We investigate neuronal circuit computations underlying learning across different time scales: from long-term learning in naive animals to fine tuning of decisions in experts. In my talk I will present some of our recent work on behavioral, dopaminergic and computational principles underlying these processes. Our results demonstrate that mice learning trajectories transition through sequences of strategies, showing substantial individual diversity. Remarkably, the transitions are systematic; each mouse’s early strategy determines its strategy weeks later. Striatal dopamine signals match individual trajectories, encoding teaching signals for sculpting these trajectories. Finally, we show that a deep neural network using heterogenous teaching signals captures these results. Altogether, this work provides insights into the biological and mathematical principles underlying individual long-term learning trajectories.