Goal-directed navigation requires animals to continuously evaluate their current direction and speed of travel relative to landmarks to discern whether they are approaching or deviating from their goal. Striatal dopamine release signals the reward-predictive value of cues1,2, probably contributing to motivation3,4, but it is unclear how dopamine incorporates an animal’s ongoing trajectory for effective behavioural guidance. Here we demonstrate that cue-evoked striatal dopamine release in mice encodes bidirectional trajectory errors reflecting the relationship between the speed and direction of ongoing movement relative to optimal goal trajectories. Trajectory error signals could be computed from locomotion or visual flow, and were independent from simultaneous dopamine increases reflecting learned cue value. Joint trajectory error and cue-value encoding were reproduced by the reward prediction error term in a standard reinforcement learning algorithm with mixed sensorimotor inputs. However, these two signals had distinct state space requirements, suggesting that they could arise from a common reinforcement learning algorithm with distinct neural inputs. Striatum-wide multifibre array measurements resolved overlapping, yet temporally and anatomically separable, representations of trajectory error and cue value, indicating how functionally distinct dopamine signals for motivation and guidance are multiplexed across striatal regions to facilitate goal-directed behaviour.
