Cooperation between vision and cognition enables humans to interact effectively with objects in the world. Despite its clear importance, far too little is known about one crucial component of these cooperative interactions: visual memory. The project will synthesize concepts and techniques drawn from vision research and from memory research, generating a coherent framework for understanding short-term visual memory, particularly episodic recognition memory. The project's empirical component uses variants of the Sternberg paradigm to assess episodic recognition for stimuli with known similarity relations. These stimuli, 2-dimensional textures created by linearly summing sinusoidal gratings and synthetic human faces, are not ordinarily used to explore memory. Because inter-stimulus similarities are central to theories of cognition, including theories of visual cognition, the ability to manipulate similarity relations among stimuli affords important advantages in research and modeling. ? ? To integrate vision and memory research, the project exploits a variety of empirical and modeling tools, including a mathematical model called NEMO, for Noisy-Exemplar Model. In preliminary experiments, NEMO has been applied successfully to recognition results averaged across test subjects. In order to pursue the project's long-term goal of understanding recognition performance by individual subjects on individual sets of stimulus items, NEMO will be evaluated relative to other alternative models. To take account of individual differences in vision, memory test stimuli will be scaled according to each subject's visual capacity. This should produce more direct assays of memory per se, including assays of any important individual differences in memory function. ? ?
Showing the most recent 10 out of 35 publications
