one would maintain that it is a distinctively connectionist architecture,
rather than a connectionist implementation of a system which can also be
described in terms of representations and processing rules.
Consider now an instance of theoretical reasoning (loosely described, to
include any system which can generate new beliefs from old). Again folk
psychology would seem to be committed to the existence oftrains of
thinkingand reasoning, within which discrete thoughts are tokened, and
where, once again, common conceptual components are shared between
consecutive thoughts in the sequence. Suppose that I return home from
work to hear the cat meowing. Then usingbelto represent new beliefs
grounded in perception or inference, andmemto represent old, the fol-
lowing sequence may occur:
bel[the cat is meowing]
mem[when cats meow, it is often because they are hungry]
]bel[the cat is probably hungry]
mem[food removes hunger]
]bel[feeding the cat will probably remove its hunger]
]bel[feeding the cat will probably stop it meowing]We believe that the various steps in this sequence of thought are distinct
from one another. And it seems crucial to the rationality of the sequence
that there should be common conceptual elements shared between the
states – for example, thatcatshouldWgure in all fourbelstates, and that
hungershouldWgure in the twomemstates. And again it seems unlikely that
a distributed-connectionist network would replicate these features of the-
oretical reasoning.
Is our folk-psychological belief in the causal systematicity of cognitive
processes derivative from natural language in some way? Is it because the
folk believe that thinking is conducted in natural language, or because they
reportacts of thinking in natural language, that they come to believe that
thought is systematic? Well, as we shall see later in this chapter, we do
think that one way of instantiating systematicity is to have conscious
thought-processes involve imaged natural language sentences, in such a
way that conscious thinking is conducted in ‘inner speech’. If the vehicles
of conscious propositional thoughts are natural language sentences, then
the systematicity of conscious thought can be derivative from the sys-
tematicity of language. But we are doubtful whether this is why the folk
believe in systematicity. If it were, then they should be happy to allow
connectionists free rein in the domain of non-conscious thought, or in
respect of non-linguistic creatures – and this would then mean that animals
and infants fail to engage in genuine inference. In fact our folk-commit-
202 Forms of representation