Understanding face perception through the contributions of bottom-up and top-down processing : a psychophysical approach

• Main Author: Pan, Liu
• Other Authors: Xu Hong
• Format: Theses and Dissertations
• Language: English
• Published: 2015
• Subjects: DRNTU::Social sciences::Psychology::Experimental psychology
• Institution: Nanyang Technological University

Face perception (recognition of facial expression and face identity in particular) plays a remarkable role in our daily social interaction. This research aims to understand the face-coding mechanism using a psychophysical approach, with three studies using the face adaptation paradigm to investigate face-coding machinery from both bottom-up and top-down perspectives. Study 1 addressed the holistic nature in face adaptation. Through a series of experiments, it revealed that the facial expression aftereffect produced by a whole face was significantly greater than the simple linear summation of the aftereffects induced by its two components (an isolated mouth part and a mouthless face part). This "gestalt" effect in facial expression adaptation was consistently observed with upright cartoon faces (Experiment 1.1) and upright real faces (Experiment 1.3) but was eliminated in upside-down faces (Experiment 1.2). These findings point to a hierarchical multilevel face-coding framework where neural responses from low-level visual areas are projected via nonlinear functions to higher-level regions in forming the holistic face representation. Study 2 investigated the different roles of top and bottom half faces in facial expression and face identity adaptation. We used a whole real face and its two size-equivalent components (a top-half face part and a bottom-half face part) as the adaptors in both facial expression and face identity adaptation tests. Results reproduced the gestalt effect in both types of face adaptation and also showed that the bottom-half face part contributed most to facial expression adaptation whereas the top-half face part made major contribution to face identity adaptation. This finding suggests a multidimensional face-coding framework with dissociable mechanisms, with facial expression and face identity employing distinctive strategies in recruiting facial properties towards the holistic representation. Study 3 examined the role of attention in face adaptation, by using the orientation pattern of microsaccades to index the deployment of attention during the adaptation. We showed that the facial expression aftereffect remained unaffected when the allocation of attention towards the adapting face was not successfully increased by a brightness-change-detection task, whereas it was remarkably reduced when a target-letter-detection task successfully diverted observers' attention away from the adapting face. This finding demonstrates the prominent influence of top-down attention on facial expression adaptation, pointing to a face-coding framework with adaptive plasticity modulated by top-down mechanisms. Taken together, findings in this research shed light on a more comprehensive understanding of the neural coding mechanism underlying face perception.