ACTION, VALENCE, DOPAMINE- <EM >DROSOPHILA</EM> AS A STUDY CASE

Since Darwin’s early proposal that emotions and behaviour may mutually influence one another, the causal direction from action to affect has remained difficult to dissect experimentally. Here, we establish a neurobiologically tractable model in Drosophila melanogaster demonstrating that a specific motor program can itself serve as an aversive teaching signal. We show that optogenetic activation of “moonwalker” neurons, which elicit backward locomotion, is sufficient to induce negative valence: odours paired with backward movement are subsequently avoided. Thus, a defined motor command, in the absence of classical noxious stimuli, can instruct associative learning. Using an integrated approach combining behavioural assays, circuit-specific optogenetics, pharmacology, connectomics, neurophysiology and computational modelling, we dissect how this motor-driven punishment signal is implemented. We identify pathways linking the mushroom body to premotor circuits and demonstrate a critical interaction with dopaminergic reinforcement neurons, revealing how movement-related signals are integrated into canonical learning circuits. Beyond defining a novel form of reinforcement, our results suggest that aversive value can emerge from the internal monitoring of action itself, rather than exclusively from external sensory punishment. A normative model constrained by our circuit findings predicts how such motor-based reinforcement stabilizes learned avoidance even in the absence of overt punishment, providing a mechanistic resolution to the classical “avoidance paradox” in experimental psychology. Together, these findings uncover a causal link from action to affective learning and establish a generalizable framework for understanding how motor states shape emotional valence at the circuit level.