Medium of thought


  • [Required] Appendix P1 of Ludlow, Peter (1999). Semantics, Tense, and Time : An Essay in the Metaphysics of Natural Language. Cambridge, Mass: MIT Press. [accessible from NetLibrary, also intranet:ludlow-lot.htm]
  • [Required] Section 14.2 of Jackendoff, Ray S. (1990). Consciousness and the Computational Mind. Cambridge, Mass.: MIT Press. [accessible from NetLibrary]
  • Jackendoff, R. (2003). Foundations of Language. Oxford: Oxford University Press. isbn:0199264376
  • Jackendoff, R. (1996). The Architecture of the Language Faculty. Cambridge: MIT Press. isbn:0262600250
  • Carruthers (2002). The cognitive functions of language. In Behavioral and Brain Sciences 25:6. Online draft version at
  • Chapter 6 of Fodor (1998) Do We Think in Mentalese? Remarks on Some Arguments of Peter Carruthers. In In Critical Condition : Polemical Essays On Cognitive Science and the Philosophy of Mind. Cambridge: MIT Press. [accessible from NetLibrary]

Different theories about the medium of thought

Thoughts = certain types of mental representations. But what are they like?

  1. We think in pictorial representations (images, maps).
  2. We think in a natural language.
  3. We think in a language of thought that is not a natural language.
  4. We think in some other medium that is neither pictorial nor language-like.

Do we think in natural languages?

  • Natural language - the languages we use in ordinary social communication and which are learnt naturally - Cantonese, Japanese, German, etc.
  • As opposed to artificial languages in computer programming and logic - propositional calculus, predicate logic, HTML, BASIC, C++, etc..

Is the language of thought (LOT) a natural language?

Position #1 : LOT is a natural language

  • Introspective evidence - Subjectively, the experience of thinking is like internal speech, couched in natural languages. We think that the question "which language do you think in?" makes sense.

Position #2 : LOT is not a natural language?

  • Mental images - Many people claim that they sometimes think with mental images. It is true that mental images might still be made up of some kind of mental language, but probably not any natural language.
  • Fine-grained perceptual concepts - We can have thoughts about certain perceptual magnitudes which we do not know how to express in natural languages ("I can still remember that taste"). Think of knowing by sight the colour or size of an object.
  • Pre-linguistic infants and animals - If they can think, then it is likely that there is some thinking that does not rely on natural language. How can we learn a NL if there is no thinking without NL?
  • Ambiguity - Sentences can be (syntactically, lexically, referentially) ambiguous when the thoughts expressed by the sentences are not.
    • "We shall discuss violence on TV."
    • Applies to imagery as well. Think about Wittgenstein's old man example in Philosophical Investigations:

@I see a picture; it represents an old man walking up a steep path leaning on a stick. - How? Might it not have looked just the same had he been sliding downhill in the position?@

Position #2 is more plausible. But we also want a theory that can explain the introspective evidence.

Jackendoff's intermediate level theory of awareness

  • We are never directly conscious of our thoughts.
  • But thoughts can be projected to conscious mental imagery.
  • We seem to think in a language because thoughts can be projected to linguistic imagery.

PF and LF

  • Our innate language faculty ("I-language" according to Chomsky) contains two and only two syntactic levels of representation, Logical Form (LF) and Phonological Form (PF). LF interfaces with the conceptual- intention system. PF interfaces with the articulatory-perceptual system (speech).

From Jackendoff (2003)

  • Conscious thinking is associated with mental imagery. We have visual imagery, motor imagery, etc. The experience of talking to oneself is the result of our thoughts being expressed in verbal imagery, the contents of the imagery being determined by the appropriate PF.
  • A supporting piece of data? - Tip of the tongue phenomenon = thought without imagery.
    • We are conscious of having a thought, but we are not conscious of the contents of the thought.
    • Cases where our thoughts are about our conscious experiences.

From Jackendoff (1996)

  • Although syntactic ambiguity is resolved at the level of LF, syntactic properties do not

seem to be part of the phenomenology of thinking.

  • In the case of deaf people this might take the form of motor imagery (sign language gestures).

An objection

  1. There is a difference in subjective feeling between having a thought with content, and simply reciting in our minds a nonsensical sentence that we do not understand.
  2. The difference in feeling shows that we are conscious of the content of a thought.


  • Maybe we have a feeling corresponding to having meaning?
  • What about cases where we are entertaining two thoughts with distinct contents?
    • Feeling of congruence or dissimilarity?

Two alternative theories


  1. We think in LF. (Peter Ludlow)
  2. We have a conceptual representation system (LOT) which is distinct from LF (although it might have similar syntactic properties). (Jackendoff, Steven Pinker)

Objections to Ludlow

  • Can the same LF correspond to different thoughts?
    • What about LFs involving demonstratives and pronouns? "That is not the same as that.", "He hit him."
    • What about ambiguous terms? "What a tart." "Meet me at the bank." Are lexical entries part of LF?

@Lexical entry for a word = A database in the mind containing grammatical and semantic information about the word. e.g. cat : [countable noun] [a small mammal with four legs that purrs, usually with a tail and soft fur.] @

  • Presumably Ludlow would have to say that we think in interpreted LFs - logical forms supplemented by lexical entries and contextually supplied information.
  • What about animal cognition? Do animals have LF?
  • What about patients who suffer from Broca's aphasia? The patient's speech and writing are affected but full understanding of spoken and written language is retained.

@Yes... ah... Monday... er... Dad and Peter H... (his own name), and Dad.... er... hospital... and ah... Wednesday... Wednesday, nine o'clock... and oh... Thursday... ten o'clock, ah doctors... two... an' doctors... and er... teeth... yah source@