and tries to effect that affordance. One of the boxes works, the other one kills the person. So on further exploration more information came to light. But at all times the person was acting happily in accordance with the perceived affordances of the environment.

    This is just the shark all over again.

    ReplyDelete
  59. EP, per se, may not have problems, but it looks like Gibson and TSRM do. Gibson suggests that surfaces provide information for the perception of affordances. TSRM claim that anchoring properties for affordances provide information for the perception of affordances. The box example challenges both of these. (I'll post the Gibson text; I may post the TSRM. I hesitate because it is long and rambling.)

    It's another matter to claim subject S perceives affordance A iff there is information that specifies the affordance. I'm not challenging that (and neither I think was Campbell.)

    You suggest that further exploration brings the non-pick-up-ability to light. But, I don't think you want to say this. You don't want to say that affordances are only perceived after what they afford has been manifest. You don't want to say, for example, that one does not perceive the sit-on-ability of a chair until one sits on the chair. EP wants the perception of affordances prior to their manifestation.

    ReplyDelete
  60. But at all times the person was acting happily in accordance with the perceived affordances of the environment.

    But, this is not an innocuous concession for EP. Suppose that the person does have the perception that the exploding box affords pick-up-ability. This cannot be because the subject has perceived that the box is actually pick-up-able. (It's not.) This opens the door to the idea that the person perceives the box to be pick-up-able because she has inferred this from the surface properties of the light and her background beliefs. If the Gibsonian or TSRM account does not work in this case. Then, that opens the door for cognitivism to step into the breach. (This is, indeed, a weakness in Noe's line that it is only some affordances that are perceived by way of anchoring properties or surfaces providing information ...)

    ReplyDelete
  61. Here's yet another way to see the moral of the two box example. Subjects are not responsive to what things afford; they are responsive to surface properties of objects.

    Gibson, 1979, p. 127 suggests (roughly) that we can identify affordances with surfaces. (He talks of constitution, but ...) But, we cannot. That's the problem.

    ReplyDelete
  62. "Subjects are not responsive to what things afford; they are responsive to surface properties of objects."

    This brings to mind another way of looking at this that keeps affordances intuitive and perhaps helps with this dilemma.

    Recall the cause-effect version of "Hume's Law", which goes something like this: nothing at time T can be the cause of anything that happens at time T+e for any e > 0 because the world could disappear in between.

    When I first encountered that, it seemed wrong because it seemed to say essentially that there was no such thing as cause and effect. But after a while I realized that implicit in the concept of cause and effect is the assumption that the world doesn't end. Otherwise, the concept itself has no meaning.

    So what? Well, earlier I raised the issue of unforeseen consequences as seemingly separable from currently perceivable affordances. Also, over at 2 Phils, Andrew had addressed the distinction between immediately actualizable affordances and those that required intermediate steps. That is precisely the situation here. When you look carefully, pick-up-able is not an immediately actualizable affordance - there is an intermediate affordance of touchable that must be actualized first. And for the exploding box, that actualization ends the world from the perceiver's POV. In which case pick-up-able is not only not immediately actualizable, it isn't actualizable at all and is therefore not an affordance

    Here's the pointer to a comment about formalizing the idea of intermediate affordances:

    http://psychsciencenotes.blogspot.com/2010/05/affordances-part-3-dispositions-or.html?showComment=1295238498029#c2185286898682219910

    The approach suggested there is consistent with the informal argument about the inexistence of the pick-up-able affordance for the exploding box that I made in my penultimate comment preceding this one.

    Thus, in my formulation the quote above is close but not quite right in the case of the exploding box. Subjects are responsive to apparent affordances which, however, may not really exist.

    ReplyDelete
  63. As I understand TSRM on anchoring, these are just the underlying physical properties that actually make something have the disposition that it does. So the surface structure of the box that affords picking-up has anchoring properties that make it that way (size, shape, rigidity, maybe even the molecular composition that gives it that rigidity). The disposition is not an accident.

