As a game theory enthusiast, I recently began reading Meaningful Games because it seems to apply the theory to a new area which I never considered before. But several chapters into the book, I became sidetracked with a discussion on Mentalese. What intrigued me about this particular “language” is that it appears to be universal. In fact, it maps any word in any spoken language to a mental linguistic syntax that is itself highly structured. There are many philosophical characteristics of the the Language of Thought hypothesis (which you can view here), but I am more concerned with the most fundamental mappings and networks that make Mentalese work.
The basis idea of Mentalese is that when one processes language from the real world, one first translates the language into Mentalese and then finds the real world object that the word represents. In a model borrowed from Meaningful Games, the process the brain undergoes in understanding something is the following:
Language -> Mentalese -> World Model
The concept of mentalese might be best explained by a graph. Each node of the graph represents an idea or a “thing.” For example, the words “ice skates,” “people,” or “Washington D.C.” might translate to ICE SKATES, PEOPLE, or WASHINGTON D.C. in Mentalese, though obviously the words in Mentalese are not English words. In fact, according to the language of though hypothesis, the capitalized words in Mentalese are just abstract representations or thoughts connected to the English words but in a language not representing English at all.
However, one may ask, in terms of evolutionary efficiency, what the point of Mentalese might be. What is the point of having a middleman between the English word and the real world object? In the simple example described above, it might be simpler to state:
Language -> World Model
But this argument does not stand up to scrutiny as it leaves little room for intercommunication between people who speak different languages. For example, the Spanish word for cat is gato and the French term for cat is chat, while the English word for the furry animal is cat. If we were to assume that there is no Mentalese language, and that the word “cat” directly translates to the physical animal, then there would be no possibility of ever reaching a consensus between someone who speaks Spanish and another who speak some other language, say English. The model of the world would look like this:
If an English speaker looks at a cat, he will associate the real world model with the word “cat.” Similarly, if a Spanish speaker looks at a cat, he will associate it with the word “gato.” Thus, we see the following relationship:
Let’s suppose, however, that the Spanish speaking individual spots the cat first and wants to tell the English speaker that she had seen a cat. She points at the cat and says “gato.” The English speaker, then, looks at the cat. If we consider the two propositions listed below to be true based on the relationships we derived above, the English speaker should logically deduce that the word “gato” refers to cat because the animal he is currently looking at is a member of the same species of furry felines. But that is not the case as we will soon see.
Proposition 1: The English speaker will associate the word with the real world model of a cat if and only if the word used is “cat.”
Proposition 2: The Spanish speaker will associate the word with the real world model of a cat if and only if the word used is “gato.”
To the English speaker, the Spanish word “gato” signifies something other than cat. The second part of Proposition 1 is false. Therefore, we can conclude that the first part of the statement is also false, thus letting the English speaker believe that he is not looking at a cat. However, this reaches a contradiction as the creature is obviously a cat and if introduced as a “cat” by another English speaker, the former would suddenly realize that the creature was indeed a cat. Certainly, this is not the way we process information. Instead, the way we process information might look more like the following if there exists such a language as Mentalese perhaps.
In this context, there would be no ambiguity as to which real world model each word is referring to.
The addition of verbs and phrases adds complexity to the mental language but not much so. The higher complexity results from concatenating nouns with verbs in ever more complex syntactical structures. However, the basic idea follows a “fill in the blank” format. The sentence, “The cat chased the mouse,” becomes CAUSE(CHASE(CAT)(MOUSE)). If you would like to learn more about the structure of Mentalese, please refer to the first few chapters of Meaningful Games or a quick Google search reveals several contrasting opinions on the topic.