Perhaps this reveals the similarity between the visual
systems of animals and neural networks when they try
to make sense of the world, elaborating on common
features, using bits of what they know, and filling in the
blanks. This all suggests that the capacity for halluci-
nating is an intrinsic feature of complex visual systems
and is one more piece of evidence to make us wonder
why we banish human hallucinations to the realm of
the unreal.
HALLUCINATIONS AND THEORIES
OF CONSCIOUSNESS
Does the hard problem seem any worse to you when
you think about a hallucinated golden tunnel of light
rather than an actual yellow-painted underground
walkway with sun shining in at the far end? It should
not do so. In essence, the problem is the same: as
Chalmers puts it, how can physical processes in the
brain give rise to subjective experience? Yet perhaps
the familiarity of thinking about perceiving the ‘real’
world blinds us to the seriousness of the problem,
which may seem more obvious when thinking about
hallucinations. We know (at least roughly) what sort
of cortical activity causes someone to have a potent
hallucination of a bright golden tunnel. But how can
the experience of a yellow tunnel (that throbbing,
pulsating, realistic tunnel sucking me right now into
its golden light) be caused by, or simply be, that neu-
ral activity?
For some theories of consciousness, hallucinations
provide a special stumbling block. For example,
sensorimotor theories entail no pictorial images or
representations inside the head; instead, perceiving
means having mastery of the sensorimotor contin-
gencies between sensory input and motor responses
such as moving your head, blinking, or running your fingers over something to
change the input. This makes imagery and hallucinations a problem because
moving, blinking, or touching them has no effect. O’Regan and Noë (2001) try to
solve this problem by suggesting that knowledge of the contingencies involved
is sufficient for experiencing hallucinations and imagery. In addition, the lack of
feedback explains why imagery and hallucinations are not as accurate or detailed
as direct perceptions.
Other theories use hallucinations in support. Higher-order theories take them as
evidence that one can have a second-order thought (i.e. one can represent to
oneself ) that one is in a state when this is not true. For example, in the visual
release hallucinations of Charles Bonnet syndrome there is higher-order rep-
resentation (for example of a group of little laughing faces) without first-order
representation; so higher-order representation seems to suffice for conscious
FIGURE 14.7 • Compare these two images. One
is psychedelic art, the other is
produced by Google’s DeepDream
program.