    TSRM then seem to be saying that these properties are what ground the affordance; so it is these that lead to the information. But they also seem happy with the actual intermediate step that Gibson was talking about, namely that these anchoring properties produce a specific type of surface which then interacts with (eg) light to produce information for the affordance. You can't have thingless properties; here, the thing is a surface.

    Net result: I just don't see the disconnect. Gibson simply had the first swing at the explanation, then TSRM developed this more precisely as they worked out the dispositional ontology of affordances. There's no contradiction, just a development of the formal account.

    You suggest that further exploration brings the non-pick-up-ability to light. But, I don't think you want to say this. You don't want to say that affordances are only perceived after what they afford has been manifest.
    It so happens in this case that there is no information that the box cannot be picked up until it is touched. If there was information available prospectively then you could in principle get it prospectively (as for the information specifying that the object could be picked up).

    But, this is not an innocuous concession for EP. Suppose that the person does have the perception that the exploding box affords pick-up-ability. This cannot be because the subject has perceived that the box is actually pick-up-able. (It's not.)
    Being wrong (from some third person perspective) is neither here nor there. You can only ever evaluate a person's behaviour in terms of the information they had available to them at the time. You don't ever get to peer behind the curtain, remember.

    Hence this doesn't work, I don't think:
    This opens the door to the idea that the person perceives the box to be pick-up-able because she has inferred this from the surface properties of the light and her background beliefs. If the Gibsonian or TSRM account does not work in this case.
    The person is happily working from what they think is a complete set of information. They think this because there is no information that they are missing something (until they try to pick up the box). Inference would only be required if the incompleteness was somehow part of experience, which it is not.

    There are certainly some devilish details lurking in here; as Charles notes, you have to ask what affordance was the person actually effecting at the time, etc. We've been proceeding on the assumption that it was picking-up-ness, but maybe I was right to start thinking about prehension instead.

    Thus, in my formulation the quote above is close but not quite right in the case of the exploding box. Subjects are responsive to apparent affordances which, however, may not really exist.
    To be precise, they're responsive to the affordances that are specified in the information they have detected. If the affordance doesn't actually exist, how would it ever project into the perceptual array?

    ReplyDelete
  64. Charles, the worry to have about "apparent affordances" is what they are. Why not just stick with subjects perceive objects by hypothesizing their existence on the basis of light they receive and mental computational routines.

    ReplyDelete
  65. As I understand TSRM on anchoring, these are just the underlying physical properties that actually make something have the disposition that it does. So the surface structure of the box that affords picking-up has anchoring properties that make it that way (size, shape, rigidity, maybe even the molecular composition that gives it that rigidity).

    I'm not that concerned to show that TSRM and Gibson are inconsistent, only that either way your set up the EP approach, it's not going to work.

    I understand what anchoring properties are in the way that you do. But, I don't think surface properties are anchoring properties.

    The argument is simple: The two boxes have the same surface properties, but different affordances, hence different anchoring properties. So, surface properties are not anchoring properties.

    ReplyDelete
  66. Think of the Venn diagram: the boxes overlap in anchoring properties which lead to their surfaces, but differ in anchoring properties which lead to them being fatal or not.

    ReplyDelete
  67. the boxes overlap in anchoring properties which lead to their surfaces

    This is very precisely what I have been challenging. What makes the one box explosive is not what makes the surface look like it does. What makes a box pick-up-able is not what makes its surface the way it is.

    ReplyDelete
  68. There's a lot here, so I'm not sure if this is exactly on point, but hopefully this gets at some of the very interested points everyone has been making.

    In "The Ecological Approach to Visual Perception", Gibson has a section titled "Misinformation for Affordances". He gives two examples: an infant not wanting to crawl over a pane of glass (visual cliff) and someone walking into a glass door. He describes these as cases where support and a barrier were mistaken for air because "the optic array specified air". In another example, he writes of an animal falling into a covered pitfall or a person being shocked by a radio cabinet. All of these are "misperceived" affordances: the information in the optic array made it seem like the ground afforded support, but it did not. He writes, "a thing may not look like what it is".

