Neural Basis for a Social Internal Model in Electric Fish

The ability to predict what others will do next is fundamental to social life—from avoiding collisions to understanding intentions. Despite its importance, the brain mechanisms underlying such social predictions remain poorly understood, in part because the relevant circuits in the mammalian brain are extraordinarily complex. This project uses a simpler model system—the weakly electric mormyrid fish—to investigate how the nervous system learns to predict the sensory consequences of the actions of others and to identify the specific cells and circuits responsible. Weakly electric fish generate brief electrical pulses to communicate and to sense their surroundings, much like a biological sonar. A great deal is already known about how the fish's brain predicts and cancels out the electrical interference caused by its own pulses, allowing it to detect prey and other external stimuli. Far less is understood about how the brain handles interference from the pulses of nearby fish during natural social interactions. This project experimentally tests the hypothesis that a brain region similar to the human cerebellum generates learned predictions of signals produced by other individuals, and constructs a detailed computational model of how this is achieved at the level of individual neurons and their connections. Because the social signals in this system are simple, precisely measurable, and experimentally controllable, the research offers a rare opportunity to provide concrete answers to a fundamental question in social and cognitive neuroscience—how the brain models the actions of others. This project advances NSF’s priorities in Artificial Intelligence. Outreach activities include developing teacher curriculum instructions for electric fish recordings and behavior in K-8 schools and at the Zuckerman Education lab. Generating predictive models of the environment is a core function of the nervous system. Research in this domain has commonly focused on the problem of predicting the sensory consequences of an animal’s own behavior. Equally important, but far less well understood, is how the nervous system predicts the sensory consequences of the actions of others. This project seeks to elucidate mechanisms underlying such social internal model predictions by leveraging a tractable model system in which detailed accounts of sensory prediction are already available in the context of individual behavior. Weakly electric mormyrid fish communicate and actively sense their environments using electrical pulses, known as electric organ discharges (EODs). However, EODs are a source of interference for a passive electrosensory system that is used by the fish to detect prey. Extensive past work has revealed how copies of the motor command to discharge the electric organ are transformed into specific predictions of neuronal responses to the fish’s own EOD within a cerebellum-like circuit known as the electrosensory lobe (ELL). By canceling responses to the fish’s own EOD, such predictions enhance detection of external electrosensory stimuli. This proposal considers natural social settings in which the ELL faces additional interference due to the EODs of conspecifics. Specifically, the proposed research (1) tests the hypothesis, strongly supported by preliminary data, that the ELL generates learned predictions of the sensory consequences of the actions of others and (2) construct and test a detailed and comprehensive computational model of the cellular and circuit mechanisms. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria. NSF Award ID: 2544926 | Program: 01002627DB NSF RESEARCH & RELATED ACTIVIT | Principal Investigator: Nathaniel Sawtell | Institution: Columbia University, NEW YORK, NY | Award Amount: $1,197,230 View on NSF Award Search: https://www.nsf.gov/awardsearch/show-award/?AWD_ID=2544926 View on Research.gov: https://www.research.gov/awardapi-service/v1/awards/2544926.html