Abstract
Perceptual decision-making was studied with stimulus sequences of visual, vibrotactile, and concurrent visual-vibrotactile pulses. Subjects categorized the beat frequency of both stochastic and deterministic stimuli. We evaluated the rate at which perceptual information accumulated over successive samples, and the robustness of signals from different modalities. METHODS. A handheld computer tablet delivered 30 msec visual orvibrotactile unimodal pulses, or both together in synch (bimodal). Visual stimuli were Gabor patches that turned on and off; vibrotactile stimuli were intermittent vibrations generated within the tablet and delivered to subjects’ hands. Two beat rates, 3 and 6 Hz, were randomly intermixed; subjects had up to 1.6 seconds to sample and categorize stimulus rate. On 2/3 of trials, beat rates were stochastic: intervals between pulses were perturbed by samples from zero-mean Gaussian distributions with different variances. The random variation in beat rate was designed to influence the confusability of the two nominal rates. RESULTS. As the temporal randomness of beats increased, accuracy fell and judgment time lengthened. These changes were similar across modalities. Subjects behaved impulsively, rarely using all available information. Logistic regression showed that subjects’ judgments were disproportionately influenced by the first one or two beats they experienced. With stochastic stimuli, subjects seemed to adjust their decision criteria in real time. Additionally, judgment times for bimodal stimuli suggested more rapid accumulation of information, a result confirmed by drift diffusion analysis. CONCLUSIONS. The robustness of temporal information conveyed by visual pulses is on par with that conveyed by vibrotactile pulses. The speeded responses produced by bimodal stimulation are consistent with theories in which timing information is processed by transmodal neural mechanisms. Finally, our results suggest that multimodal stimulation may have particular value for directing time-sensitive tasks.