Connecting object recognition performance, adaptation after effects, and form processing strategies in high-level vision

Abstract

Psychophysical techniques such as adaptation after effects are commonly used to investigate face and object recognition, and have played an important role in the development of models of high-level visual shape coding. Yet the nature of the relationship between these behavioural effects and the underlying neuronal processes is currently unclear. An open question concerns the extent to which face and object after effects reflect the properties of high-level shape coding, as opposed to other stages of visual processing. A related question concerns what aspects of high-level coding can be inferred from after effects. This thesis reports four psychophysical investigations of face and object recognition, high-level shape coding, and adaptation after effects in high-level form vision. The first study introduces a new test of within-class object recognition-the Cambridge Car Memory Test-and provides norms for the test based on a sample of 153 young adults. The second study investigated whether face eye-height after effects would be correlated with face recognition ability, based on the theoretical notion that face after effect magnitude should reflect the sensitivity of neuronal face-space coding. Results showed a modest but significant correlation between face eye-height after effects and recognition ability, arguing that eye-height after effects can tap processes which contribute to face recognition ability. The third study investigated the extent to which global face distortion after effects tap face-level processes, as opposed to shape-generic processes. Results showed complete transfer of after effects from a scrambled face adaptor to non-face test stimuli, but only partial transfer from an intact face adaptor to non-face test stimuli, arguing that global face distortion after effects have both face-specific and shape-generic components.The final study investigated the coding strategy underlying perception of horse shape and tested whether adaptation to horses can enhance subsequent discriminability of horse shape. Results were consistent with norm-based coding with linear response functions and showed that adaptation to horses can improve discriminability of horse shape. In sum, the evidence reported in this thesis argues that face and object after effects can be a useful method of investigating high-level visual shape coding, but that these effects should be interpreted with caution given that they will generally reflect the combined activity of multiple stages of processing.