Abstract
The progression of visual encoding effectiveness through time was determined in a visual word recognition task using 200 ms stimuli (5-letter words) with randomly oscillating signal-to-noise ratio (SNR; ranging between 0 and .5; noise component made of white noise). SNR's were generated by combining 5-60 Hz sinewaves (5 Hz steps) of random amplitudes and phases. Stimuli were displayed at a temporal frequency of 120Hz and stimulus energy was equated across image frames. Individual (n = 8) classification images of encoding effectiveness through time were constructed by the weighted subtraction of SNRs leading to errors (average accuracy of 61%) from those associated with correct target identification. They were transformed into Z scores by bootstrapping (1000 iterations) and then averaged across participants. A similar procedure was applied on time-frequency analyses of the SNRs for each trial to obtain a classification image of encoding effectiveness in the time-frequency domain. The temporal classification image showed that visual encoding effectiveness was just above the significance threshold (p < .05; Pixel test; Chauvin et al., JOV 2005) on the first target image frame (8 ms). It then rose steeply up to 50 ms after target onset, to then decline gradually in an oscillating wave with progressively diminishing peaks at 83, 117, and 167 ms. After 183 ms following target onset, encoding effectiveness was no longer significant. The time-frequency classification image showed relatively wide bandwidth (ranging from 20 to 40 Hz in width) significant activity between 0 and 114 ms after target onset with peaks ranging between 15 and 30 Hz. Another significant 55 Hz peak occurred at 144-171 ms SOA. Visual encoding effectiveness is temporally discontinuous and appears to involve at least two oscillating mechanisms, an early one roughly in the range of 15-30 Hz and late one at about 55 Hz.
Meeting abstract presented at VSS 2018