Applications of statistical signal processing and digital image processing in the development of a mathematical model relating chemosensory electrophysiology and feeding behavior in Manduca sexta

Two sets of biological experiments were conducted toward the development of a mathematical model relating discharge rate of peripheral chemosensory neurons to feeding behavior in the tobacco hornworm (larva I Manduca sexta). The first set examined behavioral responses to combinations of two chemical compounds, one of which stimulated (inositol) and the second of which deterred [Rosmarinic Acid (RA)] feeding in this organism. In the second set, the electrophysiological responses of specific peripheral taste neurons were recorded following stimulation by the same binary mixtures that were presented in the behavioral experiments.; Two technical contributions were developed to address the goals of the biological study. First, an optimal pattern recognition algorithm was developed for identifying the activity of individual neurons whose signals were mixed together in the extracellular recordings. This algorithm was technically superior to the techniques previously used and addressed a general neuroengineering problem that occurs in the analysis of extracellular recordings of neural activity in nerve fibers or cortical tissue. As such the findings may have applications in the development of neural prosthesis. The second contribution involved the implementation of a computer-controlled apparatus that automatically quantified feeding behavior. Specifically, software was developed that controlled camera positioning and image digitization and employed techniques of digital image processing to automatically estimate food consumption. These techniques can aid in the development of alternative pest management strategies for reducing the use of neurotoxic pesticides, which represent health hazards for humans.; Several conclusions were reached from the experimental results. First, RA was confirmed as a strong feeding deterrent, even at concentration levels that were a factor of 30 lower than inositol. The second finding was an unexpected peripheral interaction where the presence of RA, at virtually any level, decreased the firing rate of the inositol-sensitive neuron. A third finding was that moderately high levels of inositol mitigate RA's feeding deterrence to a greater degree than predicted given the mild stimulatory effects of the unary mixtures of inositol. A linear relation between chemosensory firing rate and feeding behavior was not inconsistent with these data.