ON THE ROLE OF PURE INSERTION AND SELECTIVE INFLUENCE IN REACTION TIME THEORIZING: EXTENSIONS TO THE DISTRIBUTIONAL LEVEL

F. GREGORY ASHBY, Purdue University

Abstract

Choice reaction time (RT) models have tended to rely heavily on assumptions of pure insertion and selective influence. Even so, rigorous tests of these assumptions are scarce. This work attempts to develop such tests by extending these assumptions from the mean level to the distributional level. In the case of pure insertion and under the additional assumption that the inserted stage duration is exponentially distributed, several tests are suggested. One simply involves checking a condition on the observable RT density functions and another provides an estimate of the unobservable exponential rate when the assumptions are supported. This latter test is shown to be applicable not only when processing is serial but also for certain parallel models. In addition, discrimination between self-terminating and exhaustive search strategies is provided by the test, and in the case of either, both parameter estimation and tests of the model are possible. The existence of nonequivalent mimicking models is considered, with emphasis on McClelland's (1979) Cascade model. Extensions to nonexponential models are investigated and a general method of moments solution is outlined in the parametric case. Finally several general tests of pure insertion are developed in the case when no distributional assumptions are made. Several of these results are then provided an illustrative application with data from a standard memory scanning experiment. The results provide tentative support for the double assumption of pure insertion and that the inserted stage duration is exponentially distributed. In the case of selective influence, general nonparametric alternatives to Sternberg's (1969) additive factor method are developed in the case when either a serial or parallel stage is influenced. Mimicking models are again considered and the Cascade model is found more capable of mimicking the selective influence results than those associated with pure insertion. Distributional assumptions are then briefly considered. Finally, several empirical applications are reported. In the last of these, it is shown how the method can be used to identify the stage or stages which form the locus of processing errors in experimental tasks evoking substantial inaccuracies.

Degree

Ph.D.

Subject Area

Psychology|Experiments

COinS