Wednesday, May 13, 2009

Noë and Norm

Since there has been so much interest and informed discussion of Noe, let me venture another post on this idea that perception depends on the possession, rather than exercise, of sensorimotor knowledge.

Here’s the opening move in an objection …

Consider, Norm, a normal college student who is familiar with the game of ping pong. Take two white ping pong balls and cut a ¼” hole in one of them Set the two ping pong balls side by side on a table and have the hole facing away from the door to an otherwise empty windowless room with standard incandescent lighting suspended from the middle of the ceiling. Let Norm enter the room at t0 and stand erect on a spot marked with an “X three feet from the ping pong balls. Given his past experience with ping pong balls, at t0 Norm believes that both ping pong balls are the same and that neither has a hole in it. Presumably at t0 the two ping pong balls look the same. They will give rise to the same ping pong ball like experience. One, of course, will be experienced as on the left and the other on the right, but they will be otherwise alike. But, how can this be on Noe’s theory? Norm possesses tacit sensorimotor knowledge of one of the balls, but possesses mere tacit sensorimotor beliefs about the other. Norm knows what will happen if he moves around to the other side of one ping pong ball, but does not know what will happen if he moves around to the other side of the other ping pong ball. Differences in one’s tacit sensorimotor knowledge should give rise to differences in experience (right?), but they presumably don’t. Call this the Problem of Naïve Norm.


  1. Nice post, Ken.

    It seems that the most obvious solution for someone attracted to enactivism is to say that if you want to press a strong distinction between knowledge and belief, then enactivism isn't so much about sensory motor know-how, but sensory motor "beleive-how." What's wrong with that? Is Alva *really* committed to stating the theory in terms of knowledge? Or is he just using the term "knowledge" in a relatively non-philosophical sense, as one finds in common-sense and cognitive science, where there need not be such are strong distinction between knowledge and beleif?

  2. Pete, you are the best!

    So, next move ...

    Maybe the solution to the Problem of Naïve Norm is that Noe does not mean “knowledge” in any epistemological sense; instead, Noe means “knowledge” in a kind of psychologist’s sense, which amounts to something like belief. It is because Norm has the same visual stimulation and the same tacit sensorimotor beliefs about the two ping pong balls that leads them to look the same to Norm at t0 from the perspective of standing erect at the spot marked “X”.
    Ok. So, after t0, let Norm move around the ping pong balls as much as he likes without moving them. Following this examination—this active exploration—Norm sees that the 1/4” hole in one of the balls. Let him continue to explore the ping pong balls for minutes, hours, or days, if necessary. After all this experience with the ping pong balls, at t1, let Norm stand erect once again on the spot marked “X” three feet from the ping pong balls. Presumably, after all this active exploration, the two ping pong balls still look the same to Norm. But, how can this be on Noe’s theory? Doesn’t Norm’s vast experience with the ping pong balls now give him different tacit beliefs about the two ping pong balls? After as much active exploration as you please, doesn’t Norm have the tacit belief (and indeed tacit knowledge) that if he moves in a certain way, then one ball will reveal a hole, where the other does not? Call this the problem of Educated Norm.

  3. Oh, and to my knowledge, Noe always seems to phrase things in terms of sensorimotor knowledge, and never sensorimotor belief. There's also "understanding" sensorimotor contingencies, which also seems, to me at least, like it requires truth.

  4. Ok, very good. I think Educated Norm is a pretty serious problem, and I'm not aware of anything in Alva's books that can get him out of it. I think that he is committed to saying that the two balls look different, and I agree with you that that is a poor consequence to have of one's theory.

    For what it's worth, the kind of conceptualism/enactivism that I'm attracted to builds in a way for screening off which "knowledge" infects experience and which "knowledge" does not. I'd say that, given the case of Educated Norm as described, even though he knows that there's a hole in only one ball, this doesn't give rise to a difference in the experiences of the two balls, since the application of the knowledge to experience is inferential, not automatic. The point is essentially one I lift from Paul Churchland (who gets it from Sellars). I develop this in the below linked and cited paper. See especially the part about the novice and the expert looking at an actor in a realistic gorilla suit:

    Mandik, P (2006) The Introspectability of Brain States as Such. In Brian Keeley, (ed.) Paul M. Churchland: Contemporary Philosophy in Focus. Cambridge: Cambridge University Press.

  5. Hi. I had the same thought as Pete when it comes to Naive Norm, and have some questions about how Noe's claims are being understood here.

