Popping Out Versus Filling In

Although the details differ, Clark shares Block’s claim of the existence of awareness/attention-free phenomenal properties (Clark 2006: 9-10). In addition, Clark – again, like Block – attempts to utilize empirical data to bolster said claim. One example utilized by Clark is that of the “popping out” effect revealed by such visual curiosities as the Kanizsa square (Ibid. 12). What follows is an attempt to undermine Clark’s case based upon the notion that what Clark calls a “pop out” effect in the visual processing of the Kanizsa square is actually a “filling in.” Moreover, “filling in” does not show what Clark claims “popping out” does.

Clark’s case, in short, is as follows. Empirical data supports the notion that vision is a serial process resulting from the combination of distinct/modular systems. The detection of features such as “edge level” occur previous to reportable awareness (Ibid. 5). That the lines of a square are reportable in the Kanizsa square where no such lines actually exist, but are nonetheless reportable, is according to Clark an example of pre-awareness phenomenal properties. This “popping out” “indicates a special kind of ‘preattentive’ processing” (Ibid. 3). The ability to report the awareness of the line is an “example of good, clean, traditional talk of phenomenal properties” (Ibid. 13). Thus, if attention is necessarily present along with the occurrence of phenomenal properties, then, based upon the empirical evidence of the visual system, humans ought not be able to report being able to see the lines of the Kanizsa square.

The empirical data, however, does not demonstrate what Clark thinks it does. “Popping out” in such cases as the Kanizsa square demonstrates that the visual system of humans can “fill in” their phenomenally laden awareness without the stimuli in the world that corresponds to the reported experience. In the case of the Kanizsa square, the lines of a square are “filled in” where no such lines exist in the world. One need not look to such visual curiosities as Kanizsa squares to find evidence of “filling in.” During the entire life of an awake human with normal visual functions is the “filling in” present in the form of the filled in blind spot. Humans do not have a phenomenal free zone where the optic nerve connects to the eyeball, and yet phenomenal properties are there to be reported. What Clark calls the “pop out” effect is actually “filling in.” The appearance of the lines of the Kanizsa square lends support to the idea that humans can “fill in” their visual field, i.e., have awareness with phenomenal properties, without external stimuli, and not that humans can have phenomenal properties without awareness.

Blindsight Video

Greetings,

The following link is to a video of a subject with blindsight navigating around obstacles:

http://www.scientificamerican.com/video.cfm?lineup=1406165298&id=79566940001

I thought it helpful to contemplate Block's notion of phenomenal consciousness overflowing cognitive accessibility in relation to a case such as the one demonstrated in the video.

Enjoy.

Passing the Buck to Many Little Men

Baars presents the Global Workspace (GW) as an empirically testable model of consciousness. Moreover, Baars presents the GW as an “expansion” of Crick’s research on visual attention and the “searchlight of attention” metaphor (Baars 2003: 1114 and 1118). Baars, pace Crick, believes that the GW overcomes philosophical criticisms such as those that accuse other theories of consciousness of being merely another form of the “Cartesian theatre” (Ibid. 1114). Theories of consciousness that are accused of being Cartesian theatres are found lacking because they posit a single point in the brain/mind where information, perception, representations, etc. come together. This is problematic because it suggests that there is a single perceiver in the brain/mind for whom said data is presented. Often, this perceiver is called the homunculus, and the fallacy is so too named. Cartesian theatre and homunculus-like explanations of consciousness are problematic because they fall into a regress: “Well, if conscious experience is for the audience in the theater or for the homunculus, then how do we explain how the audience or homunculus experiences a unified conscious experience?”

Baars’ response to the above criticism is as follows. The GW is not a Cartesian theater and there is no homunculus because, “In the Global Workspace theory the single homunculus is replaced by a large ‘audience of experts,’” whereby there is, "no fixed, subordinate observer. Individual modules can pay as much or as little attention as suits them, based upon their particular expertise” (Newman, Baars, and Cho 2003: 1132). One need not be a nitpicking philosopher or an expert in logic to realize an obvious flaw in Baars’ reply to the above criticisms. In positing an “audience of experts” as a solution to the Cartesian theatre and/or homunculus challenges, Baars is merely shifting the burden of explanation from one audience/homunculus to many members of an audience/expert homunculi. In other words, Baars is passing the buck to many little men. Instead of explaining how one little man in the head perceives a “bound” perceptual experience, Baars’ GW now has the burden of explaining how each of the little expert men do their perceiving, and how their separate perceptions are unified into a single conscious experience.

