Abstract
Background: There is considerable evidence that faces are processed as ‘wholes’ in the human visual system; however, there is also evidence of a feature-salience hierarchy, in that some features contribute more than others to facial percepts. Whether this is true of the neural representations of faces is not known, but a question that can be explored through the use of face adaptation. Objective: We used a perceptual bias technique to determine if there are differential contributions to identity aftereffects from the lower face, upper face, and eye region. Method: We selected two unfamiliar face pairs from the HVEM face database with equivalent physical similarity, as determined by a Bayesian ideal observer technique, and created full-face morphs between each of these face pairs with 2.5% increments, which would serve as probe stimuli. For adapting images, we used (1) the whole unmorphed faces, (2) divided these unmorphed faces into upper and lower halves, and (3) isolated a horizontal band containing the eyes alone. In the first experiment we compared the magnitude of aftereffects generated by whole faces, upper faces and lower faces. In the second experiment we compared the aftereffects from whole faces, upper faces and the eye-band. Results: Upper faces generated aftereffects that were not statistically different from whole faces, while lower faces did not generate significant aftereffects. The eye region generated aftereffects that were not statistically different from those generated by the upper face. Conclusions: The upper face and in particular the eye region form a dominant component of the aftereffects for facial identity, suggesting an important role for these regions in the neural representation of facial identity. This is consistent with evidence from discrimination experiments for a feature-salience hierarchy in human face perception.
Meeting abstract presented at VSS 2013