Process training derived from a computer simulation theory

Abstract

Kotovskyand Simon (1973) identified four basic subprocesses in their computer simulation of adult and adolescent performance on Thurstone letter series completion problems. In Experiment I, children from Grades 1 to 6 were pretested on those problems, and then experimental subjects were trained on two of the four processes as an attempt to experimentally support a correspondence between the computer subroutines and human cognitive processes. A posttest administered in the experimental and control conditions revealed a significantly greater improvement for experimental subjects, although both groups made significant gains. The children's distributions of errors were consistent with Kotovsky and Simon's predictions. In Experiment n, children from Grades 3 and 5 took four series completion tests without intervening training. The additional practice was sufficient for Grade 5 subjects to make improvements similar in magnitude to those produced by training. Grade 3 subjects, however, made no gains. These results are related to Tulving and Pearlstone's (1966) distinction between the availability and the accessibility of memory traces.The psychological reality of computer simulations has generally been derived by comparing human protocols with the outputs of the computer program for the given task. While this method provides, in some sense of correspondence, a test of theory, it does not provide the usually accepted form of independent experimental verification. One possibility for experimental work is a training study in which humans are taught to perform the identified component processes. If a computer simulation model has at least approximated the processes utilized by humans in correctly solving a problem, then instructions on these processes should improve the performance of subjects who initially performed poorly on the task. On the other hand, if the subroutines involved are incompatible with the cognitive structures of the human subject, then instruction on the subroutines should either not influence performance or influence it detrimentally. The present study investigated whether a computer simulation model could suggest subroutines that were instructable and whether instruction on these subroutines could facilitate subjects' solutions to the problem task.

Individual Differences and School Learning Environments

“…It has been suggested that if the basic components of these required aptitudes can be identified and assessed, it may be possible to provide instruction that will remediate individual weaknesses. The effectiveness of such instruction in improving performance has been demonstrated in the case of analogy type test items (Sternberg & Ketron, 1982) and letter series completion items (Holzman, Glaser, & Pellegrino, 1976). There seems to be no evidence as yet, however, that this type of improvement in learning skills or aptitudes is generalized to new situations (Campione & Brown, 1978).…”

Unities in inductive reasoning.

“…Several computer programs, for example, have been devoted exclusively to the solution of analogies (Evans, 1968;Reitman , 1965) o~ series completions (Simon ~ Kotovsky, 1963), and other computer problems have dealt with analogies or series completions, among other kinds of problems ( Williams , 1972;Winston , 1974). Analogy problems have been studied experimentally in a number of investigations with human subjects (e .g., Moiholland , Pellegrino , $ Chaser , in press; Sternberg, a, 1977bWhitely ~ Barnes, 1979), as have series completions (e.g., Holzman, Chaser, ~ Pellegrino, 1976;Kotovsky ~ Simon, 1973). We are unaware of any previously published information-processing analyses of the classification task, although Pellegrino and his colleagues are currently studying this task (Pe hlegrino , Note 1), and the literature on concepts and thel attainment can be viewed as indirectly studying this sort of task (e.g.…”

Nothing fails like success: The search for an intelligent paradigm for studying intelligence.

“…There can be little doubt that some training of cognitive processes is both possible and feasible (see, e.g., Borkowski & Cavanaugh, 1979;Feuerstein, 1979;Holzman, Glaser, & Pellegrino, 1976). Although evidence supporting the durability and generalizability of such training is still i;eager, we have cause to be at least modestly optimistic regarding the feasibility of training some cognitive skills (Bro-,.…”