Identification of an Operant Learning Circuit by Whole Brain Functional Imaging in Larval Zebrafish

When confronted with changing environments, animals can generally adjust their behavior to optimize reward and minimize punishment. The process of modifying one's behavior based on its consequences is referred to as operant or instrumental learning. Operant learning makes specific demands on the animal. The animal must exhibit some flexibility in its behavior, switching from unsuccessful motor responses to potentially successful ones. The animal must represent the consequence of its actions. Finally, the animal must select the correct response based on its past history of reinforcement. Studies in mammalian systems have found competing and complementary circuits in the cortex and striatum that mediate different aspects of this learning process. The larval zebrafish is an ideal system to extend the study of operant learning due to its genetic and optical properties. We have developed a behavioral paradigm and imaging system that have allowed us to comprehensively image neural activity throughout the zebrafish brain during the process of operant conditioning. Our analysis of the neural network activity underlying this learning process reveals several classes of neurons whose activity correlates with learning and decision making. The distribution of these learning-related neurons is highly localized to regions of the habenula and forebrain. We describe, in particular, a lateralized habenula circuit that may encode prediction and relief prediction error.