<div align=center>ToC | Chapter 2 -></div>
1 The Study of Intelligence
- Intelligence is too complex a notion to be captured by a simple definition. What people in general and even scientists mean by the term varies greatly, and there is little hope that there will ever be agreement. The key aspect, implicitly or explicitly present in many conceptions of intelligence, is generation of behavioral diversity while complying with the rules. This idea is independent of any notion of levels of intelligence. It applies to abstract thinking just as much as to an animal avoiding a predator. An organism that always displays the same behavior is not intelligent.
- IQ tests were originally invented to determine whether certain children would be better off in a special school. Eventually, the IQ test was turned into a general intelligence test, claiming to measure a general intelligence factor g. It is now generally agreed that intelligence is much too complex a phenomenon to be measured by a test yielding one number.
- Emotional intelligence has recently been proposed as being equally relevant for success in life as the kind of abstract intelligence measured by IQ tests. EQ tests measuring emotional intelligence have been suggested to complement IQ tests. It is still open to debate to what extent and in what form the EQ will survive.
- The nature-nurture debate concerns the extent to which knowledge is inborn or can be acquired. The behaviors and capabilities of a human result from a complex interaction of genetic and environmental factors. Thus, the answer to the nature-nurture question can only be that the origins of intelligence come from the interaction between nature and nurture.
- The Turing test was proposed to operationalize the notion of intelligence by means of an empirical test. It is based on the idea whether the (verbal) behavior of a computer can be distinguished from that of a human. If it cannot be distinguished, the computer can be said to have intelligence. Because the Turing test is based on natural language, it is restricted to human intelligence. It is still an open question whether it is a good test for human intelligence.
- The Chinese Room is a thought experiment proposed by Searle. On the basis of a set of rules a person - Searle - locked in a room produces output sentences from input sentences, exclusively on the basis of comparing the shapes of the Chinese characters which, to the person, are just meaningless symbols.
- There are analytic and synthetic approaches to the study of intelligence. The synthetic approach can be characterized by as “understanding by building”. In traditional AI and cognitive science, the models are computer programs; in embodied cognitive science, they are in the form of autonomous agents, either robotic agents or simulated agents.
- Autonomous agents exhibit emergent behaviors. Such behaviors are not programmed into the agents by the designer but rather are a result of the interaction of the agents with their environment.
- Autonomous agents, robotic or computational, can be used in three ways: as models of natural agents, to explore general principles of intelligence, and for specific tasks and applications. There are large areas of overlap among these three modes, especially between the first two.
- “Synthetic” implies design. The design perspective has turned out to be particularly fruitful for studying natural intelligence. The autonomous agents approach capitalizes on the design perspective.
- Embodied cognitive science designates the research field outlined in this book. It employs a synthetic methodology based on autonomous agents.
<div align=right>© 1999 MIT</div>
<div align=center>ToC | Chapter 2 -></div>