Friday, February 11, 2011

Gibson on Information and Misinformation 1

From Gibson, 1979, Chapter 8:
It is because the affordances of things for an observer are specified in stimulus information. They seem to be perceived directly because they are perceived directly.

The central question for the theory of affordances is not whether they 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 in light; you can see its form and color and texture but not its palatability; you have to taste it for that.  (Gibson, 1979, p. 140)
MISINFORMATION FOR AFFORDANCES
If there is information in the ambient light for the affordances of things, can there also be misinformation? According to the theory being developed, if information is picked up perception results; if misinformation is picked up misperception results.
The brink of a cliff affords falling off; it is in fact dangerous and it looks dangerous to us. It seems to look dangerous to many other terrestrial animals besides ourselves including infant animals. Experimental studies have been made of this fact. If a sturdy sheet of plate glass is extended out over the edge it no longer affords falling and in fact is not dangerous, but it may still look dangerous. The optical information to specify depth-downward-at-an-edge is still present in the ambient light; for this reason the device was called a visual cliff by E. J. Gibson and R. D. Walk (1960). Haptic information was available to specify an adequate surface of support, but this was contradictory to the optical information. When human infants at the crawling stage of locomotion were tested with this apparatus, many of them would pat the glass with their hands but would not venture out on the surface. The babies misperceived the affordance of a transparent surface for support, and this result is not surprising.
Similarly, an adult can misperceive the affordance of a sheet of glass by mistaking a closed glass door for an open doorway and attempting to walk through it. He then crashes into the barrier and is injured. The affordance of collision was not specified by the outflow of optical texture in the array, or it was insufficiently specified. He mistook glass for air. The occluding edges of the doorway were specified and the empty visual solid angle opened up symmetrically in the normal manner as he approached, so his behavior was properly controlled, but the imminence of collision was not noticed. A little dirt on the surface, or highlights, would have saved him.
These two cases are instructive. In the first a surface of support was mistaken for air because the optic array specified air. In the second case a barrier was mistaken for air for the same reason. Air downward affords falling and is dangerous. Air forward affords passage and is safe. The mistaken perceptions led to inappropriate actions.  (Gibson, 1979, p. 142)
In comments on a post from weeks ago, Gennady Erlikmann draws our attention to some important texts from the latter portions of Chapter 8 of Gibson, 1979.   Now, it seems to me that the passages from p. 140 cited above champion the view that affordances structure light in such a way as to enable the affordances to be perceived.  And he does not there back off the view.  But, then on p. 142, he apparently does back off of this apparently admitting that sometimes affordances do not structure light in such a way as to enable them to be perceived.  (That's the italized part in paragraph 3 from p. 142.)  In the final paragraph, the two italicized sections suggest that the glass in the two cases did not structure light in such a way as to specify an affordance of, say, "stand-on-ability" or "walk-through-ability".  Instead, the glass apparently specified something else, namely, air.  Kind of grudging admissions.
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 affordance itself is specified in ambient light. Moreover, an invariant variable that is commensurate with the body of the observer himself is more easily picked up than one not commensurate with his body.  (Gibson, 1979, p. 143).
In the final paragraph here, Gibson suggests that, while maaaybe not all affordances are perceived, at least all the basic affordances are perceived. 

Now, I've been trying to give examples showing that this kind of analysis will not work.  The box and the exploding box are physically the same on the outside, so structure light in the same way.  There is no room to say that one of these structurings is information, where the other is misinformation.  As light structure, they are the same.  At most, calling one "information" and the other "misinformation" is just to rename the one light structure as coming from the normal box, where the other light structure as coming from the exploding box.

49 comments:

  1. Laying it all out like this suggests to me the problem, a matter of simple logic. Here is what Gibson says:

    affordances of things for an observer are specified in stimulus information

    That is, A -> S. But that does not logically imply that in the absence of an affordance, stimulus information can't be present that causes an observer to misperceive the affordance. That is to say, ~A -/> ~S. (That's the "looking" vs "being" distinction prominent in Sellars' EPM, with which I suspect Gibson was familiar.)

    So, only scenarios that involve affordances that are present but not perceived, are relevant. And that's what Gibson addresses: misperceived affordances - affordances present but not perceived; that is, {A & ~S}. Scenarios in which non-existent affordances are perceived as present - that is, {~A & S} are irrelevant.

    So, let's work through Ken's scenarios starting with the easiest - scenarios in which something explodes, collapses, breaks, is poisoned, et al. These are all examples perceiving non-existent affordances.

    Walking through a glass door as described by Gibson is also in that category. One could interpret the affordance as walk-into-able, but Gibson describes it as actually be-injured-by-able. Perhaps an affordance for masochists, but not for the normal animal.

    Finally, the baby sees something that overrides the crawl-on-able affordance of the glass. But since affordances are actor-relative, that can be viewed as suggesting that the baby doesn't perceive crawl-on-able as being an affordance. So, even this is arguably not a good example.

    This may be quibbling, but Ken has made quite an issue of Gibson's exact words. This appears to be - subject to error on my part - the logical consequence of those words.

    ReplyDelete
  2. Charles' concern would be a problem if Gibson were just asserting a material conditional. But, I think he is making a causal claim (not necessarily here) when he claims (in general, mistakenly) that affordances structure light. For a causal claim, Mill's method of difference is a reasonable thing to try to apply.

