Abstract
Humans are highly skilled at recognizing patterns embedded within random stimuli. Gold, Aizenman, Bond & Sekuler (2013) generated such stimuli in order to assess memory and learning with rapidly presented sequences of quasi-random luminances. Adapting their paradigm, we tested the hypothesis that visual perception is more efficient in processing spatially distributed stimuli than time-varying ones like Gold et al.’s. In each of three experiments, subjects judged whether a subset of eight quasi-random luminances repeated or not. Experiment One compared detectability of within-stimulus repetitions embedded in spatially arrayed stimuli or in stimuli presented as a sequence of luminances. Subjects performed significantly better with spatial arrays than with temporally distributed sequences, even when the spatial arrays were presented for just a tiny fraction of the duration of a temporal sequence. Also, unlike what Gold et al. found with temporal sequences, performance was particularly good when the two halves of a spatial array were mirrored versions of each other. In Experiment Two, concurrent temporal and spatial information decreased performance relative to that seen when stimuli were presented in spatial mode alone. Finally, Experiment Three assessed learning for particular stimuli made to recur intermittently, interspersed throughout a block of trials. Performance improved steadily when a stimulus recurred, and did so at about the same rate for spatial and temporal stimuli, demonstrating equivalent accumulation of information in longer-term memory. Overall, our results show that detection of within-stimulus repetitions is better when items are presented spatially, in both mirrored and non-mirrored form, than when they are presented as a temporal sequence. Importantly, the incidental learning with both kinds of stimuli shows that, without motive or instruction, information from stimuli cumulates over successive recurrences. Keywords: Short-term memory, incidental learning, spatio-temporal sequences
Meeting abstract presented at VSS 2015