NEURAL BASIS OF VISUAL LEARNING

University Of Washington, Seattle WA

Investigators

Linked publications & trials

Abstract

This subproject is one of many research subprojects utilizing the resources provided by a Center grant funded by NIH/NCRR. The subproject and investigator (PI) may have received primary funding from another NIH source, and thus could be represented in other CRISP entries. The institution listed is for the Center, which is not necessarily the institution for the investigator. We study the neural basis of object recognition by recording neural responses from the inferotemporal cortex of macaque monkeys to understand how the neural responses can underlie the remarkable primate capacity for processing images. Object recognition is poorly understood because the processing of visual stimuli depends on the goals of the subject viewing the visual stimuli. This interaction between goals and the processing means that object recognition (and its neural basis) depends on an interactive network in the primate brain. In the projects completed during this year we found evidence that this interactive process plays an important role in vision and in the neural responses in IT. Many neurons in IT cortex are sufficiently sensitive to explain the discrimination capacity during a demanding behavioral discrimination task (Liu and Jagadeesh, 2008a), suggesting that small populations of neurons in IT might provide the information for discriminating object. Furthermore, the discrimination capacity is significantly reduced when the similar images are presented in the absence of a specific behavioral task (Liu and Jagadeesh, 2008a), suggesting that the goals assigned during the viewing of stimuli modulates the neural populations. In addition, neural responses to ambiguous images are modulated by the active interpretation of the images, suggesting that conclusions drawn by the observer can alter the neuron's responses to the images (Liu and Jagadeesh 2008b). In some circumstances, the interactive processing of images may depend on the properties of the network in IT converging to stored or predicted states within the system through the operation of a network that can be simulated by an attractor network with stored representations (Akrami et al, 2008). Finally, we have found that adaptation seems to affect only the early part of response in IT, leaving the "stored" representation to which the neural response converges unaltered (Liu et al, 2009).