    Followed by, "Nevertheless, however true all this may be, the basic affordances of the environment are perceivable and are usually perceivable directly, without an excessive amount of learning. The basic properties of the environment that make an affordance are specified in the structure of ambient light, and hence the afforadnce itself is specified in ambient light." (p.142-143)

    So it seems like what he would say about the box example is something like this: both boxes _appear_ to afford picking up (that's what the optic array gives us); however, this is only actually true of the first box. We are misinformed into thinking that the second box affords picking up. Only through learning can one determine that the second box is not actually pick-up-able. I imagine that he thinks of it as a different form of visual illusion. This isn't really at odds with anything in his theory, though, and therefore I don't find it to be a problem, although it is most certainly hand-wavy.

    This is actually my least favorite section of the book since he gives up so much here (i.e., "_usually_ directly perceivable directly"), but I'm afraid he would be then be stuck with exactly the sort of problem you bring up.

    ReplyDelete
  69. Thank you genna!

    1. misperceived = apparent = conditional = !basic

    I rest my case. So, can we pick one and move on? I vote for "conditional".

    2. However, I'm now skeptical of distinguishing the conditional affordances offered by the two boxes. The omniscient wizard Ken Aizawa has warned about the dangers of trying to pick up one box, but I don't recall his revealing details about the other box. So, how do we mere perceivers w/o magical insights know that the first is pick-up-able, ie, offers "basic" affordances? It may be full of mercury, nailed to the floor, etc. If there is no perceivable distinction, they presumably offer a perceiver the same conditional affordances.

    ReplyDelete
  70. Hi, Genny,

    Thanks for stopping by. Thanks also for citing some of the relevant texts. I know that I have not read it all and have remembered even less.

    I think that you are on to the right issues, though the discussion has moved on a bit in the past few weeks.

    Now, it may be true that Gibson will want to say that his theory works for "basic affordances", but I argue that it does not. The problem is that the mechanism for visual perception of affordances sketched on p. 127 does not work. The layout and composition of surfaces does not typically constitute an affordance.

    But, let me know how you see this with fresh eyes.

    ReplyDelete
  71. So, in the exploding boxes cases (and the others), Gibson might say that one has information and the other misinformation. But, this is evidently incorrect. The two boxes are physically identical on the outside, so they structure light in exactly the same way, hence provide exactly the same information for the observer.

    ReplyDelete
  72. Charles,
    I don't see why you think there is an epistemological puzzle concerning the configuration of the boxes. For philosophers, this would be just a standard stipulation of the conditions of a scenario. For psychologists, this would just be a sketch for an experimental protocol. In both cases, we know what's going on in the set up, because we did the set up.

    ReplyDelete
  73. Sorry for the delay in response: lots to write about. OK, so I was able to look a bit ahead and catch up with what people were saying. I'm unfamiliar with the Chemero work, so I can't speak to that and a lot of the discussion has progressed elsewhere. Instead, I'd like to take a quick jab at Noe and then expand more on exploding boxes. I'm breaking this up into two parts since it's so long.

    PART I

    Re actionism:
    Here is what Noe might have in mind: "Moving from place to place is supposed to be 'physical' whereas perceiving is supposed to be 'mental,' but this dichotomy is misleading. Locomotion is guided by visual perception. Not only does it depend on perception but perception depends on locomotion inasmuch as a moving point of observation is necessary for any adequate acquaintance with the environment. So we must perceive in order to move, but we must also move in order to perceive." (Gibson 1979, p. 223)

    I personally think that Noe reads too much into this and similar sections. I haven't taken a close look at his most recent book, so I'm not sure how much his views have changed since Action and Perception, but he had a very particular reading of Gibson there. First of all, we should throw out everything about microsaccades etc. - the fact that the eyes have to constantly be moving or else the retinal image would fade to gray (is this when he brings up the notion of a ganzfeld?) because of retinal pigment bleaching / habituation. This is just an artifact of our visual system and has nothing to do with movement proper. Likewise we should ignore that whole section about how visual stimulation is needed for cortical development - this has absolutely nothing to do with what Gibson is talking about.