    You write in the opening that perception is a matter of possession rather than the exercise of sensorimotor knowledge. I confess that I haven't read the previous tread, so I don't know if this view is being attributed to Noe or simply as a similar hypothesis for discussion but in a 2007 article in Philosophical Psychology he does claim that "Perceptual presence requires that the perceiver possess and exercise sensorimotor understanding." (p. 533)

    Also, it might be of interest to know that in the same article (p. 533 again, I think) he does indeed talk about sensorimotor belief and sensorimotor expectations in the same breath as sensorimotor know-how and understanding (I believe talk about sensorimotor expectations crops up a lot in his work generally): "My proposal is that what explains the perceiver’s sense, now, that the occluded surface is present, is his or her expectation (or knowledge, or belief, or anticipation) about what would happen if he or she were to move."

  6. I think this objection gets its force from Noe’s apparent vacillation in how he is thinking about sensorimotor knowledge. At several places he writes as if such knowledge is ‘know-how’: it consists in the possession of skills, not knowledge of propositions (for now, let’s put aside worries over whether know-how involves knowledge of propositions). But then he goes on to say things like this (as pointed out in another post on this blog):

    "Your perception of the tomato as voluminous depends on your tacit understanding of the ways its appearance (how it looks) depends on movement. You visually experience parts of the tomato that, strictly speaking, you do not see, because you understand, implicitly, that your sensory relations to those parts is mediated by familiar patterns of sensorimotor dependence." (_Action in Perception_, p77)

    You understand THAT your sensory relations...blah, blah, blah. So it looks like SK is knowledge-that after all!

    How damaging is this to Noe’s overall project, and can he still avoid the objection raised by the example of Norm? First, I think Pete is right: Noe is not using ‘knowledge’ in the epistemologist’s sense, but in the way familiar from cognitive science. But the bigger question remains: is this ‘knowledge’ propositional, and if it is, then why does not Educated Norm experience the two balls differently after walking around them?

    Here’s a rough sketch of an answer: the beliefs he acquires after walking around do not become part of the tacit understanding that is supposed to account for how things look to Educated Norm. Why? Because, despite Noe’s occasion use of ‘understand that’, the tacit understanding you have is not understanding of propositions. Furthermore, there is no reason to believe that acquiring new beliefs will automatically affect your non-propositional understanding in this case (just as acquiring new beliefs about how to play tennis does not automatically affect how you play tennis).

  7. Dan,
    Yes, the previous post proposes that we weaken Noe's view to assert merely that it is possession, and not exercise, of sensorimotor knowledge. I think that Noe's considered view is that it's both poss & exer that counts, but a few people have proposed the weaker poss alternative.

    Thanks for drawing my attention to what Noe says in the Phil Psych paper.

  8. Pete,

    You must be reading my mind, as I too thought of the Churchlandian idea (in the Fodor-Churchland debate of the late 1980's) that it sometimes takes time for cognitive penetration.

    Pete and Anonymous,
    On making the beliefs/knowledge into know-how, I tried to forestall this reply with Educated Norm. I did not set Educated Norm up with Professor Pete merely telling Norm, "Hey, Norm, the ping pong ball on the right has a 1/4" hole in it." That would presumably give Norm only propositional knowledge.

    Instead, I set it up so that Norm gets to interact in protracted ways with the ping pong ball set up. That's why I invoked Noe's idea of active exploration. I thought that that might be the way to develop the requisite tacit sensorimotor understandings. But, my aim was to simply plug in what I thought would be Noe's way of acquiring tacit sensorimotor beliefs/knowledge.

    I guess I have been assuming that one's tactic sensorimotor beliefs/knowledge is plastic. (Pete would, I think, holds this view, alghouth I tend to be more sceptical on this.) But, maybe Noe could deny this. Maybe he could say that humans have a fixed repetoire.

    On the other hand, Noe does make a lot of hay out of the plasticity following inverted goggles, left-right reversing goggles, etc. So, I don't know that this is an option for him. Thoughts?

  9. Ken,

    Ah, I see your worry now! Thanks for the clarification. Maybe Noe could say something like this?

    Part of our sensorimotor understanding of ping pong balls is that there is no reliable correlation between presented-surfaces-of-ping-pong-balls-at-a-certain-location and presence-of-holes-in-hidden-surfaces-of-ping-pong-balls, and thus finding out that THIS ping pong ball has a hole in it will not have the effect of altering how the ball looks to us in subsequent encounters (even if we actively explore the ball).

