Chapter
Three
The grand illusion
what sort of, information they claim is retained, for how long, and what is done
with it (Simons, 2000).
Those who emphasise the lack of detail retained in the internal representation
might describe it in terms of a ‘sketchy higher-level representation’ (Blackmore et
al., 1995) or ‘extremely reduced visual representations’ (Hayhoe, 2000). Two other
possible ways of thinking about it are as a ‘gist’ and a ‘virtual representation’.
The idea of the gist was proposed as part of a straightforward interpretation of
change blindness initially given by Simons and Levin (1997). During any single
fixation we have a rich visual experience. From that, we extract the meaning or
gist of the scene. Then, when we move our eyes, we get a new visual experience,
but if the gist remains the same our perceptual system assumes the details are
the same, and so we do not notice changes. This, they argue, makes sense in the
rapidly changing and complex world we live in. We get a phenomenal experience
of continuity without too much confusion.
Somewhat more radically, Canadian psychologist Ronald Rensink (2000) suggests
that observers never form a complete representation of the world around them –
not even during fixations – and that there is no visual buffer zone which accumu-
lates an internal picture. Instead, object representations are built one at a time, as
needed. Focused attention takes a few proto-objects from low-level processing
and binds them into a ‘coherence field’ representing an individual object that per-
sists for some time. When attention is released the object loses its coherence and
dissolves, or falls back into an unbound soup of separate features.
To explain why we seem to experience so many objects at once, when so little
is held in focused attention, Rensink argues that vision is based on ‘virtual rep-
resentations’: these are constructed from gist, spatial layout, and a longer-term
schema of the scene. They are not ‘structures built up from eye movements and
attentional shifts, but rather, are structures that guide such activities’ (2000, p.
36). We get the impression of a rich visual world because a new representation
can always be made ‘just in time’ using information from the world itself. Some-
times such representations may be stable; sometimes they may contain a large
amount of detail. But at no point are they both stable and detailed. Indeed,
eye-tracking evidence suggests that representations of the world may last only
about the same length of time as a fixation between eye movements (Tatler and
Land, 2011).
O’Regan (1992) agrees that we do not need to store large amounts of visual
information because we can use ‘the world as an external memory’, or as its own
best model, but he goes so far as to reject the idea that we need to make our
own internal models at all. He criticises traditional theories of vision for being
based (even if they don’t admit it) on the assumption that in visual perception,
the distortions and gappiness of the retinal image are compensated for by the
brain’s construction of a detailed model of the outside world, which somehow
creates perceptual consciousness. Instead, he argues that the visual world is not
something that we have, or build up, but something we do.
O’Regan and Noë (2001) proposed a sensorimotor theory of vision and visual
consciousness – a new way of thinking about vision that owes a debt to thinkers
like the phenomenologist Merleau-Ponty, who considered the body in action
‘we can use the world as
its own best model’(Clark, 1997, p. 29)