Figure 5.

Exemplary effects of between-group variations of latent variables. Panel (a) shows the exemplary trial sequence on the cWCST as depicted in Figure 1. Panel (b) shows cognitive-learning feedback prediction values (referred to in the text as activation levels of objects of thought) for the application of the shape category across seven trials, the first three of them are shown in (a). The received positive feedback for the application of the shape category on Trial 1 causes an increase in feedback prediction values. With high configurations of the cognitive retention rate (i.e., ${\gamma }_C$), such as seen in ALS patients, feedback prediction values retain higher levels of activation from trial to trial. Note that the shape category was not applied on Trials 4 to 7. Panel (c) shows response probabilities on Trial 3. The probability of executing response 3 is highest (i.e., application of the shape category), followed by the probability of executing response 1 (i.e., application of the color category) and the probability of executing response 2 (i.e., application of the number category). With high configurations of the inverse temperature parameter (i.e., $\tau$), such as seen in ALS patients, differences between response probabilities are attenuated, biasing response probabilities towards a uniform probability of 0.33. Note that the probability of executing an odd response (i.e., responses that match no viable sorting category; e.g., response 4 on Trial 3) is close to zero on any trial. Because the response probabilities of odd responses are virtually zero, such probabilities remain mostly unaffected by the inverse temperature parameter. The presented effects of model parameters were computed by varying exclusively the parameter of interest at arbitrary values while holding all other model parameters constant.