Intelligence
Definition
Human intelligence is a measure of human ability to solve problems.
Definition ambiguities
There are dozens of definitions of human intelligence in hundreds of books and articles. However, problem solving capacity is the simplest and most universal measure that seems to encompass all other abilities.
For example, consider the following "broader" definition by expert consensus:
A very general mental capability that involves the ability to reason, plan, solve problems, think abstractly, comprehend complex ideas, learn quickly, and learn from experience. It reflects a broad capability for comprehending our surroundings: "making sense" of things, or "figuring out" what to do
It is easy to see that problem solving requires reasoning, abstraction, comprehension, and learning.
Jeff Hawkins
Jeff Hawkins defines intelligence as the power of adaptability. He is right when he says that machines can beat humans in nearly all domain-specific tests, but they are not as universal as humans. It is us who are intelligent because we are adaptable. He plans to build a model of the neocortex that could be simulated in a computer. This way he would accomplish intelligence. However, we are all born with the type of adaptability Hawkins seeks. Despite that marvelous adaptability, we do not call babies intelligent. Adaptability underlies a potential to develop intelligence. Human adaptability is clearly more universal than that of machines (in 2020). Hawkins's definition comes from his interest in building intelligent systems. However, that definition is not very useful in my considerations on human excellence at this site.
The key to the adaptability is the conceptualization of the concept network of the brain. We are all born with that precious quality. Over time, intelligence emerges as the ability to solve problem by means of conceptual computation.
Alexander Wissner-Gross
A broader definition of intelligence has been suggested by Alexander Wissner-Gross, which explains the driving force behind all intelligent actions. It may also help demonstrate that an intelligent system (or a newborn baby), need only 3 components for the emergence of intelligence: (1) senses, (2) concept network (e.g. neocortex), and (3) the learn drive.
For a goal-oriented intelligent system, we will also need (4) effectors to interact with the environment.
Abstract intelligence
There have been many attempts to separate intelligence from domain-specific knowledge. Such a separation is the purpose of IQ tests (see: IQ myth). Raymond Cattell's separation of fluid intelligence (smarts) and crystallized intelligence (knowledge-based) has the same intent. Formal definitions of intelligence may attempt to average expected performance over a large set of environments of unknown complexity. However, the inescapable property of intelligence that it is domain-specific, and adaptable. Knowledge of chemistry makes a chemist intelligently solve problems in her field while often being intellectually impotent in the area of dentistry. In a colloquial sense, we may tend to require universality as a criterion of intelligence. However, in a world of problems to solve, we need better domain-specific problem solvers. Universal genius is attractive. However, only the adaptability of intelligent systems is universal. Universality has many advantages. However, it also carries costs. We need more Newtons who would be as universal as necessary to solve the problems they decide to tackle. Throughout SuperMemo Guru, I speak of intelligence as of a tool needed for achieving specific goals. Intelligence is a mutable and trainable property rather than a birth right.
Further reading
- Simple formula for high intelligence
- How to solve any problem?
- IQ is a dismal measure of intelligence
- Concept network
- Conceptual computation
- Society as a concept network
This glossary entry is used to explain texts in SuperMemo Guru series on memory, learning, creativity, and problem solving
Figure: Creative problem solving requires (1) vast expert knowledge of high stability, and (2) rich new knowledge of high retrievability. Vast stable knowledge makes it easy to solve algorithmic problems. Those problems can be solved at low energy expenditure with the help of fast thinking. Problems that require "thinking out of the box" rely on creativity, i.e. association of remote ideas. Creativity and learn drive are powered by "hungry knowledge", i.e. fresh knowledge that can easily be molded and generalized via forgetting. This plasticity provides for good pattern matching in new learning and in creative problem solving