Figure 5.
Model comparison between model families (colors), for each model factor (panels). A. Percentage of subjects for whom all models belonging to a certain model family (FP, SA, EP, or VP) are rejected, that is, have an Akaike information criterion (AIC) higher than that of the winning model plus the rejection criterion on the x-axis. For example, when we use a rejection criterion of 10, all models of the FP model family are rejected in about 90% of subjects, all SA and EP models in about 50% of subjects, and all VP models in none of the subjects. B. The same comparison executed for number of remembered items. C. The same comparison executed for number spatial binding errors. See Table 2 for the model abbreviation key.