Lessons from physics - the independence of vision?

Recently I have been delving into the strange and wonderful world of theoretical physics, by reading Brian Greene's mind-widening book The Fabric of the Cosmos. Some of the ideas (special relativity; time's arrow; quantum uncertainty) were familiar to me from physics classes at secondary school, but back then, through seen-it-all-before teenage eyes, I had never really grasped that what we were learning about was real!

And apart from instilling wonderment, this book has opened my eyes to the parallel perplexities faced by both theoretical physicists and us cognitive scientists. Both fields are trying to know a system (reality; the mind) which refuses to give up its secrets without a little probing and experimentation. And the common thread is that the very act of probing destroys what one is trying to measure. In physics this is known as Heisenberg's uncertainty principle, which states that you can never know certain pairs of physical properties, such as the position and velocity of a particle, at the same time: measuring one peturbs the system and prevents you measuring the other. This is similar to the problem faced when  measuring conscious states and their associated neural correlates. To know whether someone is conscious of, for example, a dim light, we need to elicit a report, which itself fundamentally changes the consciousness (and attendant brain states) of the observer.

This tension I think underpins the ongoing debate over whether phenomenal visual consciousness can be separated from our "access" to that consciousness, as proposed by Ned Block. This is a fundamental and fascinating question, and one that cuts to the heart of the mind-brain relationship. Block (1995; 2007) states that phenomenal (P) consciousness is indeed independent of cognitive access to it, in the sense that we can have private visual experiences without necessarily being able to report these facts to others. He argues that a “mesh” between psychology and neuroscience – an inference to the best explanation – allows us to establish beyond reasonable doubt whether phenomenal consciousness (at “the back of the head”) requires access to be conscious. He focuses on the Landman et al. (2003) study, which found that in response to a secondary cue, subjects could only report part of an 8 object visual array, despite claiming phenomenal consciousness of the whole scene. Importantly, different cues gave access to different parts of the array, thus at some level the information is indeed “there”, ready to be accessed. Block claims that this shows that the information is phenomenally conscious at this early stage. However, epistemic correlationism would say that every piece of data requires a “correlating” report – how else are we to really know what the subject is doing, seeing, or thinking? If this is the case, then we cannot claim for sure that there is phenomenology without access, as it is not scientifically tractable. Again, back to physics: it is rather like Schrodinger’s cat – the cat theoretically might be dead, or it theoretically might be alive, but to know you have to open the box. As for the cat, as for epistemic correlationism in the neurosciences – the only way to know is to open the box and ask the subject.

Block goes on to argue that there is a candidate mechanism for phenomenology in the brain – weak “coalitions” of neural activity in the back of the head (early visual cortex – see e.g. studies by Dehaene on graded conscious processing) that do not get broadcast to frontal areas, and are therefore unreportable. But this logic assumes a priori that phenomenology without some sort of access can exist – if we assume for a moment that it cannot, then the fact that these coalitions are not broadcast (cognitively accessed) would preclude them from being phenomenally conscious. Without these strong assumptions, we cannot know the subjective correlates of these patterns of neural activity. To return to Schrodinger’s cat: imagine that we could perform a scan on the box which would give us some image or pattern to look at. If we assume the cat is alive, we then might be able to say, yes, this assumption would correspond nicely with the fact that the image has such-and-such properties. If the cat is dead when we subsequently open the box, do we have enough evidence to say that the cat was alive at the time we took the image? Of course not. The inference that the cat was alive rests exclusively on the a priori assumption that the cat is assumed to be alive. In the same way, the inference that losing coalitions of visual cortical activity are phenomenally conscious rests on the a priori assumption that the subject is having a phenomenally conscious experience at the time they are recorded.