Language Of Thought Simplified

Computational Representations and Mind Khosrow Ahsani Ghahreman

Empiric philosopher John Locke explains how external facts are written down in tabula rasa by associative learning. Behaviourists’ paradigm has added nothing new to this assumption. I will examine development of theory of meaning starting from behaviourism. I have used these symbols in the following lines: For “extralinguistic reality” inside quotation mark For Associative link by fleshÆ For language entities as separate from real In italics

What is the meaning of this phrase, Snow is white. Externalists claim that thoughts or other mental states are caused by external stimuli. In our example snow refers to “snow” which exists in extralinguistic world, and white is a certain electromagnetic wave which that thing emits. So: snow (S1) Æ white (S2) ; is derived from association or contiguity of “snow” and “white” Snow is white. is nothing more than response or image of our mind to “snow” and “white”. The assertion which we have learned from external reality. Therefore the truth of Snow is white completely depends on its correspondence to the mentioned reality. This is so called correspondence theory of truth. When we assert that Snow is white a mental image of “snow is white” is activated in our mind. What is language? Instead of seeing “snow” you see or hear the word snow.Instead of seeing “white” you see or hear the word white .Here, simply a mental image is substituted by another, note again by the means of association. (every time I saw “snow” my mom spelled the word snow.) “Snow” is substituted by snow, which evokes same mental image. So far we have discussed about concepts, but what about the meaning of propositions? The meaning of Snow is white is the condition in which the proposition would be true. In our example: If you see “snow”, then it is “white”. In the famous PÖ Q form. This is so called truth-conditional theory of meaning. --------------------------Formal language cognitive models are essentially based on these assumptions, but are a little more elaborated ; which I am going to explain.

If you see an animal, how can you recognize that it is a “bird” or not? Look at the figure, associations could be integrated by simple connections, the so called weighted connections, which in turn do not permit to pass immediately from external world to concepts, but only after passing from “hidden unit” processing filters. Here internal images do not function as simple reflective mirrors. ASSOCIATIONISM In simple associative theory of meaning, every external world object or event is associated with its corresponding word. We learn by associations, even cause and effect rules are nothing but habitually gained permanent associations.

CONNECTIONISM In neural networks representation of an external object or event is not mirror-like but it submits rules of simple representations to combine. Here we call this thing a bird, even it can not fly (Fl), because weight of other representational atoms : B (beak) and F (feather) is sufficient for molecular representation of bird; >3.

CONNECTIONISM This figure shows neurons, when they are activated transmit their message to activate the word’s representation. The subject observes that the animal does not fly so Fl neuron is not activated. In fact representations consist of rules and atomic symbols, they need not be biologic, as in artificial neuromorphs. These do not even occur consciously, in human.

† For explaining language of thought or symbolicism let us consider historic criticism on behaviourism, which give rise to cognitive psychology in 1960s, a discipline for which Jerry Fodor is one of the founders, the same person also coined language of thought. The question was how a child learns his native language, if he ought to learn by associations, it would be impossible for him, for the reason that if a child should learn association of 20-word sentence, he must be exposed to 1020 combinations, which will take age of the earth for hearing all of them. But a normal man potentially can create any of these 20-word sentences.

If a child would learn grammar by association and then generate phrases by grammatical rules, so there must be a syntactic linguistic representation unit in our mind. Either acquired or innate, it must work by syntactic rules, able to manipulate symbols and produce (generate) linguistically acceptable word combinations and propositions. Consequently our mind from “snow” to “snow is white” crosses from linguistic analytic unit. This passage is neither directly from external world stimuli to word-form; nor via extra-grammatical weighted connections. Accordingly syntactic operations act on symbols to produce meaningful statements. In fact our spoken or written language only may be distorted by the cause of the activity of the other parts of the mind, not by language module proper. Actually it is not the language which has its counterpart in internal syntactic representational operations, but on the contrary our language is the performance of them. This language unit or module (more common term) must naturally reside in hidden unit. It is not conscious. It is clear that this unit’s rules work independently of external stimuli, however our language is in contact with external world and tries to communicate about it. † The symbolicism model is not shown in figure, but if you look at neurons and imagine the relation between them not as simple weighted connections but representative operative syntactic rules, you will get the model. It is important to note that like connectionist models physical nature of these representations is not important. The dualism of symbolic and physical is essential for this type of functionalism. ------------------------