Text this: What kind of empirical evidence is needed for probabilistic mental representations? An example from visual perception.