Our research focuses on the interface between visual perception and decision-making, with a special emphasis on learning. For example, consider a medical student learning to diagnose breast cancer on mammograms. At first, each mammogram appears as an undifferentiated blob similar to any other mammogram. With practice, however, the student learns to identify cancerous formations with high reliability. The question is, what has changed in the brain of the expert relative to that of a novice? How do these changes occur? In addition to their practical importance, questions such as these are of great theoretical interest.
Even the simplest perceptual task involves significant decision-making components. In the above example, if one type of cancer is 10 times more prevalent than another, an experienced radiologist takes these base rates into account while inspecting an image. Even in the controlled conditions of the psychophysical laboratory, factors such as response biases (e.g., Petrov & Anderson, 2005), decision strategies (e.g., Petrov, 2009), and speed-accuracy tradeoffs (e.g., Petrov, Van Horn, & Ratcliff, 2011) affect the threshold and accuracy measurements. Therefore, the decision-making components must be considered even when one is primarily interested in the perceptual aspects. Conversely, the data on decision-making and categorization depend critically on the low-level properties of the stimuli, and factors such as signal-to-noise ratios (e.g., Petrov, Dosher, & Lu, 2005) and Weber fractions (e.g., Petrov, 2008a) must be considered even when one is primarily interested in the cognitive aspects. All too often, vision scientists design their experiments to minimize the cognitive aspects, whereas cognitive psychologists design their experiments to minimize the perceptual aspects. This is a valid strategy that has produced a wealth of knowledge in each discipline. The time has come, however, to integrate the two and explore the relationships between them.
This integrative thrust is a guiding principle of our research program. We have solid expertise in both disciplines and attend regularly the conferences of the Vision Sciences Society (VSS), on one hand, and of the Psychonomic, Mathematical Psychology, and Cognitive Science societies, on the other. In our laboratory, we measure how accuracy, response times, and eye-movement patterns change with extended practice on various visual tasks. We develop models that implement specific hypotheses about the underlying mechanisms, and test the quantitative predictions of these models against the behavioral data. Such models are valuable conceptual tools for bridging the gap between brain and behavior.