Attention and Mental Paint

(cross-posted at Philosophy Sucks!)
The NYU Mind and Language seminar has started up again with a really excellent line up. I hope to blog about them as they happen…
Last Monday I attended Ned Block’s session on his paper Attention and Mental Paint. I have talked about an earlier version of this before. The basic idea comes from figures like figure A below. if one fixates (stares) ate the center and, while keeping one’s gaze fixed, moves one’s attention to an individual disk that disk will appear darker. After a bit of practice one can darken any disk one wants by moving one’s attention around. Go ahead, give it a try!

Recently a psychologist, Marisa Carrasco, has run experiments trying to quantify this effect. Below is a reproduction of the stimulus. Here the two patches differ in contrast by 8% yet when one fixates on the center and attends to the 22% patch one will judge it to be the same contrast as the 28% patch.

Block wants to use these findings as the basis for an argument against both direct realism and representationism. The basic argument goes as follows.

  • First we focus only on two cases. The first case is when we fixate and attend to the center. In that condition subjects get the judgment about contrast right (i.e. they judge that the left patch is lower in contrast). In the second condition we fixate on the center but attend to the left patch. In that condition people get the judgment incorrect (i.e. they judge the two patches to be the same contrast.
  • The second step is his claim that there is no reason to think that either of the two cases above are illusionary. Both are veridical.
  • If both experiences are veridical then the thing that they are experiences of must differ in some property but all of the properties of the objects are the same. The only thing that has changed is that one has moved one’s attention from center to right.
  • Therefore there must be mental paint, or non-representational features to our experience (the anti-representational conclusion) or a mental aspect of mental experience (the anti-direct realism conclusion)

a lot of the discussion at the session focused on whether or not there was an illusion at work here. Block claims that in both cases talked about above the perception is veridical. Why? His idea is that both of the experiences play the same functional role and so are accurate. The pro-illusion folk (Jesse Prinz was in this camp) argued that when you attend to something you represent that thing more veridically and so the condition where one fixates and attends to the center is illusionary (Jesse preferred ‘distorted’). Block protested that one could just as well say that attention distorted, or magnified, the scene and so the fact that one has access to more information when one attends is not by itself an argument that the experience is more veridical. Some other issues came up about various responses direct realists or representationalism could make.

However, I am less interested in that issue as I am in the issue of whether there is an argument here against anything like the kind of higher-order thought theory that I am fond of (i.e. one very much like David Rosenthal’s). On this kind of view we have two distinct kind of mental representations. At the first-order level we have the mental states that represent the sensible qualities. So, when I am seeing red I am in a mental state that has a property, call it red*, the represents physical red. However the kind of representation that is going one here is not intentional or conceptual. It is homomorphic. Red* is the property which is related to green* and pink* in a way that mirrors the relations between physical red, physical green, and physical pink. The starred properties can occur both consciously and unconsciously. When they occur unconsciously there is nothing that it is like for the organism in which they occur. They become conscious when I am aware of myself as being in a red* state. According to Rosenthal I do this by having a thought which deploys the concepts Red. So on this view the higher-order thought is representational in the traditional sense and it is the thing which is responsible for the phenomenology of the experience.

So is there mental pain on this view? Well, as long as one agrees that there can be unconscious sensory states with no phenomenology (a big step!) then Rosenthal’s first-order sensory qualities will count as mental paint. They are not intentional and they are mental. But they do not play a role in determining the phenomenology (except in the sense that we get the concepts we deploy in the higher-order thought from them) and so if we restrict ourselves only to conscious experiences it does look like Rosenthal denies the existence of mental paint. A conscious experience of red is constituted by a ‘I am seeing red’ thought which is completely intentional/representational. So then does Ned’s argument cut any ice against Rosenthal’s account of conscious experience?

It is not clear that it does. Ned’s argument gets its force from the claim that the thing being represented would have to be different because all conscious experiences are or represent just the actual properties that the object actually has. But on Rosenthal’s view we have the first and second order mental states. So, one could hold that there is some change in the first-order representation of the two patches or one could hold that the first-order representations are the same in the two cases and what changes is the way in which we are conscious of them. In talking briefly about this with David he seems to think that attention changes the first-order state whereas I seemed to think it was teh content of the higher-order state which changed. But since we are talking about conscious experiences here and it is the higher-order state that accounts for the conscious phenomenology the difference has to be in the way that we are conscious of the patch in the two cases…this may be because of attention in a causal sense but it is the content of the higher-order state that has to account for the difference in the phenomenology.

16 thoughts on “Attention and Mental Paint”

  1. The brightening of the disks that are the target of attention is consistent with the finding that *neuronal excitation* in the brain’s representation of egocentric space is actually increased in the spatiotopic parietal coordinates of the target of attention. The structure and dynamics of the heuristic self-locus in the retinoid model explain how this happens. See Ch. 4, in *The Cognitive Brain* “Modeling the World, Locating the Self, and Selective Attention : The Retinoid System.”

    http://people.umass.edu/trehub/thecognitivebrain/chapter4.pdf

  2. I don’t see why someone would reject representationalism based on these results. We can have two (quite) different spectral profiles that look the same (metamers). How is this different? Just because attention changes the appearance doesn’t seem particularly damaging. It would seem equally damaging to point out that shining a flashlight on an object  makes it look different than an object in the dark.

