Abstract
When two images are simultaneously flashed left and right of fixation, subjects can initiate saccades to the side where a face is present in as little as 100-110 ms (Crouzet, Kirchner & Thorpe, submitted). In the present study, we tested how performance is affected by reducing the size of the target region within the image. Six different scales were used so that the percentage of the pixels in the image that corresponded to the head (not including hair) was set at 20%, 10%, 5%, 2%, 1% or 0.5%. We generated sets of 100 target and distractor images by taking two photographs of the same scene, one with a human present (target), and another without (distractor). On each trial, a fixation cross was presented for 800-1600 ms followed by a gap lasting 200 ms. Then, a target stimulus at one of the six scales was paired with either one of the matched distractors, or one of 500 other highly varied distractor images. 8 subjects were required to saccade towards the side containing a human target. Although accuracy decreased when face size was reduced, overall performance remained surprisingly high, even for the smallest sizes. For example, accuracy dropped from 94.6% to 84.2% when the face was reduced from 20% to 0.5% of the image. Furthermore, average reaction time was under 150 ms for all six sizes, and the minimum reaction time defined as the bin where correct responses statistically outnumber errors was still only 100-110 ms, even for the smallest size. Although there is now increasing evidence that these ultra-rapid saccades towards faces may depend on relatively low level information contained in the power spectrum, the current results demonstrate that this sort of analysis must be performed locally, rather than being a feature of the global power spectrum for the image.