Spatial memory shapes active sensory sampling in pulse-type electric fish

Abstract

Event Abstract Back to Event Spatial memory shapes active sensory sampling in pulse-type electric fish James J. Jun1, 2, 3*, Erik Harvey-Girard2, Andre Longtin1, 3 and Leonard Maler2, 3 1 University of Ottawa, Department of Physics, Canada 2 University of Ottawa, Department of Cellular and Molecular Medicine, Canada 3 University of Ottawa, Center for Neural Dynamics, Canada An animal's past experiences influence its behavior and its focus of its attention as it investigates its environment. Active sensing behavior therefore implies that an animal could use its past experiences to choose an optimal search strategy and modulate its sensory sampling rate accordingly. Weakly electric fish are nocturnal and navigate using an active electric sense by generating an electric organ discharge (EOD) and detecting distortions in its electric field caused by environmental features. Some electric fish (pulse-type) species generate brief EOD pulses (~1 ms) at a varying rate; the EOD rate rapidly rises when fish encounters a novel sensory stimuli. This phenomenon is known as a novelty response and the fish’s sensory sampling rate, in the dark, is simply equal to its EOD rate. We studied the South American pulse-type weakly electric fish Gymnotus sp. to test whether they can be trained, without visual cues, to learn a spatial location of food using electrosensory-detectable landmarks and, if so, whether they can use their past experiences to minimize their search area and modulate their EOD rate when hunting for food. A live mealworm was tied and presented at a fixed location with stable plastic landmarks until the fish was trained to find the food by traveling the shortest path; we determined the EOD rate as a function of landmark location and analyzed how it changed over learning trials. We presented four peripheral landmarks within a circular arena (1.5 m in diameter, Fig. 1) for all groups, and a central landmark adjacent to the mealworm for a cue-learning group; the other group was assumed to rely path integration for locating the prey. During probe trials (after learning was complete), we took out only the food and determined the fish’s trajectory and recorded its EOD rate as a function of location. During perturbation trials, we rearranged the landmarks by translation, expansion, or mirroring, and observed the most frequently visited location and associated EOD rate. Finally, fish were relocated and re-learned the food location by entering the same arena from a different entrance which was rotated 90 degrees from the original entrance. We found that fish revisited the missing food location with high spatial accuracy. The most visited location was within ~1 cm of the missing food location for the cue-learning group, and within ~5 cm for the path-integration group. Fish specifically increased their EOD rate near the missing food or changed landmark locations. The location of search was influenced by the geometric arrangement of landmarks. Translation of the landmarks caused the most visited location to be displaced by an equal amount for the cue-learning group, but the displacement of the most visited location followed less closely for the path-integration group. After the relocation, fish re-learned the food location from an alternate entrance and achieved the shortest path more quickly than in the initial learning phase. Our studies show that Gymnotus sp. can choose its search trajectory and modulate its sensory sampling rate based on its past experiences of landmarks and path-integration. Figure 1 Acknowledgements The authors acknowledge Lee Hunton, Declan Lu for their help with experimental setup and Chris Anderson for data collection. Authors also thank Dr. Gary Marsat and Dr. Erik Harvey-Girard for helpful discussions.