Language of thought – Wikipedia

The Language of thought (not Latin: tongue ) or mentalais (in English : mental ) is a theory of linguistics and psychology that the human mind would use a kind of clean language for its mental processes. Reflections on a language of thought have irrigated Western science and philosophy since the Middle Ages.

The question of the language used by thought occupied philosophers and thinkers for several centuries. Plato already defined thought as inner dialogue, but without questioning the nature of the language used by this inner dialogue ^{[ first ]} . This thesis will be preserved by the scholasticism and is still found in Guillaume d’Ockham. The Middle Ages is rich in reflections on the existence of a tongue (Language of the Spirit), which is found for example in Thomas Aquinas ^{[ 2 ]} . Later, in the Logic of Port-Royal (1662), Antoine Arnauld and Pierre Nicole maintain that “If the reflections we make on our thoughts had never watched that ourselves, it would have been enough to consider them in themselves, without taking them any words, nor any other signs” ^{[ 3 ]} .

The modern expression of “Language of thought” seems to have been used for the first time by the American philosopher Gilbert Harman, in 1973, in Thought ^{[ 4 ]} . However, the thesis was popularized in this 20th century by the American philosopher Jerry Fodor. He defends the thesis that there is a language proper to the mind, to which mental processes use, and which makes it possible to develop complex thoughts from simple concepts. This thesis, supported in 1975 in The Language of Thought , may, according to ANSGAR Beckermann, be schematized as follows ^{[ 5 ]} :

1. Mental representations are structured.
2. The components of these structures are “transportable”; The same components (that is to say components of the same type) can appear in different representations.
3. Mental representations have compositional semantics; The meaning of complex representations results, through rules, from the meaning of components.

The language of thought would therefore have a structure composed from atomic elements (like the words of a natural language). These semical components can appear in various representations, as well as words or constituents of sentences can appear in different sentences. As in other languages, the meaning of individual representations consists from the meaning of their constituent elements.

However, the Mentalae differs from other languages ​​in that it is realized, not acoustically or optically, but through neural configurations, or bits in the memory of a computer.

The purpose of the Mentalais hypothesis is to explain the existence of intentional states in a physicist framework. Physicalism, understood in the strong sense, claims that there are only objects, events, and physical properties. Mental states and properties would thus be reducible to physical states and properties (reductionism). However, intentional states (such as wishes, beliefs, etc.) such as other types of mental states (for example qualia) do not seem first reducible to physical states. The intentional states, also called propositional attitudes, are mental states which relate the subject with a proposal, for example Paul believes that Marie is happy implies that Paul has a certain relationship, a relationship of belief, vis-à-vis the proposition Marie is happy . But how can he devil, Paul, a physical entity, can he be related to a proposition, something that seems outside the physical world? The Mentalais hypothesis makes it possible to explain how propositional states can be realized physically. The thinking subject establishes a relationship with mental representations identifiable with neural states and performs operations on it similar to that of a speaker building a sentence by obeying syntactic rules.

In order to carry out the propositional states, the Mentalais thus presupposes a computational model of the spirit and is therefore part of the dominant paradigm of the cognitive sciences: computationalism. Such a model postulates that the human mind works like a Turing machine, says more simply as a computer. Thought is therefore conceived on the model of calculation as a succession of elementary operations, which execute themselves thanks to the syntactic properties of the components of language.

Introspection leads us to believe that we think in certain natural languages, especially most often in our mother tongue. Why then should we assume that there is a language specific to thought, different from natural languages? The first answer is that there is a “thought without language”, understand an unused thought in natural language. This is proven in particular in animals and infants who have not yet acquired their first language but also in adults (see in particular the work of Lev Vygotski). Furthermore, the Mentalais hypothesis joins the central hypothesis of the generative grammar of an innate faculty/language provision which would be necessary for the acquisition of language.