Figure 6. Self-made vs. given category assignment.
(a) Experiment 3: Instead of performing the discrimination judgment themselves, subjects were provided with a cue indicating the correct category assignment right before the stimulus was presented. Then, after stimulus presentation, subjects first performed an unrelated color discrimination task in place of the orientation discrimination task (they needed to remember the randomly assigned color (red/green) of the cue indicating the correct category) before indicating their perceived stimulus orientation. (b) According to our model we should see similar estimation biases in Exps. 2 and 3, which is indeed what we found. (c) Again, the fit model well accounts for the overall distribution of orientation estimates (combined subject; see Figure 6—figure supplement 1 for distributions for individual subjects)). Because the discrimination judgment was given and always correct independent of the noise in the sensory measurement , estimates only occurred in the ‘correct’ quadrants. For the same reason the model also predicts slightly smaller bias magnitudes (compared to Exp.2), which is also matched by the data (see also Figure 7b).
Figure 6—figure supplement 1. Full distributions of individual subjects’ estimates in Experiment 3.
