Nisbett, R. E. and T. D. Wilson (1977). "Telling more than we can know: Verbal reports on mental processes." Psychological Review 84: 231-259.
(JL note: This is one swing in a continuing back and forth debate going back to at least the earliest experimental psychology. Topics to look for include introspection, the unconscious mind, and imageless thought.)
They note cognitive psychologists ca 1977 seem to believe the following three points but really don't support these conclusions. "In the absence of evidence ... social scientists are not likely to abandon practice of quizzing subjects ... 232"
e.g evaluation, judgment, problem solving, and the initiation of behavior.
Ss may not report stimuli, their responses, nor any inferential mediating process. For example in the Maier "functional fixedness" research, most subjects do not report that a swinging string led them to use an object as a pendulum even though without the slight movement, solutions are reduced and take much longer.
"they may base their reports on implicit a priori theories about the causal connection between stimulus and response"
to the incidentally correct employment of a priori causal theory.."
(JL) This does NOT mean we are never accurate; indeed the process of science offers methods to help us. Also consider the evidence against the idea that we can "learn" without awareness.
they conclude little evidence for connection between verbal report change and
Subjects sometimes do not report the evaluational and motivation states produced in these experiments , and even if they do, they may not report any change that occurred in themselves.237
various by Leon Festinger (1957) "when prophecy fails!"
Freud, Piaget, G have discrepancy as a motivator for cofabulation.
no evidence of overt verbal report involved even though there are changed evaluation of stimuli (e.g. snakes, shocks) and altered motivational states. 234.)
subjects don't "experience" the changes and even deny it occurred p.236
All including dissonance studies show failure of subjects to report on inferential processes supposedly occurring.
controversial after all these years- yet some minimal effects are clear.
(JL) Two conclusions seem warranted at this point, 11/00. Brief affective stimuli can sneak beneath awareness, biasing the affect of stimuli in full awareness. (Zajonc in several studies). There is no evidence that more than one word's meaning can sneak through subliminally -- the "two word" rule (Kihlstrom in Velmans, Greenwald, 1998?)
studies on Maier's tasks show little awareness that clues help and much that subjects don't know which do. E.g. inadvertent bumping of "pendulum" cord is not reported yet is facilitative. (These protocols could have been Gazzaniga's split brain patients!) 241.
While these studies do not show people could never be accurate, they do suggest within a few minutes or more introspection fails.246
The above evidence is consistent with the most pessimistic view about peoples' ability to report accurately about their cognitive processes.
Many subjects give similar "explanations" as well as do observers of those subjects asked to predict their behavior. 247
When asked about stimulus effects, people don't consult memory but a priori plausibility based on their "causal theories."248
notes there's no implication these "causal-theories" must be wrong; just that
they are used instead of introspection to respond... a pool of culturally
(See Linde, C. (1993). Life stories: The creation of coherence. New York: Oxford University Press.)
Linde asks people about their lives and in the process uncovers the sources of common beliefs and "common sense" -- beliefs and principles everyone in a community holds. She includes professional theories that escape into the culture such as Freudian ideas like "repression", religious doctrines, self-help manuals, media influences, etc.
(Think of Joe in the Gazzaniga video. His confabulation is obvious as we have privileged information just as the experimenters in the attribution studies do.)
Sometimes a subject's verbal account may seem totally bogus yet when used, it may have a degree of correlation with the "true" account held by the experimenter.