Purchase this article with an account.
Jianhua Wu, Hong Xu, Peter Dayan, Ning Qian; Motion-gradient defined facial expressions and the nature of face representation. Journal of Vision 2009;9(8):513. doi: 10.1167/9.8.513.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Face perception and motion analysis are standard functions of the ventral and dorsal visual pathways, respectively. Although it has long been recognized that the separation of the two pathways is far from absolute, little is known about the relationship between face and motion processing. Based on the well-known phenomenon of motion mislocalization, we show that without varying any form cues, happy and sad facial expressions can be solely defined by local motion gradients in second-order cartoon faces. Moreover, we probe face representations by measuring how these second-order cartoon faces interact with first-order cartoon and real faces in cross-adaptation experiments. We find that the pattern of interactions is better explained by statistical similarities and differences between the backgrounds of the adapting and test stimuli than by the common notion of separate mechanisms for first- and second-order processing. Our work suggests that both form and motion cues, and both local features and background statistics, contribute to face representations.
This PDF is available to Subscribers Only