[Cross-posted at Brains]
So much has been going on in nyc that I have had trouble keeping up with it. However I have a bit of free time today and wanted to jot down a few notes about Sid Kouider’s recent presentation at the CUNY CogSci Colloquium. Sid was addressing a topic close to my heart, which the the issue of phenomenological overflow (I am pretty sure his talk was based on his recent TICS paper). In his talk he first gave several arguments against the notion of overflow and then gave his account of what is going on in the cases at issue.
One of his arguments was against the idea of inaccessible phenomenology in general. His idea seemed to be that phenomenology that was inaccessible could not, in principle, make any contribution to the awareness of the individual and so the fact that subjects were able to report that they had rich experiences was some evidence that they had at least some access to the information and so counted against the inaccessibility of those states. Ned Block objected that there was a confusion between something being inaccessible and something being necessarily un-accessed. Something is inaccessible (roughly) when there is no possibility that it could be accessed. Ned admitted that he probably does believe in inaccessible consciousness but was very clear that he does not think that the phenomenological overflow argument relies on this claim. Rather, the phenomenological overflow argument relies on the claim that some phenomenology is necessarily un-accessed at any given time. This is compatible with the claim that it could be accessed at some other time. So, the argument requires only that there is always more to what we are consciously experiencing than we can access at any given time not that there is some conscious experience that is completely inaccessible.
I objected at this point that Sid’s argument seemed to withstand this point. In general I think that this distinction of Ned’s is useful for clearing up a potential confusion about how the argument is supposed to work but it does not abolish the point that Sid was making. In fact, this point has been made, in a slightly different way by David Rosenthal and myself (in the paper linked to above). Subjects report that they see a bunch of letters in the Sperling type cases and that they see all or most of the rectangles in the Sligte type cases so they must have at least partial access to the first-order state. Taking their reports at face value actually counts against the notion of overflow.
A large part of the discussion at this point centered on Sid’s argument that there is an observer effect here that should push us away from thinking about unaccessed phenomenology. His argument seemed to be that any way we could possibly test for it would run afoul of the confound that access was involved (a version of the methodological puzzle). A few people objected that Sid was raising the bar to high here and demanding standards which exceed those of ordinary science. Shouldn’t we avoid the trap of thinking that there is something special about consciousness and accept regular scientific standards of when it is and isn’t around? If so it seems we could overcome the observer effect by accumulating enough ‘circumstantial’ evidence to convince us one way or the other. Dave Chalmers at this point made a comparison to the way we think about tables and chairs. I know that there is a table here because I see it (I access it), and I can’t prove that it is there unless I access it, yet none the less I go on being reasonably confident that the table continues to exist when I am not accessing it. Might not the same be true for consciousness? For my part I think that this kind of thinking is often behind the intuitions of those who endorse phenomenological overflow but I don’t find it very convincing. I agree that I think the table is there when I am not looking at it, but I do not assume that the table is there as it appeared to me! That is, the table as unaccessed (unseen) does not look like the table that I see! The table as unaccessed is, according to our best theories, either a swarm of particles or a local collapse in a wave function, or what have you…this is not how it appears to me. So too, we can agree that the accessed thing is still there without thinking that it is there as it was when I accessed it. The qualitative state is there, but when unaccessed it is not like anything for me to have it and so there is no phenomenology present.
Sid then went on to discuss his partial awareness hypothesis, which amounts to the claim that we can have access at many different levels. So, in the Sperling type cases, the subjects will have access to the semantic meanings of the letters in the cued row and only partial access to the letters in the uncured row. To have partial access is to access a lower stage of the processing hierarchy. For instance, he presented data that showed that subjects in a Stroop-like paradigm that were presented with fake color words (like ‘geren’) would treat them like color words (i.e. exhibit strop interference) only when they expected that there were going to be some color words (strong prior confidence) and also were in conditions where the words were very hard to detect (see his paper for details).
At this point Ned objected that he and Sid agreed on how to interpret the experimental results. For Sid the subjects in Sperling cases have full access to the semantics meaning of the letters in the cued row and only partial access to the letters in the inched rows. Ned pointed out that on his view he thinks that the subjects have a conscious experience of the letters in the cued row and have degraded conscious experience of the others in the uncured rows. He has always said that the phenomenology in the uncured rows is unable to be brought under the correct concepts that would allow the subject to know the identities of the letters. This sounds just like what Sid has said and so it sounds like Sid is endorsing phenomenological overflow. But this is incorrect. Sid has agreed that the subjects have the phenomenology that Ned says that they do, but on Sid’s view they have that conscious experience because they have partial access to the first order states! On Ned’s view they do not have access to those states in any way and still have phenomenology. So Sid has not endorsed overflow of any kind. The subjects have just as much conscious experience as they can access. Since they have access only to low level processing of uncured row their conscious experience will be partial as well, but since they del so confident that they are seeing letters they mistakenly report that they consciously see all of the letters. All of this can be said without having to endorse the notion of phenomenological overflow. All in all then, the no overflow view is the most parsimonious; a conclusion with which I totally agree! :)