    ReplyDelete
  3. Ken, thanks for the summary of your view and reposting that quote. Also, thank you Andrew for the in-depth response on your blog. Sorry for the delay in my response - the end of this week has been unexpectedly hectic. That final paragraph was exactly what I was going to point to! Great comparison with the other sections. Unfortunately, my response is again quite long, so I'm going to break it up into a few sections.

    Gibson is a little loose with his wording throughout the book. As a result, he seems to be contradicting himself a fair number of times. For example, he writes things like "The basic properties of the environment that make an affordance are specified in the structure of ambient light, and hence the affordance itself is specified in ambient light" (p.143) while at the same time arguing that affordances are not physical; rather, they are not properties of the environment per se but are both psychical and physical simultaneously or neither.

    I'd like to see if I can convince you of either one of these by looking at cases we've discussed:
    (1) Weak notion: Most of the time, surfaces constitute/specify/structure affordances.
    (2) Strong notion: Surfaces constitute/specify/structure affordances all of the time.

    ReplyDelete
  4. Glass door:

    Does glass (of an appropriate size) afford collision (for humans)? Yes, it's a rigid surface. Does the information in the optic array specify that a region of the environment affords collision? Sometimes - if there's a smudge on the glass or you catch a reflection or glare. Other times the optic array specifies air. This is the gist of the misperception section and his section on perceiving glass (p.152-153). It also doesn't help that Gibson isn't very clear on this point. Take a look at this quote about glass floors:
    "The animal or child can be put down on this [glass] surface under two conditions: when it is visible, by virtue of texture paper just under the glass, and when it is invisible, with the paper placed far below the glass. The glass affords support under both conditions but provides optical information for support only under the first. There is mechanical contact with feet in both cases but optical information for contact with feet only in the first." (p.156-157) Here he seems to be suggesting that even if the glass floor doesn't structure the optic array, it still affords support. I don't think we necessarily need to accept that reading; instead, we need to remind ourselves about what it means for something to be specified in the optic array.

    Suppose I see a bucket from the bottom. Even though the bucket affords containing liquid, the information from the optic array from this particular viewpoint fails to specify that information - it doesn't structure light in the appropriate way. However, a main part of Gibson's argument (and I think Andrew was hinting at this too) is that we don't have what Noe calls as a Machian view of the world - perception is not the analysis of a series of static frames. When we think of optic array, it is important to remember that this doesn't mean "the information flowing past you at time t." Gibson writes, "[...] the structure of an optic array at a stationary point of observation is only a special case of the structure of an optic array at a moving point of observation." (p.72) and "Optical information [is] the information that can be extracted from a flowing optic array" (p.63). We shouldn't think of affordances as specified by surfaces that structure the optic array unambiguously at time t, but, rather, as specified by the structure revealed by optic flow. Returning to the glass door: as I move around the glass, depending on the nature of the illuminant, there are certain angles from which light will be reflected in a way that would specify a collidable surface. The surface affords collision and this structures optic flow - the optic array of an active observer (here I'm being a little loose with terms, but I think it's clear what I mean). It is only in special cases and from special viewpoints that light passes perfectly through the glass so as to make it imperceptible. So maybe (2) is ok.

    ReplyDelete
  5. Ice pond:

    I'm going to side with Andrew on this one. I would contend that the structure of the substance of the ice is specified in the optic array and that this determines whether something is walk-on-able or not (for a person A in environment Y under conditions Z). This is an empirical question. To test this, we would need to construct an iced-over pond that affords walking on for person A and then see if we can change it's "internal" properties to make it not walk-on-able for the same person. I would contend that this is impossible. Gibson hints at this as well in his section on "The Niches of the Environment."

    Ultimately, I take issue with Ken's statement in an earlier post, What structures light is typically on the outside of objects; what constitutes affordances are typically on the inside of objects. I would argue that almost all of the time, surfaces tell us everything we need to know because that is all there is to know. Gibson writes, "The physical, chemical, meteorological, and geological conditions of the surface of the earth and the pre-existence of plant life are what make animal life possible. They had to be invariant for animals to evolve." (p. 128). I think this is his argument against Ken's point - if how we could interact with objects was really determined by some magical, hidden properties (if every object could be a proximity mine), then we would never have survived. The human race would have rapidly vanished in exploding-object world. The fact of the matter is that virtually all of our interactions with the environment are made possible by the fact that the properties of objects are reflected in their surfaces. Perhaps the evolution of trichromatic vision in humans is what helps us tell apart ripe from unripe fruit - even "internal" properties of objects are often reflected on their surfaces. These are the invariants that we pick up. Sure, on a rare occasion this might not be the case (trick chairs, exploding boxes, pit-traps, quicksand), but most of the time it is.




    I'll post on trick chairs and exploding boxes in a little bit.

    ReplyDelete
  6. I'd like to see if I can convince you of either one of these by looking at cases we've discussed:
    (1) Weak notion: Most of the time, surfaces constitute/specify/structure affordances.
    (2) Strong notion: Surfaces constitute/specify/structure affordances all of the time.


    Nope. I reject both of these. I think that the composition and layout of surfaces rarely constitute affordances.

