Purchase this article with an account.
Oh-Sang Kwon, David Knill; Humans adaptively use temporal correlations in stimulus history to estimate velocity. Journal of Vision 2011;11(11):988. doi: 10.1167/11.11.988.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Motivation: We have previously shown that perceptual estimates of stimulus velocity are biased toward both the velocity mean of previously viewed stimuli and the velocity of the immediately preceding stimulus (Kwon and Knill, VSS 2010). The bias toward the mean was modulated by the variance of stimulus velocities, consistent with optimal statistical inference. Subjects' behavior, however, showed that they assumed temporal correlations that didn't exist in the stimulus set. Here, we tested whether observers can adapt their estimation strategies to take into account different temporal correlation functions in stimulus history. Method: We used a motion extrapolation task in which a target moved and disappeared behind an occluder and subjects had to hit the target when it was supposed to be in the designated hitting zone. The velocity of the target was randomized from trial to trial, but could have either a positive or negative correlation with the previous velocity (r = .6 or −.6). The occluder width was randomly varied in such a way that there was no temporal correlation in target's occlusion time from trial to trial. Results: As before, the timing of subjects' hitting behavior showed velocity biases toward the mean velocity of the stimulus set. In the positive correlation condition, they showed a strong bias toward the velocity of the immediately preceding stimulus, but the bias disappeared in the negative correlation condition. Subjects' behavior was well fit by an optimal model that adaptively took into account sensory uncertainty in velocity estimates, occluder width and the temporal statistics of the stimulus sequence, but with a strong positive bias in the estimated temporal correlations in the stimuli. Conclusions: The CNS accurately and adaptively accounts for the temporal statistics of stimuli when estimating velocity, but has a positive bias in its estimated temporal correlation function.
This PDF is available to Subscribers Only