Next term I will be teaching a new course on intelligence that compares machines, humans, other animals, and maybe even plants. Many people would think it natural to start with a definition of intelligence, on the assumption that we don't know what we’re talking about unless we can define your terms. In cognitive science, however, there is widespread appreciation that the definitional view of concepts doesn't work. So how can we go about characterizing a concept such as intelligence?
Chris Eliasmith’s ideas about neural representation can be used to integrate the three standard ways that psychologists talk about concepts. These ways are sometimes presented as alternative theories of concepts, but are better viewed as different aspects of concepts. First, understanding of concepts is partly based on exemplars, which are standard examples to which the concept applies. For human intelligence, exemplars include Albert Einstein, Thomas Edison, Jane Austen, and Ludwig von Beethoven. We can also note standard examples of intelligent actions, such as generating scientific hypotheses, inventing new machines, writing insightful novels, and composing beautiful music.
The second important aspect of concepts is a list of typical features, constituting a prototype or stereotype. There is little prospect for providing a list of necessary and sufficient conditions for being intelligent that would allow us to say that something is intelligent if and only if that has those conditions. But it is easy to come up with typical features of intelligent persons and actions. Many of these are found in dictionary definitions of intelligence that are otherewise circular, prone to counterexamples, and inconsistent with each other. Nevertheless, they tend to mention many of the same features, such as abilities to solve problems, to learn, to understand, and to reason. We can appreciate that effective problem solving, learning, understanding, and reasoning are typical features of intelligence without worrying about whether they provide necessary or sufficient conditions.
The third aspect of concepts discussed in contemporary psychological theories is explanation. We frequently use concepts to explain important observations, and we also look for explanations of the phenomena described using the concept. For example we use the concept of intelligence to explain why some people seem to be faster and more effective at understanding, learning, and reasoning. Going in the other direction, psychologists attempt to explain the origins of intelligence in terms of factors such as genetics, social environments, and motivation.
Therefore, even though no dictionary has succeeded in giving a strict definition of intelligence, it would be silly to say that nobody knows what intelligence is. There is a wealth of knowledge about intelligence in the form of exemplars, typical features, and explanations. Failures of philosophical attempts to analyze concepts using definitions sometimes lead to skepticism about the comprehensibility of concepts, or to the obscurantist conclusion that concepts like good or knowledge just have to be taken as primitive and known mysteriously.
Many other important concepts in psychology and the philosophy of mind are open to the same kind of three-part characterization that I have used for intelligence. Here are just a few important ideas that can benefit from being discussed in terms of exemplars, typical features, and explanations: mind, understanding, learning, emotion, consciousness, creativity, and even the self. This kind of naturalistic analysis shows how we can have some understanding of a concept without the ability to strictly define it. Of course, our understanding of a concept can increase dramatically when we expand its range of exemplars and notice additional typical features, and especially when we develop much richer theoretical explanations. There is much more to be learned about intelligence by developing better theories about how it works in machines, people, and other animals. Conceptual change and theory change unfold together.