    ReplyDelete
  7. Does glass (of an appropriate size) afford collision (for humans)? Yes, it's a rigid surface. Does the information in the optic array specify that a region of the environment affords collision? Sometimes - if there's a smudge on the glass or you catch a reflection or glare.

    Ok. But, here you are missing the point. So, we agree the glass affords collision. Does this affordance structure light? You say sometimes, suggesting that it does when there is a smudge. But, this is an incorrect analysis. It's not the smudge that is the impediment to passage; it's the glass. So, you don't have case in which it is the affordance structuring the light. It's the smudge, which is the surface, structuring the light and the surface does not constitute the affordance. This is the basic problem.

    ReplyDelete
  8. Suppose I see a bucket from the bottom.

    Sure. Throw in moving around. It's not going to help you with the exploding boxes. It's not going to help you with the stool. It's not going to help you with the ice.

    ReplyDelete
  9. I'm going to side with Andrew on this one. I would contend that the structure of the substance of the ice is specified in the optic array and that this determines whether something is walk-on-able or not (for a person A in environment Y under conditions Z). This is an empirical question. To test this, we would need to construct an iced-over pond that affords walking on for person A and then see if we can change it's "internal" properties to make it not walk-on-able for the same person. I would contend that this is impossible. Gibson hints at this as well in his section on "The Niches of the Environment."

    Ok. I would have thought that everyone agrees that humans fall through the ice on ponds are rivers because the surface does not structure light in such a way as to specify its weight bearing capacity. But, ok. Switch examples, yet again. Take two sheets of plexiglass. One thick; one thin. Spray paint them black. One will afford walk-on-ability, the other will not. The structure of the light from their surfaces will not indicate their weight bearing capacity.

    ReplyDelete
  10. Trick chairs:

    Unlike the pond example, it is realistic to expect to be able to build two chairs that look exactly but have different substance structures. Gibson does consider some cases on p.137 where he talks about some food affording nutrition and other food poison or one metallic object affording grasping and the other electric shock. Unfortunately, he doesn't explain what happens in these situations. Perhaps he would say, as earlier, that optical information is not enough to determine whether the chair is sit-on-able or not. You might be able to avoid this with higher-order invariants.

    The bucket example is informative in this respect because it highlights another point of confusion: affordances are not properties of objects! To see the bucket from the bottom and have it still afford containing liquid, it would have to be suspended overhead; it would not be enough for me to see a bucket, upside-down, on the ground. The object would not afford containing liquid. Gibson: "An affordance is not bestowed upon an object by a need of an observer and his act of perceiving it. The object offers what it does because it is what it is. To be sure, we define what it is in terms of ecological physics instead of physical physics" (p.139) and also "But we must, of course, learn to see what things really are - for example, that the innocent-looking leaf is really a nettle [...] and this can be very difficult." (p.142)

    One possible response: Ken's argument relies on restricting the visual world to the chair. But recall that objects do not afford! And they do not exist in a vacuum. Surfaces apart from those of the chair structure the optic array. Chairs located in prank-shops might be trick-chairs more often than chairs in dining rooms (see above about niches). This kind of information might be captured in whatever Gibson might mean by higher-order invariants. Indeed, I think that's what he's suggesting when he writes about palatability and invariant combinations of invariants. However, I have a really hard time talking about what exactly he means here without referring to events, experience or other modalities and I was trying to stay away from that. In that sense, higher-order invariants might not be the best response.

    ReplyDelete
  11. Exploding boxes:

    Unlike trick chairs, I have a hard time coming up with a context for exploding boxes. Think about "suspicious packages" - all of them look pick-up-able, but we can only determine whether that's the case if we have a bomb-sniffing dog or a robot that we make pick it up. I'm reminded of Andrew's comment on my Zippo example: a fuel-filled Zippo and an empty Zippo structure light in the same way even though one affords lighting and the other one doesn't. You could always find out what it affords by trying to light it, just like you would try to write if you wanted to know if your pen had any ink in it or not. But I agree with Ken that in these cases the surfaces and structure of light alone do not uniquely specify one or the other affordance (without additional information, like the weight of the lighter).

    One alternative response, then, is that we (humans) treat man-made objects as if they were natural objects with respect to the affordances that we extract from the optic array even though their affordances are not uniquely specified by their surfaces. It's important to recall that we're actually talking about trick-chairs and exploding boxes and that sit-on-able and pick-up-able are supposed to be a basic affordances unlike lighting or writing, so the distinction cannot be made along basic/complex grounds. So we would go with something like (1) : most of the time, affordances are relayed through the way light is structured by surfaces. This is especially true of the natural environment. Things get more complicated with man-made objects. We treat them as if invariants about their surfaces can tell us about how we can interact with them. But this isn't actually the case.

    ReplyDelete
  12. I think this is his argument against Ken's point - if how we could interact with objects was really determined by some magical, hidden properties (if every object could be a proximity mine), then we would never have survived. The human race would have rapidly vanished in exploding-object world. The fact of the matter is that virtually all of our interactions with the environment are made possible by the fact that the properties of objects are reflected in their surfaces.

