Preliminary Comments Designed to Suspend
The Introspection Controversy
An excerpt from Sampling Normal and Schizophrenic Inner Experience
by
Russell T. Hurlburt
published by
Plenum Press
This book describes in detail people's inner experiences; shows how schizophrenia may be
more a disorganization of inner images than of association; and suggests that depression may be
due to lack of words and images in the thinking process rather than to negative self-thoughts.
These observations are based on a new method of eliciting careful self-reports of inner
experience, which some readers may unfortunately characterize as "introspective" and then
dismiss because they accept the conclusion that accurate introspection is impossible. This
introductory section shows such skeptical readers that the present study is different from, and
quite possibly much better than, pervious studies that have been labeled introspective.
See...
The debate over the validity of verbal reports
Nisbett and Wilson's position
Hurlburt's critique
Ericsson and Simon's position
Hurlburt's critique
The most accurate verbal reports are about moments
References
See also Table of Contents
All these texts are from Sampling Normal and Schizophrenic Inner Experience by R. T. Hurlburt. Copyright c 1990 by Plenum Press.
All rights
reserved. No part of this material may be reproduced without permission in
writing from the publisher, Plenum Press, 233 Spring Street, New York, NY
10013.
The debate over the validity of verbal reports
We will take the "Nisbett and Wilson versus Ericsson and Simon debate regarding the validity
of verbal reports" (Grover, 1982, p. 205) as typical of the present polarized positions on
introspection. Nisbett and Wilson (1977) argued against the possibility of accurate introspection,
concluding that "the evidence reviewed is then consistent with the most pessimistic view
concerning people's ability to report accurately about their cognitive processes" (p. 247), and ". .
. there is by now enough evidence discrediting introspective reports to allow us to ignore any
argument based on introspection" (p. 255). On the other hand, Ericsson and Simon (1984)
argued in favor: "It is now time for verbal reports to reassume their position as a rich source of
data, combinable with other data, that can be of the greatest value in providing an integrated and
full account of cognitive processes and structures" (p. 373). We will comment on both positions
in turn.
Next Topic | Beginning
Nisbett and Wilson's position
Nisbett and Wilson's (1977) article cited a wide variety of reports, mostly from the social
psychology literature on cognitive dissonance and attribution theory, where subjects were unable
to describe accurately their own inner processes. We will summarize two of these studies here
to give a typical flavor of the kinds that Nisbett and Wilson reviewed.
In our first example, Bem and McConnell (1970) asked each subject in an experimental group
to write an essay that took a position in opposition to their own opinion on a topic. Subjects in a
control group wrote no such counter-attitudinal essay. Comparisons between the groups showed
that the subjects in the experimental group had changed their attitudes as a result of writing the
counter-attitudinal essay, a result typical of the cognitive dissonance literature. However, Bem
and McConnell found that none of the individual subjects in the experimental group detected any
change in their own attitude. Nisbett and Wilson concluded, "Thus subjects apparently changed
their attitudes in the absence of any subjective experience of change" (p. 236).
In our second example, Storms and Nisbett (1970) gave insomniac subjects in an
experimental group placebo pills to take before bedtime, with a deceptive explanation that the
pills would cause breathing irregularities, rapid heart rate, etc.-in short, the symptoms of
insomnia. The subjects then reported the amount of time necessary to fall asleep on a series of
nights. As expected, the subjects reported getting to sleep faster on the nights they took placebo
pills. At the conclusion of their participation in the study, subjects were asked to explain their
ability to get to sleep faster on the nights they took the pills. None of the subjects thought that
the pill had affected their ability to sleep; they gave instead a variety of plausible but incorrect
explanations. As Nisbett and Wilson summarized, "the explanations that subjects offer for their
behavior in [these] experiments are so removed from the process that investigators presume to
have occurred as to give grounds for considerable doubt that there is direct access to these
processes' (p. 238).
After reviewing these two and many other studies, Nisbett and Wilson concluded: "In
summary, it would appear that people may have little ability to report accurately on their
cognitive processes" (p. 246).
Next Topic | Previous Topic | Beginning
Hurlburt's critique of Nisbett and Wilson's position
Our own evaluation of Nisbett and Wilson's conclusion is that it is too broad: Nisbett and
Wilson reviewed studies in the insufficient-justification, attribution, subliminal perception, and
creativity paradigms and found in those paradigms that subjects were unable to report accurately
their cognitive processes. However, the cognitive processes associated with these paradigms
are but a subset of the total category "cognitive processes"; so Nisbett and Wilson's conclusion is
an overgeneralization from their review. Perhaps a better summary statement would have been
that people have little ability to report accurately on some (or even many) of their cognitive
processes.
