Abstract
When searching the visual world, memory must be allocated across eye movements (i.e., saccades) to make efficient use of the visual information available. Past research of transsaccadic memory has focused on comparisons with visual short-term memory (e.g., Bays & Husain, 2008, Irwin 1991, 1992) and measurement of transsaccadic memory capacity in visual search (e.g., Kleene & Michel, VSS 2017). Our project takes a further step, explicitly characterizing the temporal allocation of memory across multiple fixations. In particular, we used a multi-fixation temporal integration task to determine how humans use visual information as a function of (1) the temporal position of a fixation within a sequence of saccades, and (2) the set size, or number of locations to be encoded. In two experiments, participants were asked to report the average luminance polarity ('bright' or 'dark') of a Gaussian blob whose luminance varied over four sequential frames. The blobs were presented at one, two, four, or six locations and the target location was cued at the end of the trial. In Experiment 1, we simulated saccades by inserting blank intervals between stimulus frames, with the observer maintaining a central fixation. In Experiment 2, participants performed a set of actual saccades between two fixation points while the luminance blobs were presented. Using reverse correlation, we characterized the contribution of each display frame to the observer's luminance polarity decision and how the temporal pattern of contributions changed as a function of set size. Results from both experiments show a recency effect. Information from later fixations contributes more to the luminance polarity decision than does information from earlier fixations. However, the strength of this recency effect depends critically on memory load. It is enhanced with increasing set size and attenuated with decreasing set size, virtually disappearing when observers need only encode a single location.
Meeting abstract presented at VSS 2018