Dashboard

Disconnected

Dopamine Model

Signal d'erreur de prédiction de récompense

Schultz, 1997

Dopamine Signal

Reward prediction error (Schultz 1997)

Parameters

τ_DA0.2
α0.1
γ0.99

State

DA Level0.0000

Mathematics

Reward Prediction Error:

δ=R(t)−V(t)\delta = R(t) - V(t)

Dopamine dynamics:

τDAdDAdt=−DA+δ\tau_{DA} \frac{dDA}{dt} = -DA + \delta

Value update (TD learning):

V(t+1)=V(t)+α⋅δV(t+1) = V(t) + \alpha \cdot \delta