Temporal Ventriloquism: Perceptual shifts forwards and backwards in time predicted by the maximum likelihood model

Jess Hartcher-O’Brien, School of Psychology, University of Sydney

Abstract
Can shifts in time occur that are analogous to the shifts in space seen in ventriloquism, and does the Maximum Likelihood Estimation (MLE) model account for it? Time discrimination was measured unimodally using 2IFC procedure and a 1.25s tone as a frame of reference. Observers indicated which stimulus (flashed disc or noise burst) was later in the time frame? One stimulus (the reference) was always at the midpoint, the other (comparison) varied adaptively to find subjective midpoint. Resulting psychometric functions yielded subjective temporal midpoint and its variance (precision). In audiovisual trials, one interval contained both components presented asynchronously (±20 or ±40ms) around the midpoint, and the other a synchronised “probe” pair varied in time to find temporal location (and precision) of the asynchronous pair. Auditory temporal precision was better than visual. Localisation precision depended weakly on envelope type (square vs. Gaussian) and duration (10 vs 30ms). Bimodal data generally agreed well the MLE predictions. Bimodal localisation was intermediate between the components, consistent with a weighted sum location based on each component’s precision. Temporal discrimination at this point was usually greater for bimodal stimuli than for either unimodal component, showing that indicating statistically optimal integration of temporal information.

Not available

Back to Abstract