Previously, I have blogged about meaning in language, and especially the dichotomy related to it. Now, I feel, it's the time to look at the issue from another perspective. Sure, our language models meaning in its own way, but how is it represented in the brain? This post will be about concepts and how they are understood. 🤓

Definitionalism

This view states that there must be a set of initial concepts that are used to define all the other concepts. As an example, a higher level concept such as dog could be defined by using lower level concepts such as thing and living. The question that arises from here is that what exactly are these concepts to begin with? 🤔

Classical empiricism states that the initial concept are perceptual, like mental images. I.e. what we have perceived from the surrounding world. Those perceptions are then used as anchors in defining concepts. Another view, namely nativism, states that that basic concepts are abstract in their nature and that they are innate, hardwired to our brains. These both views pose problems. If concepts are perceptual, how are our perceptions categorized? And if they are innate, is it possible that there are concepts we will never comprehend because we lack the innate concepts to define them? Luckily we have enough innate concepts to understand computer. 😆

Prototype theory

Concepts are seen as prototypes. A prototype is sort of a mean of the properties of the thing it's a prototype of. Or at least it is something representative to all the possible tokens of the prototype. For example, a prototype of a bird would be somewhere in between a small bird like a great tit and a big one such as an eagle. It would be a bird that knows how to fly, although there are non prototypical birds such as penguins that can't do that. 🐧

In contrast to the definitionalism, prototypes have fuzzy boundaries and they don't require exact definitions based on other concepts in the brain. Prototypes are complicated from the point of view of compositionality. We can have a prototype of pet (probably something related to a dog or a cat) and a prototype of fish. However, when we try to combine them into a pet fish, we run into problems. Is the prototype of pet fish a prototypical pet, like a dog, combined with a prototypical fish? Likely not. 🤣

Exemplar theory

Concepts are a list of all the examples we have encountered in life of a given thing according to this theory. This means that your concept of dog might include your relative's dog, your neighbor's dog and so on. This would mean that we store in our brains huge lists of things we have encountered in life. But the question would still remain that how do we know how to group these different kinds of things into certain categories? How do we know that dogs and cats should be in different categories? Also, abstract concepts would be rather difficult to explain with this theory. Just think of how many times you have encountered different instances of world peace in the real world? 🙃

Conceptual role semantics

In this view, concepts are defined by how they relate to other concepts in the brain. The difference to definitionalism is that here concepts are linked to others based on their semantic relations (read more about semantics here), and they are not strictly defined by some initial concepts.

This view can be approached from the point of view of molecularism or holism. Molecularists claim, that the meaning of a concept is formed only by some of its relations. This means that what the concept animal means, doesn't depend on all the possible concepts for different species of animals, but rather a small subset of its semantic relations is enough to explain the concept, such as dog, cat and bird, for example. Holism, on the other hand, states that all the relations are needed for the meaning of the concept. So in order to understand animal, one would have to use all the concepts related to it: cat, dog, wolf, bird, salmon and so on.

Conceptual atomism

Conceptual atomism rejects the previous theories entirely. I.e. concepts don't gain their meaning by definitions, examples, prototypes or relations with other concepts. Instead, they are atomic in the cognitive level. What this means in practice, is that concepts are like words in a language. The word for dog doesn't depend on the words animal or bulldog, but it's atomic on the lexical level. Of course it's a combination of phonemes on a lower level, but so are concepts combinations of neurons. It doesn't make them any less atomic.

When atomic concepts are combined together, like pet fish 🐠, predicate logic or something similar is used. This would make compositionality possible in language and thought. However, I tend to think that conceptual atomism doesn't really solve the issue of concepts at all. It just introduces a new language that is known as the language of thought 💭(LOT), which just makes me ask how are the concepts formed in the LOT? Can anything in such a high level really be atomic?

That's all for this time. In the future, I will blog about language and the brain which won't be on such an abstract level as this post. 😄