Fig 11. The additive incentive salience model by Zhang et al. [38] predicts different patterns in dopaminergic teaching signals.

Utility of conditioned stimuli in a study by Cone et al. [26] was computed according to Eq (24). The reward value was set to R = −1 for the animals trained in the balanced state and R = 1 for animals trained in the depleted state. The gating parameter κ = 1 was when the training state is equal to the testing state. An up-shift in state is represented by κ = 3 and a down-shift in state is represented by κ = 1/3.