    What is the difference between a mental represenatation and mental paint? A painting is a representation, after all.

    More interesting would be implications for the modularity thesis. This seems like clear top-down effects on something modularity-philes would assume is modular.

  3. In the flashlight case we have an actual physical property that differs between the two patches. One of them is actually brighter than the other. In the attention case the two physical patches are unchanged yet there is a phenomenal difference. This is supposed to show that there can be a phenomenal difference when there are no properties of the objects that are different. This in turn is supposed to make trouble for anyone who thinks that phenomenal properties just are the represented properties that objects actually have (recall the transparency intuition. We are supposed to ‘see through’ the representation to the object).

    Mental paint, for Block is a kind of representation; mental red represents red (one implication of the view he defends is that many different ‘mental paint chips’ can all veridically represent the same physical property which is odd). However it is not representational in the intentional sense, truth-evaulable sense.  So a painting has this kind of representational content but the actual color of the paint doesn’t. Mental paint, for Ned, just is the phenomenally conscious color experience that is mental, non-conceptual and non-intentional (he thinks this is identical to a brain state).
    I don’t really know much about the modularity wars so I’ll have to take the 5th on that one…
  4. Hi Eric,

    Not sticking up for Ned here, just spelling out his view:

    When you look at the mona lisa, you can be aware, in some sense, of the woman and the smile (how friendly she looks!). Alternately, you can be aware of the paint (a little bumpy there, kinda reddish ochre too!). One might say that the woman and the smile are the content whereas the paint is the vehicle. Ned’s thing about mental paint might be put by saying he thinks that one can be aware of the vehicular properties of mental representations, not just their contents.

    Dig?

  5. In the flashlight case we have an actual physical property that differs between the two patches. One of them is actually brighter than the other. In the attention case the two physical patches are unchanged yet there is a phenomenal difference. […] This is supposed to show that there can be a phenomenal difference when there are no properties of the objects that are different.

    What representationalist would not absorb this in a trivial way? The flashlight example hooked into the spotlight of attention metaphor to make a certain point.

    Let me spell it out with a real-world case rather than a metaphor. Let’s say we go out at night in the desert. You see a dozen or so stars. In a half hour, after you are dark-adapted, the stimulus is the same, but you now see a sky full of hundreds of stars. So, obviously ‘there can be a phenomenal difference when there are no properties of the objects that are different.’ This is (partly) because our brain states (e.g., state of our photoreceptors, attention, etc) change, and this influences how the brain represents the world.

    Mental paint, for Block is a kind of representation; mental red represents red (one implication of the view he defends is that many different ‘mental paint chips’ can all veridically represent the same physical property which is odd). […] Mental paint, for Ned, just is the phenomenally conscious color experience that is mental, non-conceptual and non-intentional…

    So perhaps it would be more accurate to say he is a representationalist, but thinks it is a nonconceptual sort of representation which is captured by this term ‘paint’? That’s fine with me. It is the claim that such psychological data refute representationalism that bugged me.

  6. Pete: That is interesting. My first reaction is to say there are different levels at work. Wrt the content ‘mona lisa’ the color is a vehicle. Wrt the content ‘photons hitting the viewer’s eye from her skin in the depicted scene’ the paint is content. That’s just my first impression I’d have to think about this a bit more.

  7. It seems to me that my immediate experience of the Mona Lisa is a global non-reflective phenomenal experience. Then, I might have phenomenal experiences of an analytic kind — smile, serenity, attractiveness, pigmentation, etc. I might also think about the paint as the medium (not the vehicle) for the painting. All of these are transparent phenomenal experiences in the sense that the causal mechanisms for their occurrence are not directly experienced (we “look through them”). From the perspective of a cognitive neuroscientist, the vehicle for all of these phenomenal experiences is the neuronal state of the causal brain mechanisms that constitute the particular phenomena wrt the Mona Lisa.

  8. In the real-world case you talk about the first percept is (arguably) not veridical while the second one is. Block argues that in the above cases both percepts are veridical and so the kind of move you want to make isn’t available (Pete and I are currently discussing this over at Philosophy Sucks! if you want to join in). 

    As for whether Block should call himself a representationalist; I think it would cause more problems than it would solve…
  9. In the real-world case you talk about the first percept is
    (arguably) not veridical while the second one is. Block argues that in
    the above cases both percepts are veridical and so the kind of move you
    want to make isn’t available

    I wouldn’t say it isn’t veridical, as not seeing a star seems an accurate indicator that
    there are only X suprathreshold stimuli in the sky. However, rather than
    get bogged with such a quibble let me tweak the example.

    Consider a different situation in which during dark adaptation you
    don’t begin to see new stars, but the same stars you saw, which
    originally were very dim looking, appear brighter (same shape, same
    number of stars, but brighter). This is much like the attentional
    effects in Block’s story.

    I just don’t see what is interesting here. Of course
    transducers alter features of representations that are triggered in
    response to the same stimulus. Switching from rod vision to cone vision
    (which takes about a half hour) produces substantive changes in our
    experience.

