On the Apparent Paradox of Learning and Realism

This article first summarizes how the definition of perception as the detection of information follows from the assumption of realism (e.g., Shaw, Turvey, & Mace, 1982). The realist position appears to be inconsistent with the empirical finding that novice perceivers often use nonspecifying variables and converge on the use of information only after practice with feedback (e.g., Michaels & de Vries, 1998). We argue that the appearance of inconsistency is due to the application and evaluation of realist principles beyond the scale of phenomena to which they apply. If the relevant principles are considered at the appropriate scales, convergence on information and realism imply each other. We also argue that the possibility of convergence and the associated use of nonspecifying variables should always be considered in the analyses of experimental results, especially if the information-granting constraints prevailing in the experiment are different from those prevailing in natural ecologies.