GRAMMAR LEARNING AND SERIAL REACTIONS - 2 PARADIGMS - ONE LEARNING MECHANISM

Three experiments were carried out to investigate whether implicit eff ects in the paradigm of grammar learning and in the paradigm of serial reaction time tasks can be explained by only one strictly associative learning mechanism, as suggested by Frensch and Miner (1994). Based o n the theory of finite automata it was possible to construct experimen tal conditions in which the essential aspects of the two paradigms cou ld be combined independently. In Experiment 1, sensitivity towards gra mmatical structure, measured by grammaticality judgments, was found if subjects had to memorize presented sequences of events in the learnin g phase (grammar learning). No grammaticality effect was found if, pri or to the discrimination task, subjects had to react according to sign aled events in a sequence (serial reaction time task). These differenc es between learning conditions could be due to differences in the appr opriateness of learning processes for the test task, thus reflecting a high degree of specificity of implicitly acquired knowledge. To rule out this alternative explanation, a second experiment was run, in whic h the same learning conditions were combined with the task to react as quickly as possible to signaled events. Comparing the reaction times of critical transitions between grammatical and non-grammatical sequen ces revealed a decisive effect of grammaticality for both conditions. In Experiment 3, it was shown that the effects of Experiment 2 were du e to the learning conditions, rather than artifacts. The implications of these findings for mechanisms of learning are discussed.