    Similarly, if you look at research on attention, it has drastic
    effects on firing rate and (possibly) synchronous activity among
    neurons that represent things in the attended region. Why wouldn’t we
    expect this to change the threshold of responses to stimuli in said regions, or their
    apparent brightness, or the number of bits coming in from that region? It’s a no-brainer.

    Brightness, in particular, seems susceptible to such effects, because of the interesting brightness adaptation effects in our visual system. Such effects seem to generate not illusions but are adaptive in terms of improving information transmission (e.g., Simon Laughlin’s work on the utility of brightness adaptation in retinae).

    These scenarios (dark adaptation, attention) are straightforwardly describable as
    neuronal effects on representational systems, not a refutation of
    representationalism.  Perhaps it a refutation of a straw
    representationalism in which the same stimulus must always produce the
    same appearances. I’m not sure, in this age of neuroscientifically
    sophisticated philosophers, who would believe that.

    I must be missing some of the steps in the enthymeme, as Block is a smart guy.

  10. Again, in the cases you talk about there is an actual property <em>of the object being represented</em> which we come to have a more accurate ‘fix’ on. So you are right that in those kind of cases there is no threat. But the attention cases is different.  What is the corresponding property of the represented object (i.e. the Garber patch) that we are representing? Both experiences are veridical so both must be (according to the representationalist) representing actual properties of the Garber patch. 

    Block will agree with you that attention has all of those neuronal effects, some of which just are the phenomenal experiences in question but he denies that those things –the conscious phenomenal experiences– are representations, in the sense of having accuracy conditions. He thinks this because the attentional effects show that you can have two identical representations that differ in phenomenal character. 
  11. Richard: No, my tweaked adaptation scenario can’t be described as ‘getting a more accurate fix’ on the stars. I stipulated that you see the exact same stars after adaptation, but they just look brighter. I constructed it precisely to be a case where we have two different experiences but the same “representation” (i.e., the exact same stimulus). Hence, he should agree that this helps makes his case against representationalism, or if not that there is something wrong with his general argument.

    Indeed, it seems my tweaked example should be better than Block’s attention example, as in my
    example we can (reasonably) stipulate that the same number of bits are flowing
    through the visual system as the visual system adapts. In Block’s attention example we would expect that more information is getting in from the attended
    region because of the attentional “eyeglasses” (see below).

    On defining representations, it is useful that you mention ‘accuracy conditions’ as a necessary component of his concept of a representation. That doesn’t seem to help him. If I look at a movie with my glasses on versus off, I expect things to look different. Attention is like adding an additional filter to our brain, like adding intracerebral glasses instead of real glasses (again the glasses thing is a metaphor please don’t get all literal on my ass).

    In each case (glasses / attention) I’m representing the same things with different degrees of accuracy, different amounts of information flowing through my visual system (or through a certain part of my visual system in the case of attention). In both cases it would be a mistake to talk about one case as veridical the other as not veridical (he admits as much, so the two examples pace each other quite well). We (neuroscientists) would quantify the difference according to differences in the amount of information about the source making it through the (neuronal or eyeglass) filter. In both cases, I’d expect a phenomenal difference to track the informational changes.

    Block should be commended for looking into the psychology, but probably ought to follow things into the brain a bit more and think about how to quantify statements about representations. In the attention case the representational properties (quantified using information theory, signal detection theory, or some such) are not going to be the same. Using my dark adaptation example he could stipulate that even these quantitative measure of information flow are the same in order to see if this stipulation buys him anything useful in his arguments against representationalism (note I already addressed that case in my last post).

    Anyhoo, don’t see the hubub over his argument. It seems a fairly obviously flawed line of thought so I must be missing something. No need to get all HOT and bothered in response to his argument. Just look at the neuroscience.

  12. After hearing this I don’t think that your kind of view is the kind that the argument is supposed to pose a problem for. Block wants to present a problem for people, like Michael Tye and Fred Dretske for instance, who think that when we have a veridical conscious experience we are representing the actual properties that the physical objects truely have. Neuroscientists may not talk like that but there are people who do and this is supposed to be an argument against them. 
  13. A decisive counterexample to the claim that a “veridical” conscious experience represents the actual properties “that the physical objects truly have” is given on pp. 29-32 in Trehub, A (2007) Space, self, and the theater of consciousness. *Consciousness and Cognition*. Here, my seeing-more-than-is-there (SMTT) experiment demonstrates that two tiny line segments in vertical oscillation (the physical object) induce a vivid conscious experience of a triangle in horizontal motion.

    See: http://www.theassc.org/files/assc/SSTC_Preprint_edited.pdf

  14. Arnold: I think in all cases of illusions they would admit things aren’t represented veridically (e.g., Muller Lyer, moon illusion, etc), but that it is the brain’s attempt to get it right, and they would say in nonillusion cases it does get it right.

  15. Yes, it is tricky. I wasn’t endorsing their view, just trying to articulate it. I’m not even sure what it means to say that our color experience is meant to represent something real in the object.  It seems wrong to me.

Comments are closed.