Limits on perception of cognitive processes: A reply to Nisbett and Wilson.

Abstract

Nisbett and Wilson's argument that people do not have access to cognitive processes that cause behavior is criticized on theoretical and methodological grounds. It is argued that Nisbett and Wilson's position is stated in a nonfalsifiable fashion, that they define causality in a way that denies subjects access to such information, that Nisbett and Wilson have no definition of the mental processes that subjects cannot report on, and that there is evidence that people do have access to causal processes. Moreover, several of the studies that Nisbett and his associates conducted to support their position used inappropriate statistical tests. A reanalysis of one such study by Nisbett and Bellows demonstrates clear evidence for subjects' access to their mental processes contrary to the original authors' interpretations. It is recommended that researchers focus not on the question of whether people have access to process but on the more productive question of the conditions of such access. Suggestions for such research are offered. Nisbett and Wilson's (1977) recent Psychological Review article has attracted much attention in social psychology. In a field that values counterintuitive findings highly, Nisbett and Wilson's counterintuitive argument that people have no direct access to their mental processes certainly has the potential to startle and provoke thought. Though counterintuitive, this argument is not totally unexpected. An underlying function of the article appears to be to defend current formulations of attribution theory from a perplexing finding: In certain lines of research (the Storms & Nisbett, 1970, insomnia study or phobia reduction studies, e.g., Valins & Ray, 1967), the effects predicted to be mediated by attribution processes have appeared in subjects' behavior, but verbal self-report measures of attribution have entirely failed to show supporting results. If Nisbett and Wilson's arguments are correct, though, these findings are not damaging to attribution theory,