Thursday, June 4, 2009

Another Noe Question

Here is a simple Flash demo of amodal completion:

While the black square occludes the white thing, I do not, strictly speaking, see a circle, but I visually perceive a white circle. I think this is common ground.

Ok. Now, no matter how long I look at this demo-no matter how long I take in the sensorimotor contingencies this demo affords, (I’m assuming that) I will continue to visually perceive the white occluding thing as a circle. Contrary to what Noe seems to me to predict, we apparently have a case in which SMK does not influence my visual perception. Contrary to what Noe seems to me to predict, even though I expect that white thing to be revealed as a non-circle, I still visually perceive a circle during the maximal occlusion.

How about this?


  1. Gennady ErlikhmanJune 4, 2009 at 5:52 PM


    Cool stimulus and very interesting question!

    In Noe's shoes I would fall back on a probablistic account - SMK is rarely a case of one-shot learning about particular objects. Rather, we acquire over time a general set of skills that tells us things like, "circles seen in these conditions from these viewpoints look like ovals".

    In this case, we might say something like "occluded pacmen are often completed as circles". Maybe if we see enough deformed circles we could come to expect, to virtually complete the pacmen as deformed circles. As noted in my longer response to the May entry (again, sorry about the length!), there is an unfortunate lack of an explanation of how we come to have SMK in the first place, so really this is just a guess at what Noe might say.

    As an additional thought: would the enactive account be able to deal with Fake Barn County cases? Given enough experience with deformed circles, could we complete pacmen as deformed circles in some contexts and as full circles in other contexts? How does SMK explain this? I think there are some studies on this, but I can't remember any off the top of my head.

  2. Hi, Gennady,

    Glad you like the demo. One nice thing about this example is that there is a relatively widely understood sense of what one experiences in this case. I think it avoids, to some extent, your concern about HOW one experiences something.

    I think that, as you suggest, Noe's most plausible line is it takes a long time to unlearn more familiar SMK. (This is the Churchland line to which Pete Mandik alludes in my earlier post.)

    Re fake barn cases. This is exactly the issue. Can one learn to discriminate metamers. Take any two things that are metamers under conditions C. Then vary C (a lot or for a long time or for many times) so that the things aren't metamers. Then return to C. It looks like Noe's view predicts that the things will look different.

    Oh, and how about another version of this same problem with an occluded cat:

  3. Hi Ken,
    This is indeed an interesting problem. For me there is a sense in which I perceive the deformed circle but still can't shake the perception of an occluded circle. Saying that it takes time to learn SMK doesn't seem to be much a solution to me.

    I think context is important here and would normally resolve this kind of thing (e.g. pack man in pack man world moving behind an occluding object would I think be less likely to be perceived as a circle). Noe discusses visual perception and context here...

    Still I think this is a real problem to explain, not just for Noe either!