The effect of different training experiences on object recognition in the visual system

This dissertation is aimed at comparing the effects of different experiences on human visual object recognition. Functional specialization occurs in the visual object recognition system in terms of selectivity for certain categories (e.g., faces, letters, scenes, and body parts) over others in different regions of the ventral occipitotemporal cortex, as well as specific behaviors associated with different categories. A number of factors have been proposed to explain the functional specialization, yet few empirical studies addressed the unique contribution of different factors. In this dissertation I investigated the role of recognition experience by comparing the effects of two types of categorization training. The use of novel objects allowed me to control for stimulus properties and higher-level factors (emotional, social, and linguistic processing). I found that different training experiences resulted in different changes in behavior and neural activity.;This dissertation begins (CHAPTER I) with a review of the functional specialization for object categories, with an emphasis on faces and letters. I describe the different proposed factors affecting category selectivity in the brain, and discuss the difficulties associated with testing the contribution of individual factors using real-world objects. I then discuss the use of novel object training as a useful tool in answering this research question, and the study of more than one type of training as a valuable way to extend the tool.;Following the review of this literature, CHAPTER II reports a novel-object training paradigm inspired by Gauthier et al. (1997, 1998, 1999, 2002) to address one facet of the problem: can different recognition experiences with one set of novel objects lead to qualitatively different behavioral and neural consequences? Two groups of participants went through training procedures designed to mimic experience with either faces or letters. The face-like group learned to categorize novel objects at a subordinate level, similar to how people discriminate among faces in person identification. The letter-like group learned to recognize, at the basic level, multiple novel objects presented in spatially organized clusters with coherent font-like styles, similar to how people process letters when reading a text. The two training procedures led to different behavioral phenomena associated with face perception (e.g., a smaller basic-level advantage and holistic and configural processing effects) and letter perception (e.g., a larger basic-level advantage and faster recognition of objects in a string).;CHAPTER III reports the effects of the two types of training on fMRI activity. The face-like group showed enhanced activations around the right fusiform face area. The degree of change also correlated with the magnitude of the configural processing effect found in the behavioral post-training test. The letter-like group did not show enhanced activations in the letter-selective region as predicted. Instead, activations for this group changed in a more distributed manner. A greater emphasis was put on the medial parts of the ventral occipitotemporal cortex after training. These medial parts also showed a preference for object presentation in the periphery of the visual field, and for basic-level categorization relative to subordinate-level categorization.;I conclude my dissertation with CHAPTER IV in which I discuss the implications concerning the role of experience in forming functional specialization in the visual object recognition system. I also discuss how the comparison of multiple types of training can help answer broader questions concerning perceptual expertise.