    I think what Noe really wants to get at some deep points about perception like the nature of perceptual constancy or something. For example, given that our retinal image of the top of a mug is almost always (except under special viewing conditions) elliptical, how is it that we are able to consistently experience the top of the mug as a rigid (non-deforming) circle? But this is NOT what Gibson is talking about.

    Gibson is saying (in the above-quoted and immediately preceding sections) that, for example, sequences of occlusions and disocclusions of surfaces in the optic array (either from self-motion or object-motion) can inform us about the structure of the world in a way that a static, monocular image cannot. This interpretation makes perfect sense in a historical context (history of psychology/psychophysics) - the study of perception had been operating with (and continues to, to a large extent) the framework that the brain gets poor, distorted, low-quality signal from the eyes/world ("booming, buzzing confusion") and has to "build up" the world from this paucity of information. Gibson's revolutionary point is that this is all hogwash - all the information you need to perceive the world veridically is out there, you just have to get it and you do so by moving. For example, different 3D shapes can project to the same retinal image. You can resolve the ambiguity in two ways: by having two eyes and stereopsis, or simply by seeing the object from a different viewpoint. The quote above is merely saying that the function of seeing is to enable us to locomote around obstacles and get our hands on food and shelter etc. and that in order to recover 3D structure of objects accurately, you have to move around them (or have them move around you). But there is nothing inherently contradictory in this; problems only arise if you take Noe's peculiar reading about what Gibson is trying to say here.

    ReplyDelete
  74. PART II

    Now on to exploding boxes.

    Ken, I can imagine several objections that you might have, and I'm not quite sure which one you have in mind, so I'll try to address the ones that are the most obvious to me. I think you are trying to say something akin to (2), let me know if I'm misinterpreting your point.

    1) There is not enough information in the optic array to determine which box (exploding or not) affords picking up, since both are inaccurately perceived as being pick-up-able from their surface properties.

    Response: I agree! You can't distinguish the pick-up-able from the not-pick-up-able! BUT that's not a problem. The reason that we are misinformed is because we've had sufficient experience with pick-up-able objects to expect all similar objects (surface layouts) to be pick-up-able. (Think of being fooled by quarters superglued to the floor.) That is, we expect it to be pick-up-able. A much trickier question is where do these expectations come from. Unfortunately, Gibson does not provide a very satisfying account of perceptual learning in this text (and, tellingly, neither does Noe). But, if we take it for granted that the observer is the kind of person that can perceive pick-up-ability (as I think Gibson does), then both box-structures would appear to afford pick-up-ability.

    1.1) Let's imagine that we live on a booby-trap-world designed by an evil mastermind where any object can potentially explode upon being touched and there is no way to visually discriminate between booby-trapped objects and non-booby-trapped objects. How can we ever know by visual information alone whether something affords picking-up or not? What saves us from these worries in our own world? How can we ever learn anything?

    Response: I'm not too confident in this response, but I'd venture that people in booby-trap-world fail to learn and therefore fail to perceive the affordance. This analogy is a little imperfect, but maybe it'll work: some phones are smart-phones and have internet access and some do not. Which phones afford "internet-connectivity"? We used to live in a world/environment where no phones did; possibly some time in the future all phones will; currently, some phones do and some phones don't and many people (or at least I) cannot determine by visual means alone which phones do and which ones do not. Or maybe think of iPads - some are 3G enabled and some are not. There is no way to determine from the structure of the object whether it affords internet-connectivity or not. I realize this isn't a great example because internet-connectivity isn't a basic affordance and therefore has its own slew of problems, but I think the argument is ok. Or imagine a future world where holograms are a lot more pervasive than they are now; we might not be as sure of grasp-ability.

    ReplyDelete
  75. PART II continued


    This, I think is what you are actually arguing:
    2) Surfaces, in general, are not sufficient to specify affordances, not just in these tricky cases, but always, like your frozen-pond example in a later post.

