Fig. 3: Computational hypotheses in temporal anticipation.
From: Neural signatures of temporal anticipation in human cortex represent event probability density
a Canonical hazard rate hypothesis. Left: temporal blurring (“tb”) poses that uncertainty in the neural estimation of elapsed time increases linearly with elapsed time, affecting the estimate of event PDF (Methods). Middle: HR computation requires a sequence of three mathematical operations (Methods). Right: The HR hypothesis predicts a linear, mirrored (“mir”) relationship between event probability over time and anticipation (Methods). b PDF-based hypothesis. Left: Probabilistic blurring. Event PDF determines uncertainty in elapsed time estimation. In modeling, the objective event PDF is convolved with a Gaussian kernel whose SD is inversely related to the event PDF, resulting in a subjective estimate of the event PDF: the probabilistically blurred (“pb”) PDF (Methods). Right: Reciprocal probabilistically blurred PDF (“1/pb PDF”) predicts RT. c Effect of reciprocal computation on RT prediction. 1/pb PDF implies a nonlinear “weighting” of event probability in temporal anticipation. For comparison, probabilistically blurred PDF was linearly inverted (“mirrored pb PDF”) to illustrate the effect of the presumed nonlinear reciprocal computation. For comparison, both 1/pb PDF and mirrored pb PDF were scaled by their respective ranges. The y-axis in arbitrary units.
