Abstract
Humans are voracious recognizers of visual patterns, an ability that depends upon cooperation between perception and memory. Recent studies of auditory pattern recognition showed that after just a few encounters, observers develop and maintain an implicit memory for a particular sample of purely random noise (Agus et al., 2010). To explore the generality of those findings, we adapted Agus’ approach to study the perception of random spatio-temporal visual stimuli. Specifically, we generated 1000 ms, 8 Hz samples of 1D and 2D Gaussian temporal contrast noise whose luminance levels were either uncorrelated across the entire stimulus duration (‘Noise’ trials) or identically repeated between the first and second 500 ms of a stimulus (‘Repeat’ trials). Presented with equal numbers of interleaved ‘Noise’ and ‘Repeat’ stimuli, observers judged each as a ‘Repeat’ or not. Additionally, half of the ‘Repeat’ trials comprised a fixed sample of repeated noise generated at a session’s start (‘Fixed repeat’ noise). Similar to results from the auditory domain, d’ was significantly higher for ‘Fixed repeat’ than ‘Repeated’ noise. Trial-by-trial performance with the ‘Fixed repeat’ noise gradually improved across a session, but the improvement was much more gradual than was seen in the auditory domain. In another experiment, the repeated half of the contrast noise sequence was presented in reverse temporal order, creating stimuli with temporal mirror image symmetry. This stimulus proved to be impossible for observers to learn, likely because of limits on working memory. Reverse correlation analyses of our main results reveal that observers placed relatively more weight upon information from a stimulus’ temporal mid- and end-points. Our results show that observers can reliably develop implicit visual memories of random temporal sequences with a relatively limited amount of exposure, and that such memories are remarkably resistant to interference from multiple exposures to other temporal sequences.
Meeting abstract presented at VSS 2012