Cross-modal identification: effects of contingent changes in the stimulus series.

Two computer controlled experiments in an olfactory cross-modal matching task, using two-component odour mixtures matched against bar diagrams, were designed so that stimulus presentation was contingent upon the recent performance of the subject; stimuli that were relatively poorly (in experiment 1) or well (in experiment 2) matched were more frequently presented. Analysis shows that the autoregressive structure of the performance is modified by such contingent presentation and that there is a weak relationship between transmitted information in matching and the time series structure of the matching errors. It is suggested that the process is nonlinear.