Journal of Computational Neuroscience 17, 137–147, 2004 c  2004 Kluwer Academic Publishers. Manufactured in The Netherlands. A Neural Network Model of Chemotaxis Predicts Functions of Synaptic Connections in the Nematode Caenorhabditis elegans NATHAN A. DUNN AND SHAWN R. LOCKERY ∗ Institute of Neuroscience, University of Oregon, Eugene, OR 97403, USA ndunn@cs.uoregon.edu shawn@lox.uoregon.edu JONATHAN T. PIERCE-SHIMOMURA Ernest Gallo Research Center Suite 200, University of California San Francisco, Emeryville, CA 94608, USA jonp@egcrc.net JOHN S. CONERY Department of Computer Science, University of Oregon, Eugene, OR 97403, USA conery@cs.uoregon.edu Received September 8, 2003; Revised February 6, 2004; Accepted April 15, 2004 Action Editor: Karen Sigvardt Abstract. The anatomical connectivity of the nervous system of the nematode Caenorhabditis elegans has been almost completely described, but determination of the neurophysiological basis of behavior in this sys- tem is just beginning. Here we used an optimization algorithm to search for patterns of connectivity suffi- cient to compute the sensorimotor transformation underlying C. elegans chemotaxis, a simple form of spatial orientation behavior in which turning probability is modulated by the rate of change of chemical concentra- tion. Optimization produced differentiator networks capable of simulating chemotaxis. A surprising feature of these networks was inhibitory feedback connections on all neurons. Further analysis showed that feedback reg- ulates the latency between sensory input and behavior. Common patterns of connectivity between the model and biological networks suggest new functions for previously identified connections in the C. elegans nervous system. Keywords: chemotaxis, Caenorhabditis elegans, spatial orientation, recurrent neural networks, sensorimotor integration 1. Introduction The complete description of the morphology and synaptic connectivity of all 302 neurons in the nema- tode Caenorhabditis elegans (White et al., 1986) raised ∗ To whom correspondence should be addressed. the prospect of the first comprehensive understanding of the neuronal basis of an animal’s entire behavioral repertoire. The advent of new electrophysiological and functional imaging techniques for C. elegans neurons (Lockery and Goodman, 1998; Kerr et al., 2000) has made this project more realistic. Further progress would be accelerated, however, by understanding