Project Summary / Abstract This grant focuses on how very recent experiences––over the past few seconds to minutes––allow brains to update expectations about the world and then use these expectations to guide behavior. The ability to flexibly adjust one's course of action in this manner is a hallmark of adaptive human behavior. At the neural level, relevant cellular-activity correlates have been described in non-human primates and other vertebrate model systems. For example, ramping neural activity has been observed in the few hundred milliseconds, or seconds, leading up to behavioral decisions and the rate of rise of these ramps tracks the gradual accumulation of information relevant for the decision being made. Ramping activity is thus a correlate of an increasing expectation that an important decision needs to be made and the moment at which the ramp reaches a threshold level typically signifies when a final decision is taken. Another salient correlate of internal expectations are reward-prediction error signals: bursts of dopamine-neuron activity when an animal receives an unexpected reward or a reward is surprisingly omitted. Reward-prediction error signals seem well poised to adjust animal and human behavior based on learned expectations. A clearer picture of how quantitative internal signals––like ramping and reward-prediction error activity––contribute to behavioral flexibility would be an important step forward for cognitive neuroscience. Here, we propose to develop two new behavioral tasks in tethered Drosophila, where we can perform simultaneous neurophysiology. Our first aim is to use one of these tasks to test the hypothesis that ramping neural signals are fundamental in forming behavioral decisions over tens-of-seconds to minutes timescales in ethologically relevant contexts, rather than just on much shorter timescales and in laboratory defined tasks (as has been shown to date). Such a discovery would argue that expectations built over minutes in real-world conditions are ultimately fed into slowly ramping neuronal signals so as to guide natural decision-making. Our second aim is to discover reward-prediction error signals in fruit flies actively performing a trial-by-trial conditioning task and to define a circuit mechanism through which such signals allow brains to form quantitatively precise expectations––updated on a trial-by-trial basis––on the likelihood of rewards arriving or not arriving in the near future. Such discoveries in a genetically tractable model will inform our thinking on how our brains generate expectations that allow for flexible, adaptive behaviors, ultimately informing new therapeutic approaches to neurological conditions in which flexibility is impaired, such as obsessive-compulsive disorder.