    Of course, the surfaces of objects are often good guides to what they afford. That's common ground between EPists and cognitivists. The difference, however, lies in the accounts EPists and cognitivists provide of why surfaces are often good guides to affordances. Gibson maintains that surfaces are good guides to affordances because basically surfaces are affordances. That's nuts. The cognitivist answer is that surfaces provide evidence about what's inside (that there are correlations between surface properties and affordances (and it's not magic)) and that humans can typically make the inference from the surface properties, such as colors, to affordances. According to cognitivists, a human might perceive that an object is red and spherical, then infer that is is an edible ripe apple. The EPist says that one just perceives edibility.

    How do you test what a human perceives? Do an experiment. Hold the surface properties constant and vary the affordances, then see how humans respond. This is no trick.

    ReplyDelete
  13. Ken, I agree with most of your points. I think higher-order affordances are totally bonkers. For example, that we can perceive mail boxes as "postable" - as you saw, I have a hard time arguing that view.

    But, I think we can salvage some notion of basic affordances in cases - particularly with respect to natural environments.

    ReplyDelete
  14. Quick point:
    Ok. I would have thought that everyone agrees that humans fall through the ice on ponds are rivers because the surface does not structure light in such a way as to specify its weight bearing capacity.

    I'd argue that individuals just don't know what the relevant information is to take the appropriate action. Perceptual learning helps here.

    ReplyDelete
  15. How do you test what a human perceives? Do an experiment. Hold the surface properties constant and vary the affordances, then see how humans respond. This is no trick.
    Actually no, because all you're doing is building irrelevant, unecological, equivalent configurations. This has been the experimental problem all along.

    The experiment to see what human perceive is to perturb the candidate information variables until you find the one that breaks the ability to use the affordance or perceive the event.

    ReplyDelete
  16. But, Andrew, by Gibsonian light the exploding boxes, the thick and thin ice, the strong chair and the weak chair are not equivalent configurations! Gibson's claim is that the affordances structure light, so that differences in affordances should lead to differences in the structure of light.

    The problem is that he's just wrong about what configurations are equivalent.

    ReplyDelete
  17. Ok. I would have thought that everyone agrees that humans fall through the ice on ponds are rivers because the surface does not structure light in such a way as to specify its weight bearing capacity.

    I'd argue that individuals just don't know what the relevant information is to take the appropriate action. Perceptual learning helps here.


    But, Genny, you're missing the point. This is about ecological physics, if you will. The core is that the composition and layout of surfaces does not typically constitute an affordance. Whether or not an object will bear your weight (hence whether it affords sit-on-ability, climb-up-ability, walk-on-ability) depends on more than just the composition and layout of the surfaces. It depends on the internal structure of the thing, the structure that is not exposed to light.

    ReplyDelete
  18. But, Andrew, by Gibsonian light the exploding boxes, the thick and thin ice, the strong chair and the weak chair are not equivalent configurations! Gibson's claim is that the affordances structure light, so that differences in affordances should lead to differences in the structure of light.
    I think you're vastly overestimating the odds of two things that are actually different coming to produce the same information. Note the chairs and the boxes must be carefully constructed; in general, two things that aren't exactly identical will not produce exactly identical information. Runeson's point was that just because you can carefully construct an equivalent configuration, doesn't mean it has any bearing on an evolved perceptual system's abilities.

    ReplyDelete
  19. But, Andrew, you're missing the point. This is about ecological physics, if you will. It's not about evolved perceptual system abilities.

    Contrary to what Gibson suggests, the composition and layout of surfaces does not typically constitute an affordance. Whether or not an object will bear your weight (hence whether it affords sit-on-ability, climb-up-ability, walk-on-ability) depends on more than just the composition and layout of the surfaces. It depends on the internal structure of the thing, the structure that is not exposed to light.

    This has nothing to do with how hard it is to make a chair or a floor or a fake room.

    ReplyDelete
  20. Your strategy in defending Gibson here is clear, Andrew. Controlled experiments are a cheat, since they are not ecologically valid. But, "naturally occurring experiments" are inconclusive, since the natural world rarely if ever produces exactly identical things. So, you can't lay a glove on Gibson's hypothesis that the composition and layout of surfaces typically constitutes affordances.

    ReplyDelete
  21. Your strategy in defending Gibson here is clear, Andrew. Controlled experiments are a cheat, since they are not ecologically valid
    Controlled experiments are indeed the right way to go at this: but 'keeping the surface the same and varying the affordances' is the incorrect way to do things because you are simply building an equivalent configuration and making the difference imperceptible.

    The correct experimental design (the one we actually do) is to perturb candidate information variables to try and prevent the specification of the affordance; you then measure changes in behaviour to see if you succeeded.

    I'm not ruling out controlled experiments: I'm saying your proposed design is incorrect.

    But, Andrew, you're missing the point. This is about ecological physics, if you will. It's not about evolved perceptual system abilities.
    Actually it is. The exploding box poses a problem because it is an equivalent configuration set up. If such set ups were ubiquitous, we would have reason to believe perception would have evolved to cope with it. But they are extraordinarily unlikely, because of the ecological laws that govern specification and the extensive access we have to that information via multiple perceptual systems, making the space sufficiently high dimensional to make ECs unlikely by chance. So anything that entails an EC (eg your experimental design) is asking the wrong question.

    You don't have to keep repeating your argument. I understand what you are claiming. You are incorrect, however, and the reasons are on my blog. Also, these comments have arisen in response to specific things you've said, eg about how to do the science. They're about why you don't have the right design.