    The plausibility of this story will turn, in part, on whether we have good reasons to think that sensorimotor knowledge does not track particulars. By this I mean that, even though I can re-identify a particular ball, and even though I can remember actively exploring that ball, such explorations will matter to how things look only insofar they affect the generalizations I am prepared to make. In other words, whether the ball with the hole in it will look different to me will depend on whether I implicitly accept a generalization concerning the location of balls and the presence of holes.

  10. Hmm, there is something I'm missing here.
    Why is it a problem to say that the perception of the two balls, following exploration, is different?

  11. Anon,
    Glad that worked for you. So, switch examples.

    Let U be a uniform gray 10' x 10' square of a particular luminance and let G be a 10' x 10' square grating of frequency F with the same average luminance as U. At t0 let Naive Norma stand at a distance D0 (maybe 100 yards) such that U and G are metamers. This is like Naive Norm.

    Now, let Norma actively explore the two squares, walking closer, etc. Then, have her return to D0.

    Now, I've been told that with various sorts of practice one can get improvements in one's ability to resolve gratings, but let D0 and F be large values and let the active exploration involve Norma getting very close, perhaps using highly sensitive light meters.

    The basic idea is begin with things that are metamers under conditions C, then vary C by active exploration.

  12. Anthony,

    My example being a mere thought experiment, one can, of course, not be certain of the outcome of an actual experiment. So, that I must concede.

    Second, even if one were to do some sort of psychophysics experiment that revealed indiscriminability, I suppose that it would still be possible to stay that the perceptual difference in looks is hard to detect.

    So, I do worry about Noe taking this line. This worry made me throw in a lot more details than I would have liked, perhaps cluttering up a simple idea.

    Those concessions in place, I would wonder what account might be given of the difference in the perception of the two balls following exploration. How is it that they look different? Is it that one looks like it is whole, where the other looks like it has a hole? How do U and G look different?

  13. Hi Ken,
    I take Noe to be arguing for a much more holistic view of perception, especially in the new book. Perception is not like a picture, Noe rejects the snapshot hypothesis. So while the visual image of the two balls is unchanged, the perception of them is changed.

    This makes sense I think and is clear when we consider more extrema examples: Norm is presented with two identical looking doors, behind one door is a vicious looking dog (and Norm is very scared of dogs). Following exploration even the idea of opening the door with the dog behind it elicits fear in Norm, an emotion that he perceives accompanied by many somatic changes which are also perceived (tightening in the stomach, hairs standing up etc.). So it seems clear that his perception in this case is changed.

    This is not however a passive perception, it is the thought of an action (the covert application of SK) that elicits such a perceptual change. Is his perception of the scene different? I would say yes definitely.

    Even with the ball example, if you ask Norm, post-exploration which ball has a hole in it, he points to the correct one. How can you possibly explain this if you suggest that his perception of the two balls is the same. If like Millikan you invoke the different histories that Norm has with the 2 balls, this amounts to saying that Norm has different SK of the 2 balls.

    So U and G are perceived differently. Do they look different from here? that all depends on what you mean by look. I suspect that you mean, if Norm took a photo and we analysed the image, would we discover that one ball has a hole in it, that one is not going to behave in the right way to play table tennis with etc. in which case the answer is no.

  14. Anthony,
    In this thought experiment, the visual image of the two balls is supposed to be unchanged. That is now common ground. And I can concede, for the sake of argument, perhaps, that the (total) perception is changed. But, I'm proposing that the *visual* perception is also unchanged. That is the problem.

    Visual perceptions are delimited by Noe's theory of sense modalities. So, that is kosher.

    Noe also draws a distinction between what we perceive and what we believe (cf. Noe, 2004, p. 60). So, it seems to me that Norm does not visually perceive the presence of the dog (cf. Noe, 2004, pp. 64-65). In fact, my guess is that he perceives an emotion (fear) in addition to the door.

    How do I explain why Norm correctly points to the ball with a hole upon query? I say that he visually perceives the balls to be the same, but knows that the one with the hole is on the right.

  15. Hi Ken,
    I think we are pretty much in agreement then. Norm visually perceives the balls to be the same, but knows the the one with the hole is on the right.

    Similarly in Noe's example the visual perception of the tomato does not include the back, though the whole perception (vision plus) of the tomato does include the back.

  16. Anthony,
    I don't think this last paragraph is Noe's view. On p. 63, AIP, he writes,
    "One experiences the presence of that which one perceives to be out of view."
    In looks like he's only talking about vision, not vision plus. Where do you find this idea of vision plus in this part of his text?

