| Over the past thirty years, studies in single cell neurophysiology, neural imaging and neuropsychology have spun a debate in the field of visual object recognition on the issue of domain specificity in face processing. There is now a growing body of biological evidence pointing to the presence in the primate brain of mechanisms selective for the processing of faces. In parallel with the notion of domain specificity for faces, the idea of process specificity has also gained prominence. A variety of behavioral findings are consistent with the notion that faces are processed as undivided wholes. This "holistic" processing strategy is thought to be restricted solely to upright faces, whereas inverted faces and objects undergo part decomposition. This Thesis presents three studies that quantify the processing specificity of faces with psychophysical techniques.; In Study I, a summation technique was used to test the perceptual independence of face parts degraded by noise. The results indicate that parts of inverted faces are independently discriminated, whereas performance with upright faces is consistent with the integration of parts into a spatially undivided whole.; Study II explored boundary conditions on performance with noise-degraded faces. Contrary to the results in the spatial domain, there appears to be no difference in temporal integration properties between upright and inverted faces. Furthermore, sensitivity declined non-linearly from the upright to inverted orientation, suggesting a relatively narrow orientation tuning operating in an egocentric frame of reference.; Study III further quantified the characteristics of the orientation tuning of face processing by using transparency displays and contrast matching. The notion of an orientation tuning function for face perception is then discussed in relation to its probable underlying neural substrates. |
