Chapter
Fifteen
Dreaming and beyond
Unpredicted stimuli produce a robust spike (a startle response) in these waves
in both waking and in REM sleep. Startle responses mean that a mismatch has
occurred between what was predicted and what occurred, so their presence in
REM sleep suggests that hypotheses are being tested against sensory evidence
here, too. But a problem for this theory is that there is less habituation in this PGO
startle response during sleep. This seems the wrong way round because surprise
during dreams is very rare. Indeed, this is one of the oddities of dreaming: that
we are so little surprised by truly bizarre dream events. The authors suggest that
maybe we are incapable of surprise in dreams because what we experience is
the result of top-down predictions in the thalamocortical system. This fits with
the idea that dreaming is more similar to imagination (being driven by top-down
intentions and predictions) than to sensory perception (driven in a bottom-up
fashion by environmental percepts) (Nir and Tononi, 2010).
The brain, they say, is propelled in both sleeping and waking to infer the causes of
its sensory sampling, like scientists driven to test their hypotheses (Hobson and
Friston, 2014). The Cartesian theatre is a metaphor for the virtual reality models
this creates. But there is no inner audience observing the show – only stories and
fantasies being rehearsed and tested against sensory evidence. Is this a perni-
cious Cartesian theatre? Interestingly, Hobson and Friston arrive at a new kind of
dualism, a duality between the conscious processes of inference and the physical
brain states that encode them, claiming that this may help ‘dissolve some of the
mysterious aspects of consciousness’ (p. 6).
Overall, however, Hobson and Friston are much more interested in the processing
aspects of sleep than in the phenomenology of dreaming. For them, sleep is what
does the important work, while dreams are merely ‘the subjective epiphenomena
of the nocturnal products of our virtual reality generator and contain no new
information’ (2012, p. 87). Maybe, they suggest, this is why it isn’t usually worth
our while to remember them. But to dismiss something as important as dreaming
as just an epiphenomenon raises its own philosophical problems (see Chapter 1).
Another example of how to connect physiology with experience comes from the
dream bizarreness research by Revonsuo’s group. They argue that three types of
bizarreness can be understood as failures of three types of binding: feature bind-
ing, contextual binding, and binding across time. They conclude that ‘more global
forms of binding flounder much more frequently than those concerned with only
local bundles of features’ and relate this to the number of distinct processing
modules involved in generating different kinds of dream images (Revonsuo and
Tarkko, 2002, p. 20). In other words, the harder it is for the brain to construct a
certain kind of integrated image, the more likely it is that such an image will fall
apart or show bizarre failures of binding during dreams.
This suggests that even the most peculiar of dream features may yield to a study
of brain mechanisms during sleep. Even so, we are still relying on correlations, and
as with all other aspects of conscious experience, we cannot say with confidence
that dreaming and brain states are reducible to each other or are the same thing,
nor can we confidently describe them in terms of ‘brain–mind states’.
So far we have been assuming that dreams are conscious experiences, but is this
true? Some philosophers have questioned whether dreams are experiences at all
(Malcolm, 1959; Dennett, 1976).
‘[Dreams are]
the subjective
epiphenomena of the
nocturnal products
of our virtual reality
generator’(Hobson and Friston, 2012,
p. 87)