    A bit later he says,
    "In this way, we can explain our sense of the perceptual presence of, say, the whole tomato. Our perceptual presence of the whole tomato's wholeness--of its volume and backside, and so forth--consists in our implicit understanding (our expectation) that movements of our body to the left or right, say, will bring further bits of the tomato into view"

    I think it is Noe's view that you perceive the whole tomato including its back, even though you do not see the whole tomato including its back. The model for this, I take it, is amodal completion in which one perceives a corner of a square occluded by a circle, even though one does not see the corner of the square occluded by the circle.

  17. Hi Ken,

    I'm not sure what Noe's view is on this but I don't see a problem here. I experience the world as 3D, as having form, as including solid objects etc. I think if I set up a flat 2D picture of a tomato and sat in just the right place so that my view of it was identical to a real 3d tomato, my experience would be different. I would predict that my movments would have different results in the 2D case than the 3D case. With the real tomato I "experience the presence of that which I [visually] percieve to be out of view". On my reading of this I take the experience of presence to be more than visual, including such things as what type of grip I would use to pick it up etc. From the second quote my movements (non visual modes) can bring the back into view (visual) but that does not imply that I can see the back now. In fact the opposite seems to hold, this is the knowledge of how to bring the back into view and so the back cannot already be in view.

  18. Gennady ErlikhmanMay 24, 2009 at 11:02 AM

    I think Noe's response here might be something along the following lines:

    Look, perception is all about _how_ we look at things, not just about looking at them. Take a different example (I'm not sure if he uses this one in his book...) - if one looks at a novel degraded image, one might simply see a mass of black and white squares, but when informed of the content, suddenly the image organizes itself and we see a dalmatian or Darth Vader (unfortunately I couldn't find any good links to these, but they are online somewhere). What has changed? Surely the visual input is still the same, the image hasn't changed. But it _looks_ different; or, and this is what I think Noe's main point is: _we_ look at it differently. _We_ organize the image differently.

    In many ways, this is analogous to cases of seeing-as, as in the classic duck-rabbit or the old-woman-young-lass images: some people can only see one or the other and only when they learn to look at it in a particular way can they see it. Note that it is insufficient to have merely semantic knowledge! That is, if a person is told, "here is the nose", that might not be enough to get them to see the image the right way. Instead, people need to acquire an ability (SK?) to _look_ at it the right way.

    Something similar could be said about the Educated Norm case. Sure, the _image_ of the balls looks the same from this vantage point, but how we _look_ at the balls, _how_ we see them is different. For example, we know which way to move to see the back of a ball and which way to move to see a hole. I think that's the use of "know" that Noe had in mind and is Anthony's point. It is only in that sense that we "perceive" the back of each ball. One might want to say here that we represent the ball as having _a_ back, for that is what it means to be a ball and not a bowl, but not a _particular_ back. So we can perceive the moon as having a back without knowing anything about it.

    (There's an interesting point to be made here about enactive perception, for on this account, then, if I were to come into a new room with two new ping pong balls I could _choose_ what I would see them as - whole balls or balls with holes. I'm not sure the extent to which one might want to endorse that view...)

    So, like Anthony, I'm having a hard time seeing why this case in particular is such a big problem for the theory; although I do think there are plenty of other holes.

  19. Hi, Anthony,

    I've been out of town for a bit ....

    Let's see if we can find some common ground on this tomato.

    I think all agree that we only strictly speaking see the front of the tomato.

    I think all agree (if only for the sake of argument on my part) that we non-visually experience the front and back of the tomato.

    Do we also then agree that we only visually perceive the front of the tomato and not the back? It is this last case that I am pushing for. I am trying to argue that we do not visually perceive the back of the ping pong balls.

  20. Hi, Gennady,

    You raise an interesting interpretation of what Noe is up to, one that I had not thought of. Nonetheless, I think the problem remains.

    My point is that the ping pong balls case is unlike the high contrast dalmation case. Additional info does not change how I experience the ping pong balls, where additional info (let us say) does change how I experience the dalmation.

    I agree that we know which way to move to see the back of a ball and which way to move to see a hole. That's part of the set up of the thought experiment. The problem, then, is that this does not change my visual experience.

    It seems to me that the general drift of your and Anthony's reply is to deny that one's visual experience is unchanged by the acquisition of SMK.

  21. Gennady ErlikhmanJune 4, 2009 at 5:40 PM


    Thanks for the reply! My response is rather long so I'm going to have to break it up into two posts.