    Response: If this is your claim, then I'm afraid that the problem comes from taking p.127 out of context with respect to the rest of the book. Chapter 2, for example, outlines kinds information that can assist in determining and distinguishing surfaces, including properties like texture, composition, cohesion, reflectance and viscosity (as well as "extra-surface" (my term) properties, like illumination). All of these constitute the surface. This helps us with seeing the surface as ice. Whether ice affords support or not is a matter of experience. Whether this particular pond affords support is a matter of familiarity with the environment - given my weight, size and choice of footgear and knowledge of time of year, recent weather, and whether this pond often freezes over or not, I could probably determine, just by looking at the pond, whether it affords support or not. Basically, I think there is a lot more information in the structure and nature of the environment than you give credit to. The visual system has evolved to be accurate as often as possible. Given our environment, it is sensible to believe that we accurately perceive affordances such as sit-on-able because most things that appear sit-on-able actually are. However, just like there can sometimes be visual illusions that trick our visual systems, there can likewise be "trick-objects" that trick the visual system in exactly the same way.

    Sorry for the long post! I hope it makes some sense.

    ReplyDelete
  76. Genna,
    There is a lot in your comment on which I would like to comment in return, but let me cut to what I take to be the nub of the issue. This is your interpretation 2). This is precisely what I am pursuing. Still, I don't think your reply works.

    So, just to focus our attention precisely, here is the claims:
    2) Surfaces, in general, are not sufficient to specify affordances,

    (Now, technically, I think the Gibsonian jargon is that light specifies affordances, which will probably raise Andrew's hackles, but what doesn't? I hear he hates puppies!)

    For the purposes of this argument, the next part I can entirely agree with:
    Response: If this is your claim, then I'm afraid that the problem comes from taking p.127 out of context with respect to the rest of the book. Chapter 2, for example, outlines kinds information that can assist in determining and distinguishing surfaces, including properties like texture, composition, cohesion, reflectance and viscosity (as well as "extra-surface" (my term) properties, like illumination). All of these constitute the surface. This helps us with seeing the surface as ice.

    So, I can agree, for the sake of argument, that there are all these properties that constitute the surface. And I can agree, for the sake of argument, that these help us with seeing the surface of the ice.

    But, here you are simply wrong:
    Whether ice affords support or not is a matter of experience.
    Whether ice affords support or not is a matter of such things as how much weight it will bear, etc. Your knowing or believing that the ice affords support is a matter of experience. So, that's my first rejoinder.

    But, my second is to have you look back at the claim 2). It is a claim about surfaces specifying affordances. Or, better, whether the composition and layout of surfaces constitutes affordances, or whether light specifies affordances. My point is that the composition and layout of surfaces does not constitute affordances; light does not specify affordances. It does not. Here I am challenging the account on Gibson, 1979, p. 127. So, really, the collapse of the Gibsonian account occurs before an organism is struck in the eye by light. The problem is not one of experience. (But, Andrew, Gary, and you have all hit on this. So, that's a consensus reply. It just doesn't work.)

    ReplyDelete
  77. Re: "conditional affordances", I think if one reads through Gibson, 1979, Chapter 8, this is not anything that Gibson takes seriously. It's just a fleeting comment for which he builds no theory. Genny is right to draw attention to this penultimate section of Chapter 8, but my take on it is that Gibson fails to really come to grips with the problem. (More of which in future posts.) Genny, however, seems to think that Gibson concedes too much. But, maybe I am reading G incorrectly.

    ReplyDelete
  78. Re: Noe and actionism, it seems to me that he might well be drawing on more than just Gibson for his actionism. I think that there are roots in the Phenomenological tradition as well, although I cannot speak to that with any degree of authority. Gary Williams can probably comment in more detail.

    Re: Noe and saccades, I agree that he misinterprets the significance of retinal stabilization experiments. I noted this in my "Understanding the Embodiment of Percepton", J. Phil., January, 2007. But, (in a paper I keep plugging) Nivedita Gangopadhyay has offered a scientifically and philosophically detailed reply to this.

