The Fourth Asian Conference on Vision (ACV2006)i]j

Effects of Backward Masking on Discrimination Performance of Facial Identity and Facial Expression
Takayuki Wakatsuchi1, Mikio Inagaki2, Ichiro Fujita2
1 School of Engineering Science, Osaka University, Osaka, Japan
2 Graduate School of Frontier Biosciences, Osaka University, Osaka, Japan

1. Introduction
The face of an individual provides information about his/her identity and facial expression. Perception of facial identity and expression is suggested to depend on separate systems at some points in neural processing. In this study, we examined temporal characteristics of perceptual processes for discriminating facial identity and for discriminating facial expression.

2. Methods
By using a backward masking paradigm, we assessed elapsed time needed to discriminate identity and expression of faces. A face stimulus (cue stimulus) with a backward mask stimulus was followed by 4 face stimuli (test stimuli) after a delay (100 ms). In identity discrimination trials, the test stimuli were faces of 4 different persons with the same facial expression. In expression discrimination trials, the test stimuli were faces of 1 person with 4 different facial expressions (neutral, happy, fear, anger). Subjects were asked to choose the same face as the cue stimulus among the 4 test stimuli. We can control the effect of masking by varying the interval between a cue stimulus and a mask stimulus (stimulus onset asynchrony, SOA); shorter SOAs generally cause stronger masking effects. Task difficulty was controlled by varying the contrast variance of the cue stimulus.

3. Results
Psychometric functions were obtained by plotting correct performance rates against the contrast variance for identity and expression discrimination tasks with different SOAs. We defined a psychophysical threshold as the contrast variance yielding 72% correct responses. At short SOAs (23, 35, 47 ms), the threshold for the expression discrimination task was lower than that for the identity discrimination task. The ideal observer analysis showed that the difference in discrimination performance between the two tasks was not attributable to graphical differences in the cue and test images between the two experiments. At a long SOA (940 ms), no difference was observed in the thresholds between the two tasks.

4. Conclusion
The results showed that people can discriminate facial expression with shorter duration of stimulus presentation. We suggest that the neural representation for facial expression is more quickly formed than that for facial identity.

This work was supported by a MEXT grant (17022025)
Correspondence: Ichiro Fujita Ph. D. (