Abstract
The visual system is constantly bombarded with continuous streams of information. As such, a primary challenge in visual perception is determining which stimuli should be integrated or segregated across time. Take the example of a film being presented at the standard rate of 48Hz. Subjectively, we perceive the individually presented frames as a continuous stream of information. However, if that film strip is slowed to a presentation rate of 1Hz, we then perceive the frames as individual units. Critically, there seems to be a threshold presentation rate that characterizes a shift in the perception of the film as continuous stream of information to the perception of segregated frames. This study focused on the temporal integration of visual word recognition. The experiments employed a novel word identification paradigm wherein target words were broken down into parts and presented, with noise, along a rapid serial visual presentation (RSVP) stream (e.g., HXMX, XOXE, XOMX, HXXE, would be a typical presentation stream for the target word ‘HOME’). Experiment 1 demonstrated that changing the presentation rate (43Hz, 22Hz and 11Hz) of the RSVP stream led to qualitatively different approaches to target integration. At fast rates, all target letters were integrated automatically. However, at slow rates, target identification necessitated a conscious, ‘string building’ strategy. Experiment 2 validated the claim for qualitatively different approaches to integration; performance on a concurrent working memory task was impaired only at the slow presentation rates. Together, the results suggest that temporal integration in visual work recognition works according to ~100ms integration windows. Integration beyond that window is still possible, but necessitates top-down working memory influences.
Meeting abstract presented at VSS 2013