Abstract

Decision-making strategies in the face of conflicting or uncertain sensory input have been successfully described in many species. We analyze large behavioral datasets of larval zebrafish engaged in a "coherent dot" optomotor assay and find that animal performance is bimodal. Performance can be separated into two "states"-an engaged (attentive) state with high performance, where fish consistently turn in the direction of coherent motion, and a second, disengaged (inattentive) state, where performance drops to chance. A hidden Markov model is sufficient to model these transitions and can be incorporated into a drift-diffusion model framework that has previously been used to model perceptual decision-making in larval zebrafish. Furthermore, we fit a mixture model of performance distributions and extract two latent variables termed "focus" and "competence" that are largely influenced by parents and environmental context, respectively. This quantitative framework can potentially help to identify a genetic basis and a neural mechanism for attention that extends across organisms.