Unsupervised neuronal and perceptual learning of invariant object representation

点击翻译按钮获取中文摘要

DESCRIPTION (provided by applicant): Visual object recognition (categorization and identification) is one of the most fundamental cognitive functions for our survival. The inferior temporal cortex (IT) is the gateway to brain's decision and memory centers and it conveys visual object and category information in a manner that is largely tolerant ("invariant") to the exact position, size, pose of the object, illumination, and clutter. We recently found that one type of tolerance ~ position tolerance ~ can be rapidly altered by targeted manipulation of the statistics of natural visual experience. This newly discovered type of neuronal learning suggests that IT learns tolerance from the spatiotemporal contiguity of natural visual experience, as different images of the same object tend to play out smoothly on our retina during natural vision. The overarching goal of this proposal is to examine the role of natural visual experience in shaping the IT object representation, and the perceptual consequences of that experience in the same subjects (non-human primates). Animals will acquire experience in an artificially altered visual world where we make specific changes to the spatiotemporal contiguity of their visual experience. Intermittently, we will measure animals' IT neuronal position tolerance and perceptual tolerance by means of single-electrode recording and object discrimination task to track any changes produced by that experience. Our first aim is to examine whether the targeted, experience-driven changes in IT position tolerance is accompanied by specific, stable changes in object perception. Our second aim is to examine whether the naturally acquired spatiotemporal experience instructs learning of tolerance ("invariance") to other identity-preserving image variations (object size, 3D pose). Together, the outcome will elucidate the potentially key role of unsupervised experience in the development of "invariant" object representation. IT projects directly to brain areas responsible for decision, action, and memory, thus an understanding of the IT object representation allows us to understand the basic building blocks of memory and cognition. IT is analogous to structures in the human brain: area Lateral Occipital Complex (LOC), Fusiform Face Area (FFA), and Parahippocampal Place Area (PPA) have been implicated in the processing of objects, faces, and places. Deficits in recognition are symptoms of many neurological disorders: e.g., agnosia, Alzheimer's, autism, dyslexia, thus an understanding of the circuitry underlying the recognition process is likely to provide insight into their conditions.