Figure 1.

Experimental paradigm and behavioral results. A, Event-related three-choice visual categorization task where subjects were instructed to select the category (face, car, or house) of the image after each stimulus presentation. On each trial, the image was briefly presented for 100 ms, following by a 2–2.5 s decision period. Subjects responded with different button presses to indicate their choice. B, Example of images at two phase-coherence levels. Images were presented at high-coherence (50%) and low-coherence levels (35%), where high coherence indicates high stimulus evidence, i.e., an easy decision. C, Mean RTs for face trials and nonface trials at the high-coherence and low-coherence levels. RTs were significantly modulated by the amount of stimulus evidence. At the high-coherence level, mean RT for faces was significantly lower than that for nonfaces. D, Behavioral accuracy for face trials and nonface trials at two coherence levels. Low stimulus evidence led to less accurate decisions. At the high-coherence level, accuracy for faces was significantly higher than for nonfaces. Though subjects responded significantly faster and more accurately to faces than to nonfaces when stimuli were presented at high coherence, their behavioral performance was not significantly different between faces and nonfaces when stimuli were presented at the low-coherence level. Error bars indicate the SEM. Asterisk (*) indicates significant difference at p < 0.05.