Intermediate Level Theory of ConsciousnessIntermediate Level Theory of Consciousness
More specifically, we can investigate visual object recognition. Though Marr’s (1982) treatment of
this process is somewhat out of date, it can still provide clear answers to our questions. The idea
is to understand the representations that play a part in his construal of this particular process,
and then locate them on a low, intermediate, and high scale. When we locate the representations
that are where consciousness arises, we can then move on to the big question of what happens
to them to make them conscious.
At the beginning stages of visual object recognition, the primary visual cortex (V1) receives
stimulation from the retina and thalamus to form a conglomerate of mental representations of
very local features. V1 houses cells that represent edges, lines, and vertices that pop up in very
specific areas of the visual field. The composite of these features yields what Prinz helpfully
calls something approaching a two-dimensional pixel map. This map only gives information
about these very specific properties of very specific locations in visual space. Marr calls this the
‘primal sketch’. The representations that make up the primal sketch are low level representa-
tions. Next comes the processing that yields the ‘2.5 D sketch,’ where figure is separated from
ground, surface information is calculated, and a coherent, vantage point-dependent represen-
tation is formed of the object(s) found in the visual field. The representations of items here
are considered intermediate level representations. Finally, the last stage of processing involves a
3D representation of an object, which is formed by a collection of basic shapes like cubes and
cylinders. These representations are supposed to represent their objects regardless of perspec-
tive. Their main job is for classification and categorization. The 2.5 D sketch is no good for
categorization because a shift in perspective yields different representations. There is nothing
invariant in the intermediate processing, then, to classify objects as the same through shifts in
vantage point. Hence the need for these high level representations to complete visual object
recognition.
The details of Marr’s account have largely stood up to the test of time. Lower level visual
areas are perhaps more capable than he realized, there are a wealth of areas that independently
process particular features at the 2.5 D level (including color, form, and motion), and the higher-
level primitives he hypothesized have not been confirmed. Nevertheless, there is much neuro-
logical and psychological evidence to back up his story. Where does consciousness seem to arise
in this process, however? Certainly not at the 2-D, pixel map stage. We are conscious of inte-
grated wholes, distinct from their backgrounds. Certainly, also not the perspective-independent,
3-D representational stage. We experience objects from a particular perspective; when they, or
we, move, our visual experience of those objects changes. The answer to our question seems to
be: consciousness arises at the intermediate level.
Is there any evidence to substantiate this story besides the loose allusion to Marr’s fairly
successful theory of visual object recognition? Prinz cites three lines of evidence, which are
generally used to support all theories about information processing in the brain: cell record-
ings from neurons in the brains of monkeys, fMRI studies about neural activation in humans,
and neuropsychological studies from human patients with brain damage. Beginning with the
first line, recordings from cells in the brain areas that correspond to the distinct levels show
that only intermediate level neurons fire reliably in response to what monkeys are conscious
of. Lower level cells, located in V1, for example, fire in response to two distinct colors that are
presented rapidly in succession, despite it being well known that such a presentation of colors
is experienced as a mixture of the colors. V1 cells also show no activity during color afterimages.
Higher-level cells show the same lack of sensitivity to what monkeys are consciously experienc-
ing. Cells in the inferotemporal cortex, those found at the back end of one of the major visual
streams in the brain, respond to the same objects regardless of size, orientation, position, and left/
right reversal. Evidently changes in these parameters make a huge difference to our conscious