Results from the COVIS simulations, with COVIS learning the Fig 3 RB categorization task from a variety of parameter configurations.

<p><i>A</i>. Underlying pre-criterion competence (Y-axis) plotted as a function of the pre-criterion performance actually expressed within the simulation (X-axis). <i>B</i>. Underlying post-criterion competence (Y-axis) plotted as a function of the pre-criterion performance actually expressed within the simulation (X-axis). <i>C</i>. The change in underlying performance competence (Y-axis) plotted as a function of the change in performance actually expressed within the simulation (X-axis). In all panels, the diagonal line represents perfect correspondence between the observed performance produced by the simulation and its underlying true competence at the point of producing that observed performance. Each data point represents the average data from one of 23,040 simulated competency transitions. All simulations were performed using Matlab and the code is available as supporting information.