    I agree with your last sentence; at least in that I think that that's what Noe is trying to argue for.

    For example, he writes,
    "[…] our sense of the perceptual presence of the cat as a whole now does not require us to be committed to the idea that we represent the whole cat in consciousness at once. […] we take ourselves to have access, now to the whole cat. The cat, the tomato, the bottle, the detailed scene, all are present perceptually in the sense that they are perceptually accessible to us. They are present to perception as accessible. They are, in this sense, virtually present. The ground of this accessibility is our possession of sensorimotor skills" (p63 of Action and Perception).

    And later, "Our perceptual sense of the tomato’s wholeness – of its volume and backside, and so forth – consists in our implicit understanding (our expectation) that movements of our body to the left or right, say, will bring further bits of the tomato into view. Our relation to the unseen bits of the tomato is mediated by patterns of sensorimotor contingency [...] My sense of the presence of the whole cat behind the fence consists precisely in my knowledge, my implicit understanding, that by movement of the eye or the head or the body, I can bring bits of the cat into view that are now hidden" (same page).

    I think he is committing himself here to precisely the claim that SMK affects our visual experience.

    I agree that the dalmatian case isn't a perfect since many would want to say that you actually look at the image differently (you organize its elements differently, or your gaze shifts from one part of the image to another in a different order).

    This gets us into the question, then, of what exactly does Noe mean when he writes that the back of the tomato (or ping pong ball) is _virtually_ there, which is I think your question as well. Some places in the book are a little sloppier than I would like, and there are sentences like "we perceive the back of the tomato blah blah blah" but I think that Noe only means that they are _virtually_ present or perceivable.

    He writes things like, "To experience detail virtually, you don’t need to have all the detail in your head. All you need is a quick and easy access to the relevant detail when you need it" (p50). And also, "The content of perceptual experience is virtual. […] experiential content is itself virtual. According to the enactive approach, the far side of the tomato, the occluded portions of the cat, and the unseen environmental detail are present to perception virtually in the sense that we experience their presence because of our skill-based access to them. Phenomenological reflection on the character of perceptual presence suggests that the features are present as available, rather than as represented. The world is within reach and is present only insofar as we know (or feel) that is it. Crucially, phenomenologically speaking, virtual presence is a kind of presence, not a kind of non-presence or illusory presence" (p67).

    So I don't think that he means that we even represent the back of the tomato/cat/ball in the same way that we do the front; rather, the back is somehow a part of our representation, a part of what it means to see a spherical object or to know that it's a sphere or to know what views of the sphere will be presented as we move around it. That's SMK. SMK changes our visual experiences only with respect to this virtual component of the representation in the ping pong ball example (whereas, as you rightly point out, it seems like something about the actual percept is changing in the dalmatian case).

  22. Gennady ErlikhmanJune 4, 2009 at 5:40 PM

    Part II

    All of this being said, I still think there are some terrible problems with the theory as described. First of all, I think the evidence is a little too carefully selected and a lot of counter-evidence is ignored. And even the evidence in support of the enactive view I think can be interpreted to go the other way (see Prinz 2006 for similar objections).

    In some places, for example, I think Noe suggests that retinal information is first converted to either motor commands or some sort of motor information, namely knowledge of sensorimotor contingencies, and only then is sent to a perceptual system. This is an empirical question and I believe the most recent consesus is that information travels on two parallel pathways instead of serially. (As a side-note, Slater and Morison 1985 suggest that infants who are a few weeks old exhibit shape constancy - I find it extremely difficult to believe that this is dependent on SMK in barely-moving tots.)

    One of my chief worries with enactive perception is the actual applciation of the theory. What does it really tell us? Suppose that one does have a store of these perspectival (P-) properties of objects in the form of SMK. How does one combine P-properties to have a coherent representation of an entire object? How does one predict further P-properties from the present ones? That is, if I see the coin as an oval, why do _those particular_ P-properties all me to "see" (in Noe's sense) it as a circle at the same time? (An account of how SMK arises is woefully absent.)

    Basically, my main objection is that the account fails to resolve any explanatory gap and isn't really an advance over other theories. Use of patterns of information (even action patterns) for perception isn't a novel concept (Brown 1911, Riesen 1961, Smith and Smith 1962, Neisser 1976, Barsalau 1999, to suggest a few).

    I apologize for the length of my response! If you'd like to chat more outside of the blog, feel free to shoot me an email at gennaer at gmail