Baars explains that useful metaphors in science “should help organize existing evidence, yield testable hypotheses and suggest conceptual clarifications” (p. 1115). Baars believes the theater metaphor, characteristic of the global workspace (GW) model fulfills these requirements. I think the theater metaphor, particularly the role of the director, however, is somewhat confusing and not entirely helpful in understanding attention and consciousness. More specifically, I think Baar’s theater metaphor fails to capture an aspect of attention and consciousness, and it also fails to accurately capture the concept if a stage director

The spotlight metaphor is prima facie appealing. It is a metaphor commonly used (see Gopnik’s reply to Block for an example) and is an intuitively appealing description of attention. When you attend to something, it is as though the object of attention is highlighted. The stage metaphor incorporates this particular sub-metaphor well and, additionally, highlights the cooperative (or even competitive) nature of brain function.

That being said, I think the stage metaphor, in general, suggests a counter-intuitive view of attention. Stage productions tend to be highly planned. Each lighting cue is worked out in advance, each exit and each entrance is timed and rehearsed (this is not always the case, but it often is). Consciousness and attention appear to be much more dynamic. The “spotlight” can switch suddenly from one object (a book) to another one (a loud noise) for a variety of reasons.

This dissimilarity is further demonstrated in Baars’ discussion of the stage director. He explains that the “left-brain “narrative-interpreter” receives its own sensory inflow from the right visual field, so that it “observes” a conscious flow of visual information” (p. 1119). He continues saying that “each interpretive system can control its own voluntary motor functions and thus there is an obvious analogy with a stage director, who observes events on stage and orders changes where needed” (p. 1119-1120). The director, though, does a great deal of his work in advance. The director guides a production (in various ways) to meet his particular vision of the play. This would include deciding what the lighting would be like and so on. The situation Baars is describing sounds more like an orchestra conductor or even a head coach or manager whose role is more dynamic and unfolds in real time.

These ambiguities and disanalogies make it difficult to imagine exactly what determines where the spotlight is directed. Does the audience of experts decide or does the director, who is perhaps more like a conductor, direct the spotlight to whatever the audience might find interesting or important? I think the theater metaphor needs some modification (some modifications are made in Newman et al.) to be useful in investigating attention and consciousness.

Premature Speculation

Block is premature in arguing that the allegedly relevant empirical data shows that phenomenal consciousness overflows cognitive accessibility (481). Although they state that they agree with Block’s point, Koch and Tsuchiya undermine a key portion of Block’s chosen empirical data (509). Koch and Tsuchiya cite Block’s utilization of fMRI data, particularly data on the fusiform face area, which is alleged to demonstrate phenomenal consciousness without cognitive access (510). Koch and Tsuchiya object to this evidence on the grounds that “vigorous” fMRI signals are not enough to account for consciousness. As they note, the relationship among the signals measured by fMRI is “very complex” (510). At most, activity in the fusiform face area is correlated with conscious face recognition. Moreover, even Block notes that, “Mere correlation is too weak” and that, “At a minimum, one wants the neural underpinnings of a match of content between the mental and neural state” (483).

Perhaps, it can be argued, Block’s point remains potent even if the fMRI data on the fusiform face area is speculative at best. Unfortunately for Block, if data such as the aforementioned are undermined, then so is his overall case. Block is counting on the empirical data to support his distinction between phenomenal and access consciousness. It is far too early in the neuroscience game to make such strong claims as Block’s that phenomenal consciousness overflows cognitive access. As Koch and Tsuchiya’s criticism of the fusiform face area fMRI data demonstrate, Block’s utilization of his chosen empirical data overextends said data’s power of explanation. When, or if, a mature neuroscience of consciousness comes to fruition, that is, one that does not merely correlate phenomenology with brute neural activity but posits clear and evidenced explanations, then such hypotheses as overflowing phenomenal consciousness can compellingly be tested. Until then, Block is exploiting data that is correlative at best or just plain wrong at worst.

