UC San Diego

The interface between living nervous systems and hardware is an excellent proving ground for precision experimental methods and information classification systems. Nervous systems are complex (104 - 1015(!) connections), fragile, and highly active in intricate, constantly evolving patterns. However, despite the conveniently electrical nature of neural transmission, the interface between nervous systems and hardware poses significant experimental difficulties. As the desire for direct interfaces with neural signals continues to expand, the need for methods of generating and measuring neural activity with high spatiotemporal precision has become increasingly critical. In this thesis, I describe advances I have made in the ability to modify, generate, measure, and understand neural signals both in- and ex-vivo. I focus on methods developed for transmitting and extracting signals in the intact nervous system of Hirudo verbana (the medicinal leech), an animal with a minimally complex nervous system (10000 neurons distributed in packets along a nerve cord) that exhibits a diverse array of behaviors. To introduce artificial activity patterns, I developed a photothermal activation system in which a highly focused laser is used to irradiate carbon microparticles in contact with target neurons. The resulting local temperature increase generates an electrical current that forces the target neuron to fire neural signals, thereby providing a unique neural input mechanism. These neural signals can potentially be used to alter behavioral choice or generate specific behavioral output, and can be used endogenously in many animal models. I also describe new tools developed to expand the application of this method. In complement to this input system, I describe a new method of analyzing neural output signals involved in long -range coordination of behaviors. Leech behavioral signals are propagated between neural packets as electrical pulses in the nerve connective, a bundle of neural transmission fibers. These signals control and coordinate sophisticated behavioral motions allowing the animal to combine several stereotypical behaviors to performs actions such as hunting. I developed a blind source separation technique to isolate individual axon activity patterns from noisy, highly overlapping local voltage measurements of the intact bundle. These axon activity patterns correspond to single neural sources. My unsupervised algorithm can extract candidate signal patterns that are undetectable by established techniques. Using a propagation-sensitive electrode I developed, I am able to acquire neural signal velocity and direction information. With this set of information, I successfully identify rhythmically active multifunctional neurons that participate in interganglionic signaling during swimming, crawling, and whole-body shortening. I also demonstrate tracking of single neural sources across distant measurement sites. These results demonstrate a new way to search for behaviorally important biological signals, and help locate and identify signals involved in specific behaviors in Hirudo