    ReplyDelete
  22. Andrew, whether or not the composition and layout of surfaces constitutes an affordance is not a matter of perception. It's not even about the structure of light. It's not even about equivalent configurations. You could test using equivalent configurations, but you don't like that. Ok. There are other ways. Use physics. It's simply a matter of physics that the composition and layout of the surfaces of a chair, a floor, or a box does not determine the weight that it will bear, hence whether the chair, floor, or box affords sit-on-ability-for-an-adult-human, stand-on-ability-for-an-adult-human, walk-on-ability-for-an-adult-human. I repeat the point, since your typical reply is to ignore it and, say, "Look at this other thing ...."

    ReplyDelete
  23. It's simply a matter of physics that the composition and layout of the surfaces of a chair, a floor, or a box does not determine the weight that it will bear
    Actually this isn't true: the composition and layout is precisely what determines these things. The issue is whether there is information for the composition and layout; you say no, but only because you restrict yourself to vision and the various other problems I've already covered.

    I have yet to hear anything new that I haven't already rebutted, so at my end there's nothing new to worry about.

    ReplyDelete
  24. So, you don’t need a further iteration of the core argument. How about something on Runeson?

    So, Runeson claimed that a normal room and an Ames room were equivalent configurations, but that humans perceive both rooms as normal, since the Ames room was such an improbable configuration. So, let’s extend that reasoning to the exploding boxes. Since the exploding box is so unlikely, one should visually perceiving the exploding box as a normal box, hence as affording pick-up-ability. But, the exploding box does not afford pick-up-ability. So, applying the Runeson story doesn’t help secure the visual perception of affordances.

    ReplyDelete
  25. Spray paint the surface of a chair. The composition and layout of this surface does not constitute the weight bearing affordance of the chair.

    ReplyDelete
  26. Spray food coloring on the surface of some food. The composition and layout of this surface does not constitute the edibility of the food.

    ReplyDelete
  27. It's simply a matter of physics that the composition and layout of the surfaces of a chair, a floor, or a box does not determine the weight that it will bear
    Actually this isn't true: the composition and layout is precisely what determines these things.


    Ok. At least now you are talking about what I am talking about. This is what I have pressing. Do the composition and layout of surfaces typically constitute an affordance?

    ReplyDelete
  28. But, Genny, you're missing the point. This is about ecological physics, if you will. The core is that the composition and layout of surfaces does not typically constitute an affordance. Whether or not an object will bear your weight (hence whether it affords sit-on-ability, climb-up-ability, walk-on-ability) depends on more than just the composition and layout of the surfaces. It depends on the internal structure of the thing, the structure that is not exposed to light.

    Spray paint the surface of a chair. The composition and layout of this surface does not constitute the weight bearing affordance of the chair.

    According to Gibson, the internal properties of the object are reflected on the surface. See his section on the substances of surfaces (earlier in the book - I left it in the lab and don't have it on hand). He believes that the internal properties of an object are visible in the sense that their relations with surface properties are represented by invariant features of the surfaces. I didn't think I had to argue for this since you agreed to accept, in an earlier post, that we can get this information from surfaces.

    As I suggested in my examples, I don't think this is true for trick chairs or exploding boxes. I think Andrew concedes this point as well. (Is that correct?) However, I don't think this is true for ice ponds and other natural objects. As Gibson points out, it's difficult to quantitatively describe an affordance, so the only way we can test your theory is to see if someone can learn which ice ponds as walk-on-able or not. Of course this is a proxy for what you actually want to test - whether affordances structure optic flow - but the only way that we can test that is by observing behavior.

    ReplyDelete
  29. One more quick point - I was hoping my examples would emphasize the point that affordances are not properties of objects. It's important not to forget that when talking about ponds, experiments or boxes.

    ReplyDelete
  30. So, Runeson claimed that a normal room and an Ames room were equivalent configurations, but that humans perceive both rooms as normal,
    Actually, as identical.

    Since the exploding box is so unlikely, one should visually perceiving the exploding box as a normal box
    As an identical box.

    hence as affording pick-up-ability. But, the exploding box does not afford pick-up-ability.
    It does. The box has everything you need to be able to pick it up; it just happens to also have one additional imperceptible property. The same is true of the Ames Room (although of course it's easier to reveal the problem with the Ames Room).

    ReplyDelete
  31. Do the composition and layout of surfaces typically constitute an affordance?
    I would say that given affordances are non-accidentally anchored in physical properties, surfaces will often be the way they are because of those properties. Their composition (eg rigidity) will also typically figure, I guess.

    I guess it means, what do you mean by surface? Your examples all involve cosmetic changes which hide the actual surface; I'm thinking more 'tree branch' and 'concrete surface' than 'painted bike'. Actual surfaces often happily reflect their properties, internal or otherwise.

    Take palatability; something being edible often does have consequences on the surface, which they often try to camouflage with varying success. The fruit example is an interesting one, actually.

    That was just some thinking out loud; typically? I really don't actually know. Often Sure. I'm also, of course, not concerned about affordances that are not constituted this way, because we do, of course, have plenty of perceptual access to those too.