Phenomenology and Workspace Interaction

I question Block’s final step towards establishing the modular independence of phenomenal consciousness from workspace activity. He relies on insufficient behavioral data to make a conclusion about an underlying neural mechanism. His argument goes something like this: (1) There is evidence suggesting that workspace is located in the prefrontal cortex; (2) It is established that visual experience correlates with activity in the back of the head; (3) Computational models seem to represent well the interactions between these two areas suggesting that there is competition between coalitions of neurons in the back of the head for the activating of coalitions of neurons in the front of the head. There also seems to be competition for dominance in the frontal area and feedback signaling that may affect the success of given coalitions in the back; (4) Sergent & Dehaene’s 2004 experiments show that subjects tend to ignore completely one of a couple of target stimuli when they have already detected the first; (5) Therefore even though there were two coalitions of neurons that were active in the back of the head only one succeeded in the activating of a coalition of neurons in the front of the head. This is to say that even though the subjects had phenomenal experience of two stimuli they only had cognitive access to one of them. As I see it (5) does not follow from the premises. Block ignores an alternative interpretation of the data that is consistent with what is assumed in (3). One of the options suggested by the computational model is that there is a feedback mechanism from the front to the back of the head. Given that the subjects tended to report having seen the first stimulus and not the second, it is possible that once a perception was detected in the front of the head, signals were sent to the back of the head that blocked perception and not just attention to the second target stimulus.

Attention Deficit

Block argues that, with respect to attention and phenomenology, “neither is necessary for the other” (p. 540). For attention without phenomenology, he cites a study (Jiang et al. 2006) showing that subjects are drawn by nudes of the opposite sex in “conditions in which the nudes are invisible” (p. 540). Block adds that attention, in this experiment, was gauged by a “subsequent task to which distribution of attention is crucial” (p. 540). To show phenomenology without attention, he notes that subjects in some experiments, as in the work by Alvarez and Oliva, can see “scene gists”, like the presence of a face in their peripheral vision despite “maximal siphoning off of attention by a very demanding task” (p. 540-541).

Block believes these and similar experiments counter Prinz, Lycan and many others who believe that attention is necessary for phenomenology. I am not entirely convinced, however. One important issue is the definition of attention. In his reply, and even in the target article, the term is used somewhat loosely. In the Jiang et al. experiment discussion, it is not completely clear what is meant by attention—it certainly does not match a typical understanding of the term. If one does not notice the nudes or actually study them then it is difficult to say one is attending to them ordinarily speaking. Block might be helped by drawing on some of what Alison Gopnik discusses. She distinguishes between exogenous and endogenous attention. Exogenous attention is characterized by “eye movements, decelerating heart rate, and parietal activation” while endogenous is goal directed, top-down attention (p. 503). She compares exogenous attention to lantern light and endogenous attention to a spotlight.

Block does not specify what type of attention he is talking about, but it appears he is referring to exogenous attention in the Jiang et al. experiment. In another experiment (Volpe 1979; Verfaeille 1995), which he uses to show that, against Lycan, people can have representation without attention, he claims that there is unconscious seeing (p. 540). In this experiment he says that subjects are not attending to what they actually see. In the nudes experiment, Block says that the subjects’ attention is drawn by the invisible nudes. Here he seems to be relying on an exogenous notion of attention while in the unconscious sight experiment he seems to be employing a more endogenous version of attention. If he were to apply the same, broader notion of attention in both discussions the subjects in the unconscious sight experiment could be said to be attending as well. This would mean that his empirical evidence does not tell against the view of Lycan, Prinz and others without further modification.

I would have to review the experiments to see if the role of attention is clearly defined (especially in the unconscious sight experiment), but it seems that Block is equivocating between two types of attention in his replies to Prinz, Lycan and other responders in reply 5.

Do fMRI results show what they claim to show?

We talked briefly in class about some potential problems with relying on fMRI results. Here's one such study I've come across. No doubt the neuroscientists among us know others. I'm not sure exactly which study Luis was referring to yesterday, but it might be this one (here's a link to the study itself). I take it both studies could be consistent with one another. My understanding is the former claims that sometimes blood flows to brain areas in anticipation of brain activity that might or might not occur, meaning that researchers could be mislead about whether or not the haemodynamic signals the MRI picks us indicates brain activity. Actually, many of the google scholar citations for the former study look interesting.

UPDATE: You might also find this article by Tyler Burge interesting and an introduction to some of the issues we will be discussing.