| Description |
In the last score of years considerable enthusiasm has been generated over expert systems in the artificial intelligence and the business communities. Expert systems are one of the more successful developments in the search for artificial intelligence, with substantial applicability in the business world. There are many different types of expert systems, but the majority of the population can be divided into two types: symbolist and connectionist. The two schools of thought, symbolism and connectionism, have engaged in an ongoing debate over the architecture and definitions of expert systems. The symbolist school defines expert systems in terms of explicit symbolic representation of the world and serial processing. The connectionist school rejects symbolism and serialism in favor of weighted connections and parallel processing. The architectural differences between the two types of expert systems produce inherent advantages and disadvantages in each system. To a large degree, the strengths and weaknesses of each type of system are complementary. Symbolist expert systems are adept at complex problem solving but weak in tasks of perceptual recognition. Conversely, connectionist expert systems excel at perceptual recognition but perform poorly in problem solving. In addition, as expert systems are a product of the field of artificial intelligence, they have relevance to the phenomenon of human cognition. The debate between the two schools of expert system design often breaks down to the dispute over which expert system philosophy is most similar to human cognition. This approach is faulty, as the different types of expert system are similar to different aspects of human cognition. Fodor's theory of cognitive modularity provides a model which reveals the complementary nature of the two types of expert systems. Connectionist expert systems are analogous to human unconscious perceptual systems that Fodor names "input systems." Symbolist expert systems are analogous to the conscious "central processes" of Fodor's theory which control logical reasoning and associative thinking. However, the problem remains that both types of expert systems emulate only a portion of human cognitive activities. Because both types of expert systems fail to provide insight into the nature of consciousness, they ultimately fail to explain human cognition. |