    ReplyDelete
  32. So it looks like Andrew can get on board with some of my arguments and is ceding to Ken that EP can only go so far (perhaps only applicable to natural environments?). Except he might also believe the exploding box is pick-up-able, which I'm willing to forgo to argue on Ken's terms.

    Ken, it seems like you don't want to grant that the substances of surfaces affect the surfaces themselves (except for broad cases like moving water vs. rigid ground). I was arguing that if you walk around an ice pond, there is enough information about the surface of the pond to specify whether it's walk-on-able or not, although it might require some learning. Gibson suggests that the only way to find out if this is true is to see if people can learn to differentiate walk-on-able pond from ponds that are not walk-on-able, so perhaps this is an empirical question.

    ReplyDelete
  33. Andrew, Genny,

    In your last comments you defend something different than what Gibson is proposing on p. 127. Remember Gibson's conjecture: "Perhaps the composition and layout of surfaces constitute what they afford."

    In the first paragraph of his comment, Andrew essentially restates Gibson's view, but then moves on to articulate another view.

    Gibson is not suggesting that (the composition and layout of) surfaces reflect their affordances. He is suggesting that the composition and layout of surfaces constitutes affordances.

    Gibson is not suggesting merely that (the composition and layout of) surfaces will be influenced by their affordances.

    Gibson is not suggesting that affordances (such as being edible) or even internal properties will have consequences for their surfaces.

    Gibson is not suggesting merely that the substances of surfaces influence surfaces.

    Can we focus on what Gibson is claiming here and get straight on that, before moving on to other matters?

    ReplyDelete
  34. So, Andrew proposes a distinction between "actual surfaces" and "cosmetic surfaces" and that EP only works for actual surfaces, but not cosmetic surfaces. And, Genny interprets this to mean that EP works for natural environments, but not artificial environments.

    This, however, seems to be a departure from Gibson. Recall what Gibson writes on p. 130, just three pages after the sentence under scrutiny.

    "Why has man changed the shapes and substances of his environment? To change, what it affords him. He has made more available what benefits him and less pres' what injures him. In making life easier for himself, of course, he has made life ha for most of the other animals. Over the millennia, he has made it easier for himseff get food, easier to keep warm, easier to see at night, easier to get about, and easier to train his offspring.
    This is not a new environment-an artificial environment distinct from the natural environment-but the same old environment modified by man. It is a mistake separate the natural from the artificial as if there were two environments; artifacts have to be manufactured from natural substances. It is also a mistake to separate the cultural environment from the natural environment, as if there were a world of mental products distinct from the world of material products. There is only one world, however dive and all animals live in it, although we human animals have altered it to suit ourselves." (Gibson, 1979, p. 130)

    So, if Andrew wants to break ranks with Gibson and claim that Gibson is wrong about the artificial cases, that would be great. Maybe then we could move on to seeing how Gibson's theory fails even in natural cases.

    ReplyDelete
  35. So, Andrew proposes a distinction between "actual surfaces" and "cosmetic surfaces" and that EP only works for actual surfaces, but not cosmetic surfaces.
    I was simply noting that paint is designed to conceal some of the actual properties of surfaces. It occurs to me this might interfere with our visual access to certain anchoring properties, because it breaks the non-accidental relation between the surface and it's composition.

    Gibson's talking about artefacts and design, not paint; I'm perfectly on board with that analysis.

    ReplyDelete
  36. Oh I completely agree that I've broken from Gibson. My examples were meant to illustrate where we can still salvage some notion of affordances.

    Perhaps I was a little inaccurate with what Gibson means with respect to substances. I recommend taking a look at Chapter 2. Lots of properties of the surface depend on the nature of the substance. Whether the part of the optic array corresponding to pond ice is walk-on-able is constituted by the surface at that location - it is the nature of the surface that specifies whether that part of the optic array is walk-on-able or not. When I wrote that the substance is reflected in the structure of the surface, I meant that the substance structures the surface so that surfaces that separate different substances from the medium would be different (and would structure light differently).

    I also was trying to point out early on that Gibson is a little loose with his terminology so I wouldn't hang too much on the formulation on p.127.

    ReplyDelete
  37. Ok. Good, Genny, so you agree that Gibson was wrong to maintain that the composition and layout of surfaces constitutes affordances. (Indeed things that Gibson says in Chapter 2 correspond to my slogan "surfaces on the outside; affordances on the inside). So, why don't you help me explain to Andrew why Gibson is wrong? Then we can move on a bit.

    ReplyDelete
  38. But, Andrew, you are proposing an exception/amendment to the hypothesis that the composition and layout of surfaces constitutes affordances, namely, the composition and layout of "actual surfaces", but not "cosmetic surfaces" constitutes affordances.

    The problem is that nothing in Gibson seems to warrant this. Gibson seems to suppose that the natural environment and the artificial environment get the same analysis.

    Any, why wait all the way until p. 130 to see that Gibson doesn't make an exception for artificial environments of the sort you are suggesting? Check out p.127 above "But I have described what the environment affords animals, mentioning the terrain, shelters, water, fire, objects tools, other animals and humans displays".

    He thinks his theory works for human displays, but what are many human displays but paints (cosmetic surfaces) on canvas (actual surfaces)?

    So, are you going to conclude that Gibson was wrong in thinking that his theory does not work for paintings?

