9622297 Simmons The general objective of the proposed research is to identify activity in the brain that causes perceptual images to have their particular content. The specific question is how specific numerical values for parameters of neural responses manifest themselves in corresponding numerical values for perceptual dimensions of images. This project uses target ranging in the biological sonar of the big brown bat, Eptesicus fuscus, as a "test-bed" for addressing questions about the formation and content of perceived images. Sonar and radar systems offer an especially well- developed body of theory about signal-processing and display of information that can be used to understand the mechanisms of biological systems evolved for the same purpose. Moreover, the exceptional real-time performance of biosonar systems in well-defined tasks of target localization, tracking, and classification make it obvious that there are technological advantages to understanding how echolocating animals achieve this performance. The starting point for the project is that bats have unusually good echo-delay accuracy of 10-15 ns and "two-point" echo-delay resolution of s. The initial problem is to account for this seemingly implausible performance with neural responses, which from physiological single-cell recording experiments appear to be 2-3 orders of magnitude less precise. New physiological experiments in the bat's midbrain and auditory cortex reveal that the "missing" high-precision temporal information is carried by the latencies of multi-cell neural responses using expanded time scales. There are two prominent competing hypotheses about neural substrates of perception which differ in the relative importance of response-rate and response-timing for determining image content. These hypotheses yield widely- divergent predictions from neural data for what bats should be able to perceive about target range. The combination of behavioral and physiological data from Ep tesicus leads to rejection of the hypothesis that relies exclusively on response-rate to represent information in images, while sustaining the hypothesis that response-timing may be "read out" directly into perception. The proposed research will examine these competing hypotheses in more detail because rejection of the response-rate model has widespread implications for neuroscience and for bioengineering, specifically for development of man-made systems that seek to emulate animal sonar performance. The proposed experiments will use 3-D video reconstructions of the bat's flight in sonar-guided interceptions to quantify how well the bat can locate and recognize targets in complex multiple- target sonar "scenes." New psychophysical experiments using electronically-simulated targets will measure the structure of the bat's sonar images to determine whether specific combinations of single and multiple overlapping echoes delivered from different directions lead to predicted echo- delay values appearing in the images. New physiological experiments will record and analyze the response in the bat's inferior colliculus that carry fine echo-delay information on expanded time scales to learn how these responses are combined to form images. A large-scale computational model of the bat's sonar will be used to evaluate experimental data in the context of emulating the bat's performance. ***
