In the time-window-of-integration (TWIN) model framework multisensory integration occurs only if the peripheral processes elicited in different sensory channels terminate within temporal proximity of each other [1]. The model has been successful in predicting mean reaction times (RTs) to crossmodal stimuli under a host of different conditions including variations of stimulus intensity, spatial and temporal correspondence, and instruction, i.e., focused attention vs. redundant signals paradigm [2].

Here we investigate the model’s predictions about reaction time variability as measured, in particular, by RT variance. Because occurrence of multisensory interaction, that is, facilitation or inhibition, depends on the outcome of the race in the first, peripheral stage subsequent stages are not statistically independent of the duration of the first stage. Although the effect is not directly observable, it can be shown that under rather general conditions this statistical dependence is positive for multisensory inhibition and negative for facilitation. In addition, we derive explicit expressions for RT variance in the ex-Gaussian model version and also investigate effects of stimulus intensity and integration window size on RT variability.

time window; reaction time; variability