    ReplyDelete
  39. I don't know what I'll conclude - I've only just started considering the problem. I just got thinking about what it is that paint does, and thought it had some interesting implications for perception in a modern world. That's all.

    I don't see it being a special case, or anything; Gibson's right that the natural and man-made worlds are the same type of thing and get the same kind of attention. Camouflage, mimicry, all occur in nature and have consequences for behaviour - maybe a paint job is like that? I don't know.

    ReplyDelete
  40. "Camouflage, mimicry, all occur in nature and have consequences for behaviour"

    And in all of Ken's examples it's the instantaneous occurrence of a consequence that poses the problem, not the continuous, intentional behavior over the time interval leading up to it: being blown up, poisoned, etc; falling through/with a breaking/collapsing surface; walking into an imperceptible surface. Prior to that instanteous occurrence each "apparent" affordance does its job of motivating completely appropriate action. Hmmm.

    ReplyDelete
  41. I agree that he was wrong for many examples. Mostly, however, I just disagree with what he counted as an affordance - anything beyond his "basic" affordances seems a bit zany to me. I was hoping that Andrew would be convinced by either the trick-chair or the exploding boxes cases as I described them above.

    Incidentally, going back through the book, I think we might be able to handle the trick-chair scenario by looking at his definitions of attached and detached objects (p.34). Ken, I'd be curious to know what you think about those distinctions since they hinge on the relations between the substances of surfaces, not the surfaces themselves.

    To play devil's advocate for a bit, he also has a section talking about the chemical composition of substances. Perhaps this can be used to deal with objects that will electrocute you. I guess a lot hinges on how much of substance you allow into your definition of surface.

    ReplyDelete
  42. Genna -

    All of Ken's examples depend on denying that there can be misperceived (or apparent) affordances, ie, that a surface can appear to offer an affordance that can't be actualized. Ken (and perhaps Gibson - I'm not sure) accepts that denial and logically concludes that since there can't be apparent affordances, if one misperceives a surface as making an offer that can't be actualized, whatever was offered can't be an affordance. (This is just restating informally what I presented formally in the first comment in this thread.)

    Bringing in anchoring properties, how surfaces structure light, the conduct of empirical studies, non-visual perception modes, etc, may be important in more realistic scenarios, but in these simple ones, they just obscure the basic question: can one misperceive an affordance as being offered when it actually isn't offered? If one joins Ken in answering "no", then IMO he wins by default since for any proposed affordance one can devise a scenario in which it appears to be offered but is negated by some imperceptible feature of the environment.

    But denying the possibility of misperceived (apparent) affordances seems to make no sense. Affordances are a conceptualization of offers to actors in real situations, and real actors can make perceptual mistakes. So, I accept the possibility of misperceived (apparent) affordances, thereby negating all of Ken's examples.

    ReplyDelete
  43. Charles,

    I don't care if the visual perceived moves around, looks around, walks around. In a lot of the cases, such as the painted thick and painted thin floor cases, I don't care if the perceiver smells the floor, licks the floor, puts his ear to the floor, or places his fingers lightly on the surface of the floor. Subjects will still not be able to use surface properties to determine the affordance of the suface. What the surface affords is determined by what is beneath it. To use the Gibsonian terminology, what weight a surface will bear is determined by the substance of which the surface is a surface.

    ReplyDelete
  44. Genna,
    So, suppose we draw a distinction between "basic" and "non-basic" affordances. What is the theoretical justification for this? We don't want merely some ad hoc innovation (such as an "actual surface"/"cosmetic surface" distinction) that saves Gibson's view.

    I don't see that anything on p. 34 is of any help to Gibson regarding the composition and layout of surfaces constituting affordances. Most of the distinctions on that page concern the relationship between surfaces and the medium. I take it that the fake chair, your zippo lighter, the exploding boxes, and the statue of Elton John are all detached objects. The pond ice, the spray painted floor, and the glass floor all seem to me to be attached objects. But, the problem seems to me to be unchanged. With both types of objects, the composition and layout of the surfaces do not constitute the affordances.

    When it comes to the shark in the water, the cabinet that shocks you, it seems to me that Gibson is just noting a problem, then capitulating and retreating to basic affordances. There is nothing there to help him "solve" the problem.

    Notice that the principal thing that surfaces are supposed to do is structure light. In Andrew's distinction between "actual surfaces" and "comestic surfaces", when we have a "cosmetic surface" over an "actual surface" (a distinction I see no reason to think that Gibson endorses), i.e. when we have paint on a chair, the "comestic surface" is the one that structures light, but the "actual surface" does not. So, we still don't get a case in which the light structurer constitutes the afforder, which is what Gibson was after. This does not enable to have one thing that does the two jobs Gibson wants the one thing to do. There is no light structuring afforder.

    ReplyDelete
  45. But denying the possibility of misperceived (apparent) affordances seems to make no sense. Affordances are a conceptualization of offers to actors in real situations, and real actors can make perceptual mistakes. So, I accept the possibility of misperceived (apparent) affordances, thereby negating all of Ken's examples.

    You can surely avoid my objections by dropping the theory I object to, namely, Gibson's theory. I'm not sure that really counts as "negating all of Ken's counterexamples". To me, it seems like accepting the counterexamples and moving on to a theory that doesn't succumb to them.