    ReplyDelete
  79. But, here you are simply wrong:
    Whether ice affords support or not is a matter of experience.
    Whether ice affords support or not is a matter of such things as how much weight it will bear, etc. Your knowing or believing that the ice affords support is a matter of experience. So, that's my first rejoinder.


    You are right, I miswrote.



    OK, now that I understand your argument I'd be happy to try to argue Gibson's side. I'll wait until there's a new place to post.

    ReplyDelete
  80. Ok, Genna. I think that tomorrow morning I'll post something that will give you an entry into the debate. It's about some of the texts that you have cited.

    ReplyDelete
  81. Genna, I think learning is one of the places to go looking for the solution to Ken's problem too. I don't think we're quite learning what you suggested; experience with shoes, etc sounds like the kind of enrichment of experience with "knowledge" that the Gibsons were working on avoiding (EJ, JJ's wife, was the perceptual learning part of the team, btw).

    But our extensive experience with surfaces is indeed an important part of my reply to Ken.

    I don't know why people are so down on the constant motion of the eyes as being meaningful. Of course it's to preserve access to information - this isn't, I don't think, that controversial.

    ReplyDelete
  82. But, how is what is, or is not, going on in your head going to affect the ability of the surface of a piece of ice, or a chair, or a box, to structure light?

    ReplyDelete
  83. Andrew:
    Thanks for the link to your post. I'll take a look.
    On my reading, if I recall correctly, Noe takes microsaccades as giving access to information in a similar way that moving around an object does. This is incorrect.

    Ken:
    I'll be happy to discuss this more at length tomorrow. My main argument will center on the nature and role of invariants. Please include this quote in your post, if possible. (Everywhere, original italics.)


    p.140-141
    "The central question for the theory of affordances is not whether thy exist and are real but whether information is available in ambient light for perceiving them. The skeptic may now be convinced that there is information in light for some properties of a surface but not for such a property as being good to eat. The taste of a thing, he will say, is not specified by light; you can see its form and color and texture but not its palatability; you have to taste it for that. The skeptic understands the stimulus variables that specify the dimensions of visual sensation; he knows from psychophysics that brightness corresponds to the intensity and color to wavelength of light, He may concede the invariants of structured stimulation that specify surfaces and how they are laid out and what they are made of. But he may boggle at invariant combinations of invariants that specify the affordances of the environment for the observer. The skeptic familiar with the experimental control of stimulus variables has enough trouble understanding invariant variables I have been proposing without being asked to accept invariants of invariants.

    Nevertheless, a unique combination of invariants, a compound invariant, is just another invariant. It is a unit, and the components do not have to be combined or associated. Only if percepts were combinations of sensations would they have to be associated. Even in the classical terminology, it could be argued that when a number of stimuli are completely covariant, when they always go together, they constitute a single "stimulus." If the visual system is capable of extracting invariants from a changing optic array, there is no reason why it should not extract invariants that seem to us highly complex.

    The trouble with the assumption that high-order optical invariants specify high-order affordances is that experiments, accustomed to working in the laboratory with low-order stimulus variables, cannot think of a way to measure them. How can they hoe to isolate and control an invariant of optical structure so as to apply it to an observer if they cannot quantify it? The answer comes in two parts, I think. First, they should not hope to apply an invariant to an observer, only to make it available, for it is not a stimulus. And, second, they do not have to quantify the invariant, to apply numbers to it, but only to give it an exact mathematical description so that other experimenters can make it available to their observers. The virtue of psychophysical experiment is simply that it is disciplined, not that it related the psychical to the physical by a metric formula."

    ReplyDelete
  84. Hi, Genny,

    I was going to begin with the first half of the first paragraph, so why don't we begin with just this tomorrow? (My main point, after all, has been to challenge, essentially, the part up to taste for that.)

    And, I'll relate how the exploding box type cases handle invariants and higher order invariants. Basically, postulate all the invariants and higher order invariants you want. The problem comes in trying to make the case that the two boxes, or two chairs, or two sheets of ice, or whatever, that are physically identical on the outside are going to give rise to different higher order invariants.

