Friday, 19 May, 2000
The new Cognitive Neurosciences 2nd ed pp 1285-
focusing on the nuronal correlate of visual awareness in the primate brain � seek understanding of function and the specific neuronal structres underlying the phenomenology of conscious vision � experimental evidence, especially electrophysiological data in the awake macaque monkey
assume that consciousness is explicable and a shared phenomenon
major neuroscience question = which are the active neuronal processes in our heads which correlate with consciousness, and what is the difference between them and the others which do not
assumption: the different aspects of consciousness (pain, visual awareness, self-consciousness etc.) = basic common mechanism(s)
no precise definition of consciousness (will probably include attention and STM)
assume that some animals possess some of the features of consciousness
\ language is not essential for consciousness (assuming that they don�t have language, and that it isn�t the transforming factor between their and our intelligence), though it may enrich it
easiest scientific approach = visual consciousness
visual percepts = especially vivid and rich in information
primate visual system is similar to ours
many experiments already on visual systems (e.g. macaques)
biological usefulness of visual consciousness in humans = produce the best current interpretation of the visual scene
using past experience of ourselves, or ancestors in genes
make this interpretation directly available (for sufficient time) to the voluntary motor output (incl speech) parts of the brain
zombie = acts just as normal people do, but is completely unconscious
seems an untenable scientific idea
but in some cases: visual input �/span> relevant motor output, without being able to say what was seen
what is the benefit of conscious mental life over a suite of such specialised zombie modules?
e.g. frog�s 2 independent systems for action (Milner & Goodale, 1995)
1. snap at small, prey-like objects
2. jumping away from large, looming ones
an arrangement of many zombie systems would be less efficient than consciousness
better: single complex representation available for choosing between different plans for action = seeing
Milner & Goodale � primates 2 systems:
1. on-line = faster, unconscious (e.g. trained tennis player reacting to a fast serve without seeing it; sprinter starting to run before consciously hearing the pistol)
2. seeing = conscious
to be aware of an object/event:
the brain has to construct a multilevel, explicit, symbolic interpretation of part of the visual scene
multilevel� = in psychological terms
different levels, e.g. those that correpond to lines/eyes/faces
������������ = in neurological terms
the different levels in the visual hierarchy (Felleman & Van Essen, 1991)
explicit = a smallish (probably <1000) group of neurons that employ coarse coding (Ballard, Hinton & Sejnowski, 1983) to represent some aspect of the visual scene
e.g. neurons explicitly representing a particular face might still fire to face-like objects
a set of these neurons will be of the same type, probably fairly close together and all projecting to roughly the same place
if all such groups in a stack were destroyed, then a person would not see a face (though may be able to see the parts of a face)
other parts of the brain explicitly represent other aspects of a face, e.g. emotion or gaze
although the information needed to represent a face is contained in the firing of retinal ganglion cells, they lack an explicit representation for faces
will consist of many relevant aspects of it
likely to be fairly distributed over the visual system
how are they bound together = the �binding problem� (von der Malsburg, 1995)
constructing a representation requires a great deal of (probably unconscious) neural activity
the unconscious activity = the computations needed to find the best interpretation
the interpretation itself = the results of the computation
we become aware of some of these: winner-take-all
only some types of specific neurons will express the NCC
e.g. the firing of many cortical cells does not correspond to what the animal is currently seeing
= a �local� theory of consciousness
emphasises the role of specific neuronal/cognitive subsystems in consciousness
though with potentially global effects
the conscious visual representation = probably distributed >1 area of the cortex (+ possibly sub-cortically too)
but not in V1 (although crucial for vivid visual awareness, V1 neural activity is not correlated with what is seen)
the NCC is necessarily global
e.g. an emergent property of very large + diffuse assemblies of neurons (Libet, 1993; Popper & Eccles, 1981; Sperry, 1969)
consciousness cannot be located at the level of single neurons
�visual consciousness� = covers a variety of processes
recollections are less vivid usually than live visual input
some form of iconic memory seems essential for consciousness (transient, lasting < second) (Coltheart, 1983; Gegenfurtner & Sperling, 1993)
if we don�t pay attention to a part/aspect of the visual scene, our (STM) memory of it is transient and can be overwritten (masked)
why is it so surely probable that there are specific active, neuronal processes correlating with consciousness?
why is representation of some aspect of a visual scene by a smallish group of coarse-coding neurons �explicit�?
what does it mean to say that a person will not be able to �see a face�?
why don�t the retinal ganglion cells constitute an explicit representation?
is it because they don�t code for the face exclusively, but whatever falls into their receptive field (so it�s not an aspect but an area of the visual scene)?
is the NCC = the particular property of the arrangement of the neurons, or of the neurons themselves, which gives rise to the phenomenon of consciousness?
so once known, this property would presumably be realisable in some other form?
is Koch/Crick�s theory still a working hypothesis?