Abstract
We can easily understand the visual environment despite our eyes moving to take in new information three to four times per second. This rapid information flow may initially be registered in iconic memory, a brief high-capacity store containing literal visual representations. Evidence for higher-level semantic representations in iconic memory has not been demonstrated and it has been proposed that such representations may only be accessed in subsequent stages of memory, for example, visual short-term memory and phonological short-term memory stores. Two experiments aimed to investigate whether visual and higher-level semantic representations can be accessed in the very early stages of visual memory by using comparable iconic memory tasks. A novel partial-report paradigm was used where participants were briefly shown arrays of six pictures at various exposure durations (50 ms, 150 ms, and 250 ms). Following array offset, a cue specified full report (recall of all six pictures) or partial report (recall of only one cued picture). Experiment 1 investigated whether location information (a pointer to a spatial location) could cue recall. Experiment 2 investigated whether semantic information (instruction to report a picture belonging to a particular semantic category) could cue recall. In both experiments, performance was superior in the partial-report condition compared with the full-report condition (partial-report superiority). These results suggest that several pictures from each array must rapidly gain access to a store in which visual representations are attached to spatial locations and access higher-level semantic information. In both experiments, recall was unaffected by an increase in exposure duration from 50 ms to 150 ms but it improved significantly with the longer 250 ms exposure duration. The evidence for both early visual and higher-level semantic processing in iconic memory tasks challenges the initial conceptualisation of iconic memory and its role in subsequent stages of memory.
Meeting abstract presented at VSS 2016