Secondly, the computational approach specifies the processparameters of RL, such as PE, learning rate, reward sensitivity,and memory of previous reinforcement (which will be discussedin detail below). For instance, in the field of depression, thisreduces the broadly defined symptom of anhedonia to more spe-cific and refined constructs (Treadway and Zald, 2011; Der-Avakianand Markou, 2012. See our later discussion of ‘wanting’, ‘liking’and learning in Section 3.4). This moves the field from generalsymptom-based descriptions to direct and precise identification ofimpaired latent functions (Wiecki et al., 2015).