Abstract
Face identification relies on the use of horizontally-oriented facial structure. The degree of horizontal bias (HB) is correlated with overall face identification accuracy, as well as the size of the inversion effect, and HB to inverted faces increases after practice with identifying inverted faces. In addition, faces elicit an EEG response sensitive to horizontal structure: When the horizontal structure of upright faces is removed, the N170 is delayed and reduced in amplitude, and the N250 also is reduced in amplitude. Furthermore, the N250 amplitude difference to horizontal vs. vertical structure is correlated with upright face identification accuracy. Here, we asked if the sensitivity of the N170 and N250 to horizontal structure in inverted faces changes with practice. Eleven participants completed a one-of-six inverted face identification task. Over the course of 2 sessions, identification accuracy improved from 56% to 78% (compared to >90% accuracy for upright versions of the stimuli). Before and after training, we measured identification accuracy and ERPs to vertically or horizontally filtered faces of varying filter bandwidths. Response accuracy improved most when horizontal structure was visible, and post-training HB correlated with unfiltered face identification accuracy. N170 amplitude, but not N250 amplitude, decreased when horizontal structure was removed, but these effects of orientation were not influenced by training. Neither N170 or N250 amplitude HB were correlated with unfiltered, inverted face identification accuracy. N170 latency increased with the removal of vertical or horizontal structure, but the difference between orientations was not significant before or after training. Nonetheless, post-training N170 latency HB was correlated with unfiltered face identification accuracy. Overall, our results indicate that although a relatively short amount of practice with inverted faces improves behavioural performance and HB, such training does not differentially alter the sensitivity of the N170 or N250 to horizontal and vertical facial structure.
Meeting abstract presented at VSS 2018