THINKING ABOUT KNOWLEDGE LEARNED FROM INSTRUCTION AND EXPERIENCE - 2 TESTS OF A CONNECTIONIST MODEL

Precise characterizations of thinking processes were tested quantitati vely in 5 experiments. Newly learned bodies of knowledge were shown ti s be subject to thinking processes that were simulated by spreading ac tivation through an associative network: In 3 domains, thought pattern s of 717 Air Force recruits were successfully predicted by a power alg orithm spreading activation process, which always produces the first p rincipal component of the associative network, implying that specific form for the thinking process. Similarly, preexperimentally learned kn owledge was shown to be organized into distinct, discrete subrepresent ations corresponding to the principal components of the knowledge netw ork's associative matrix: In 120 undergraduates, principal components with large loadings were successfully induced and completed when incom plete, whereas components with small loadings were squelched.