Brit. J. Phil. Sci. 56 (2005), 469–485
You Don’t Know How You Think:
Introspection and Language of
Thought
Edouard Machery
ABSTRACT
The question ‘Is cognition linguistic?’ divides recent cognitive theories into two antag-
onistic groups. Sententialists claim that we think in some language, while advocates of
non-linguistic views of cognition deny this claim. The Introspective Argument for
Sententialism is one of the most appealing arguments for sententialism. In substance,
it claims that the introspective fact of inner speech provides strong evidence that our
thoughts are linguistic. This article challenges this argument. I claim that the Intro-
spective Argument for Sententialism confuses the content of our thoughts with their
vehicles: while sententialism is a thesis about the vehicles of our thoughts, inner speech
sentences are the content of auditory or articulatory images. The rebuttal of the intro-
spective argument for sententialism is shown to have a general significance in cognitive
science: introspection does not tell us how we think.
1 The problem
2 The introspective argument for sententialism
3 The argument for the blindness of introspection thesis
4 Objections and replies
5 Conclusion
1 The problem
The question ‘Is cognition linguistic?’ divides recent cognitive theories into
two antagonistic groups. Among each group, there are for sure some inter-
esting and deep differences. Still, this opposition is crucial. Several views of
cognition, for example the recent dynamic systems approach (e.g., Thelen and
Smith [1994]; Port and van Gelder [1995]) or the new wave of empiricism
(Barsalou [1999], Prinz [2002]), have been explicitly developed to supersede
sententialism; similarly, its adequacy has been the object of a passionate con-
troversy in psychology, philosophy and artificial intelligence (e.g., McClelland
et al. [1986]; Smolensky [1988], [1991]; Fodor and Pylyshyn [1988]).

The Author (2005). Published by Oxford University Press on behalf of British Society for the Philosophy of Science. All rights reserved.
doi:10.1093/bjps/axi130 For Permissions, please email: journals.permissions@oupjournals.org
Advance Access published on August 10, 2005.

For present purposes, the sententialist hypothesis claims that the mental
representations that feature in the cognitive processes constitute a language.
According to the Mentalese hypothesis, the cognitive processes are defined
over mental representations that constitute a non natural language (Fodor
[1975], [1998a]; Jackendoff [1989]). To put it simply, when you entertain a
thought that p, you token a sentence of a language that does not possess the
characteristic properties of natural languages. On the contrary, some claim
that human beings’ cognitive processes are defined over expressions of natural
languages (Sellars [1956]; Carruthers [1996], [1998a], [1998b], [2002]; Devitt
and Sterelny [1999]). When you, English speaker, entertain a thought that p,
you token a sentence of English that means p.
The non-linguistic conception of cognition is really a mixed bag. It includes
very different conceptions of cognition, which are only unified by their
rejection of the sententialist orthodoxy.
Sententialists have put forward many arguments for their pet view of
cognition (e.g., Fodor [1975]; Fodor and Pylyshyn [1988]; Horgan and
Tienson [1996]; Carruthers [1998a], [1998b], [2002]). Among all those more
or less technical arguments, one of them is particularly seductive, for it is
based on our own experience, on the way most of us experience our own
conscious thoughts:
In substance, the argument infers inductively from the introspective fact
of inner speech1 that our conscious propositional thoughts are natural
language sentences.2
When we introspect our own conscious propositional thoughts, we have
access to thoughts expressed in natural language sentences; in other words,
we find ourselves engaged in inner speech. This introspective fact is treated as
evidence that we do think consciously in a natural language: as Carruthers
states it ([1996], p. 50), ‘introspection informs us, in fact, that many of our
thoughts are expressed in natural language.’3 I call this inductive inference the
Introspective Argument for Sententialism.
In this paper, I purport to refute that claim:
The introspective fact of inner speech does not count as evidence that our
conscious thoughts are expressed in a natural language.
1 In the psychological literature (e.g., Smith et al. [1995]), the term ‘inner speech’ is sometimes used
in a restricted way: inner speech requires covert muscle movements. The subvocalized rehearsal
that does not require covert muscle movements does not count as inner speech, so conceived.
I use this term in a wider sense: it includes all the cognitive processes that result in the phe-
nomenology of heard sentences in the head.
2 This argument does not establish that our entire cognition is linguistic. It supports a restricted
version of sententialism: our conscious thoughts constitute a natural language.
3 Carruthers has been using this argument to support the natural language view of cognition
against the claim that all thoughts are expressed in Mentalese. But since the natural language
hypothesis is a version of sententialism, this argument also supports sententialism.
470 Edouard Machery