    ReplyDelete
  85. But, how is what is, or is not, going on in your head going to affect the ability of the surface of a piece of ice, or a chair, or a box, to structure light?
    What I said was, it helps preserve access to that information (and yes, I know you think there isn't any).

    On my reading, if I recall correctly, Noe takes microsaccades as giving access to information in a similar way that moving around an object does. This is incorrect.
    It's a little right. I'm not sure about microsaccades per se, but I do know eye movements are more than sufficient motion for access to information; you can put someone in a bite board and all kinds of stable perception is preserved.

    ReplyDelete
  86. Ok. Genny, the post is loaded and ready for action tomorrow. Unfortunately, tomorrow is a heavy teaching day for me.

    "But, how is what is, or is not, going on in your head going to affect the ability of the surface of a piece of ice, or a chair, or a box, to structure light?
    What I said was, it helps preserve access to that information (and yes, I know you think there isn't any). "

    This, Andrew, is a very nice example of what I mean when I complain that you talk around an issue. What I am challenging is Gibson's assumption about the capacity of surfaces to structure light, but then you want to talk about something else, in this case, the preservation of access to information.

    ReplyDelete
  87. Ken:
    Sounds good!

    Did you mean "... the capacity of surfaces to structure light in such a way that specifies affordances"?

    Here are two more quotes that you might find interesting.

    p.127:
    "If a terrestrial surface is nearly horizontal (instead of slanted), nearly flat (instead of convex or concave), and sufficiently extended (relative to the size of the animal) and if its substance is rigid (relative to the weight of the animal), then the surface affords support."

    p.128
    "But if a surface is horizontal, flat, extended, rigid, and knee-high relative to the perceiver, in can in fact be sat upon. If it can be discriminated as having just these properties, it should look sit-on-able. If it does, the affordance is perceived visually."



    Andrew:
    I'm on board with eye movement in general - how we look at things is incredibly important for what we perceive. I just quibble with microsaccades in particular.

    ReplyDelete
  88. This, Andrew, is a very nice example of what I mean when I complain that you talk around an issue. What I am challenging is Gibson's assumption about the capacity of surfaces to structure light, but then you want to talk about something else, in this case, the preservation of access to information.
    We were talking about eye movements. Well, I was, I thought you were too. I think there is information, and so I think an eye in constant motion is a good idea to maintain access to that information. That's all I was saying here.

    ReplyDelete
  89. Did you mean "... the capacity of surfaces to structure light in such a way that specifies affordances"?

    I'm not sure where I said otherwise than this, but this does seem to be what I want to say.

    ReplyDelete
  90. I know this thread advanced elsewhere, and stopped over a year ago... but...

    I think it would be impossible to read Gibson as saying that all perception is the fully correct perception of affordances. There are other ecological theorists you might be able to read this way, but not Gibson.

    At any rate, in Campbell's quote at the start, it is clear that he sees the the things-on-the-workbench as affording many things, even though he is uncertain about the key affordance that might justify it having a place on the bench. He does not seem to doubt that it is touchable, describable, etc. Thus Campbell's setup of "I perceive no affordance" seems vulnerable to attack in exactly the same manner as Descartes's "I doubt everything".

    There are bigger issues here, in how to set up ecological theory. As you no doubt know, ecological psychologist are not unified on many of these issues . The Connecticut school seems to want to 'perception = perception of affordances' to be true by definition. But that does not seem in line with Gibson. If nothing else, Gibson had well-reasoned arguments why we would expect to find perception of affordances - it was (at the time) an open empirical question whether evidence could be found that people are particularly sensitive to affordances. If Gibson saw it as true by definition, he would not have seen an empirical question.

    ReplyDelete
  91. Hi, Eric,

    Could you provide some references to places where Gibson writes of "fully correct perception"?

    Thanks as well for the link to "Fixing Psychology". I've added it to my blog reading list.

    ReplyDelete