Mushroom-bodies regulate habit formation in Drosophila

Abstract

AbstractOur past experience is one of the primary sources of information when faced with a choice. We ask ourselves: "what will happen if I do this?"Accurately predicting the consequences of our actions is usually modeled by operant (instrumental) learning experiments. These types of experiments are often contrasted with classical (Pavlovian) conditioning experiments in a dichotomy. And indeed, different brain circuits mediate the acquisition of skills and habits (via operant/instrumental learning) and the acquisition of facts (via classical/Pavlovian learning). However, realistic learning situations always comprise interactions of skill- and fact-learning components (composite learning). Fixed flying Drosophila melanogaster at the torque meter provide one of the very few systems where the relationship of operant and classical predictors in composite learning can be studied with sufficient rigor. The latest experiments show that the textbook operant/classical dichotomy is misleading and that instead composite learning consists of multiple interacting memory systems. These interactions between predictive stimuli (classical component) and goal-directed actions (operant component) make composite conditioning more effective than the operant and classical components alone (learning-by-doing, generation effect). Rutabaga (rut) mutants are impaired in learning about the (classical) stimuli, but show improved (operant) behavior learning. This is the first evidence that operant and classical conditioning differ not only at the circuit, but also at the molecular level. The interaction between operant and classical components is reciprocal and hierarchical, such that the classical suppresses the operant component. Experiments with transgenic flies demonstrate that this suppression of operant learning is mediated by the mushroom-bodies and serves to ensure that the classical memories can be generalized for access by other behaviors. Extended training can overcome this suppression and transforms goal-directed actions into habitual responses. This interaction leads to efficient learning, enables generalization and prevents premature habit-formation.