hide
Free keywords:
Bayesian modeling; Decision making; Response time models
Abstract:
The popular random-dot motion (RDM) task has recently been applied to multiple-choice perceptual decision-making. However, changes in the number of alternatives on an RDM display lead to changes in the similarity between the alternatives, complicating the study of multiple-choice effects. To disentangle the effects of similarity and number of alternatives, we analyzed behavior in the RDM task using an optimal-observer model. The model applies Bayesian principles to give an account of how changes in the stimulus influence the decision-making process. A possible neural implementation of the optimal-observer model is discussed, and we provide behavioral data that support the model. We verify the predictions from the optimal-observer model by fitting a descriptive model of choice behavior (the linear ballistic accumulator model) to the behavioral data. The results show that (a) there is a natural interaction in the RDM task between similarity and the number of alternatives; (b) the number of alternatives influences “response caution”, whereas the similarity between the alternatives influences “drift rate”; and (c) decisions in the RDM task are near optimal when participants are presented with multiple alternatives.
