The bare bones of object recognition: implications from a case of object recognition impairment

Three experiments were designed to investigate the performance of a patient (RK) who could name objects when presented in conventional views but showed catastrophic failures in identification from unconventional views. The aim of all three experiments was to assess the properties of the central representations that allow recognition of objects presented in conventional but not unconventional views. All three experiments showed that RK had problems in object identification not apparent from his naming performance. In the first experiment, RK was found to be extremely impaired at recognising the parts of objects even though he could name the whole object. In the second experiment, alterations in colour, shape and parts of objects were undetected in stimuli that he could name. In the third experiment, RK showed considerable difficulty with mirror-images and inversion tasks. The explanation for RK's impaired object recognition could not be attributed to defects to his early visual processing. We argue that RK's recognition is achieved through abstract (object-centred) representations that are global rather than local, and quite independent of their spatial framework. These abstract representations we take to be the essential bare bones for object recognition.