    ReplyDelete
  46. "To me, it seems like accepting the counterexamples and moving on to a theory that doesn't succumb to them."

    I thought I had made it clear that I am doing precisely that - in response to your convincing examples, suggesting that the assumption that affordances can't be "apparent" should be eliminated.

    This seems a modest and reasonable tweak of a definition that seems pretty amorphous already. If acceptable, it would get us off what seems to have become a merry-go-round going nowhere. And if not acceptable, it would be nice to know why - hopefully for some more convincing reason than "because Gibson said so".

    ReplyDelete
  47. You can surely avoid my objections by dropping the theory I object to, namely, Gibson's theory.
    This kind of logic showed up in the first pages of Bounds of Cognition, too (which my library finally bought for me, so you're welcome there). It's not patronising at all :)

    We don't want merely some ad hoc innovation (such as an "actual surface"/"cosmetic surface" distinction) that saves Gibson's view.
    ...
    Notice that the principal thing that surfaces are supposed to do is structure light. In Andrew's distinction between "actual surfaces" and "comestic surfaces", when we have a "cosmetic surface" over an "actual surface" (a distinction I see no reason to think that Gibson endorses)

    This isn't what I'm up to, by the way. Your specific example about painted surfaces got me wondering about what happens in the specific case where the surface has been cosmetically altered. I'm not suggesting Gibson holds for one but not the other.

    Having a good theoretical basis is like being a securely attached child: I'm happy to wander around a bit, kick the tires, see what's going on, without being in any doubt I'll find a solution. I may be wrong, but I see no evidence yet, and cries of 'Oo! Oo! schism!' seems a bit weird.

    Charles
    And in all of Ken's examples it's the instantaneous occurrence of a consequence that poses the problem, not the continuous, intentional behavior over the time interval leading up to it: being blown up, poisoned, etc; falling through/with a breaking/collapsing surface; walking into an imperceptible surface. Prior to that instanteous occurrence each "apparent" affordance does its job of motivating completely appropriate action. Hmmm.
    Yes. I'm still not sure I see the need to call them apparent affordances (from the first person perspective, they simply are) but yes.

    Genna; Charles' point is well taken. And I'm not ceding all that much to Ken, certainly not that Gibson only holds in natural environments. I think surfaces provide plenty of information for affordances, and when they don't, we have things other than vision to fill the gaps, and then sometimes things are imperceptible.

    ReplyDelete
  48. "I'm still not sure I see the need to call them apparent affordances"

    Because according to the definition implicit in Ken's argument, the information in light from a surface must be sufficient to allow an actor to perceive an affordance that is guaranteed to be actualizable. If there is any possibility of failure when an actor attempts to actualize a perceived affordance, then that demonstrates that affordances can't structure light so as to reveal themselves in every detail. (Ken prefers Gibson's phrase "the composition and layout of surfaces typically constitute an affordance", but I think my phrasing more clearly captures the relationship between the information in light and the details of the affordance the light is being asked to characterize). So the essence of Ken's examples is that if an actor attempts to actualize a perceived affordance and fails for any reason, that shows that the affordance didn't structure light so as to reveal itself in every detail.

    If a box is perceived as affording pick-up-able but explodes when approached or touched, then the affordance didn't structure light so as to reveal the booby trap. If a stool is perceived as affording sit-on-able but collapses on contact, then the affordance didn't structure light so as to reveal that the legs had been sawed through. If a frozen pond is perceived as affording walk-on-able but the ice breaks at the first step, then the affordance didn't structure light so as to reveal that the bearing weight of the ice was one gram less than the weight of the perceiver. Etc, etc.

    Put this way, I think it's clear that no affordance can do what is being asked of it - one can always concoct a scenario in which some critical detail of an affordance is not revealed by the light's structure. (That's why Ken has been so prolific in concocting scenarios; once you see the trick, anyone can do it - even me!) Hence, the Gibson project fails.

    That's why I argue that it's not a matter of anchoring, movement, Mill's whatever - it's a simple consequence of the assumption. Removing that (IMO, ridiculous) assumption solves the problem. "Apparent" is just my shorthand for "the following affordance isn't assumed to be required to do the impossible". Any similar qualifier will do.

    ReplyDelete
  49. the information in light from a surface must be sufficient to allow an actor to perceive an affordance that is guaranteed to be actualizable. If there is any possibility of failure when an actor attempts to actualize a perceived affordance, then that demonstrates that affordances can't structure light so as to reveal themselves in every detail.
    ...
    So the essence of Ken's examples is that if an actor attempts to actualize a perceived affordance and fails for any reason, that shows that the affordance didn't structure light so as to reveal itself in every detail.

    Huh. I guess I'm comfortable with the fact that failure to effect the affordance can happen for a variety of reasons (you blink, you're tired, etc etc). You tell the difference between the two cases by the fact that success will look highly structured and consistent while failure will be wildly variable. Failure because you're systematically missing something (Ken's suggestion) would have more structure.

    Also, there are two dispositions in the exploding box case, one a subset of the other; pick-up-and-explodability, and pick-upability. There is information only for the latter, hence your expected behaviour. I agree with your point that imperceptible properties aren't relevant, and so I guess I'm not worried about such scenarios. The ease with which you can generate them is a hint that they aren't sufficiently constrained.

    ReplyDelete