Nisbett and Wilson were aware that their review did not allow complete generalization: "We
also wish to acknowledge that the studies do not suffice to show that people could never be
accurate about the processes involved. To do so would require ecologically meaningless but
theoretically interesting procedures such as interrupting a process at the very moment it was
occurring, alerting subjects to pay careful attention to their cognitive processes, coaching them in
introspective procedures and so on" (p. 246, italics in original).
The method that we are using in the present book satisfies all three of these requirements: it
does interrupt processes at the moment they occur; it does alert subjects to pay careful attention
to their cognitive processes, and it does coach them on the procedures. There are thus two
conclusions, to be drawn: First, it would seem that Nisbett and Wilson rather specifically
exempted our present method from their sweeping anti-introspection generalization because it
meets the three requirements stated above; second, Nisbett and Wilson specifically
acknowledged that a study that meets their three requirements (as does ours) could provide
accurate reports about cognitive processes. (As an aside, we have no idea why Nisbett and
Wilson would consider such a method to be "ecologically meaningless.")
Next Topic | Previous Topic | Beginning
Ericsson and Simon's position
Ericsson and Simon (1984) provided an opposing position to that of Nisbett and Wilson, and
concluded that verbal reports of cognitive processes are deserving of a prominent place in
cognitive science. Their argument can be summarized as follows. It is commonly understood
that verbal reports have for much of this century been excluded from scientific psychology; but
this view rests on the mistaken assumption that yes/no responses or response latencies, which
have been acceptable to psychology, are not themselves types of verbal response. Once we
recognize that such responses are in fact verbal responses, then the question becomes not
whether verbal responses should be accepted but rather under what circumstances they should
be accepted. Ericsson and Simon then present an information-processing model of cognition to
evaluate which kinds of verbal reports are likely to be accurate.
Their information-processing model predicts that
...two forms of verbal reports can claim to being the closest reflection of the cognitive
processes. Foremost are concurrent verbal reports-'talk aloud' and 'think aloud' reports-where
the cognitive process ... are verbalized directly...We claim that cognitive processes are not
modified by these verbal reports.... [The] second type is the retrospective report." (p.16, italics in
original)
In the ideal case the retrospective report is given by the subject immediately after the task is
completed while much information is still in STM [short-term memory] and can be directly
reported or used as retrieval cues.... In this particular case, the subject will still retain the
necessary retrieval cues in STM when a general instruction is given "to report everything you can
remember about your thoughts during the last problem." This form of retrospective verbal report
should give us the closest approximation to the actual memory structures. (p.19)
Thus, Ericsson and Simon's model predicts that there are two kinds of "ideal" verbal reports:
concurrent (think-aloud) reports and retrospective reports given immediately. We will evaluate
each in turn by asking whether Ericsson and Simon could possibly have considered our kind of
data when arriving at their ideals.
Next Topic | Previous Topic | Beginning
Hurlburt's critique of Ericsson and Simon's position
Ericsson and Simon's model predicts that there are two kinds of "ideal" verbal reports:
concurrent (think-aloud) reports and retrospective reports given immediately. We will evaluate
each in turn by asking whether Ericsson and Simon could possibly have considered our kind of
data when arriving at their ideals.
First, Ericsson and Simon claimed that think-aloud reports are the "closest reflection" of
cognitive processes, and that these "cognitive processes are not modified by" the think-aloud
procedures. The richness of our own data, as will be seen, makes it impossible for us to accept
such claims. It is premature to summarize all of our results (see Chapters 10 and 15), but a few
observations will make evident why thinking aloud cannot possibly accurately reflect inner
experience as our subjects described it.
First, our subjects may take as long as 30 minutes to describe a one-second experience,
indicating that the nuances of experience do not translate easily into words. Second, our
subjects frequently report multiple simultaneous cognitive processes: visual images
simultaneous with words, etc. Third, verbal thoughts themselves are not experienced simply as
"strings of phonemes," as Ericsson and Simon stated (p.63), but are instead complex processes,
where, for example, the rhythm of a phoneme sequence may be present but the phonemes
themselves absent. Fourth, inner verbal experience sometimes involves visual representations
of the words simultaneous with their phonetic utterance, and the simultaneous visualized works
may be different from the internally spoken words.
We will cite other examples in Chapters 15 and 16, but the conclusion is clear: Thinking
aloud cannot provide an adequate rendering even of inner verbal experience, much less of inner
cognitive experience in general. Thinking aloud is at best only a one-dimensional representation
of a process that is multidimensional. That Ericsson and Simon could make the claim that
thinking aloud represents the "closest reflection" of cognitive processes, and that "cognitive
processes are not modified by [thinking aloud] verbal reports," indicates to us that they must not
have included our kind of data in their analysis.
Ericsson and Simon's second "ideal case [is] the retrospective report given by the subject
immediately after the task is completed while much information is still in STM" (p.19). Now it
happens that our own method of thought sampling asks for "reports on the immediately
preceding cognitive activity," and so seems to qualify as an "ideal case" under Ericsson and
Simon's analysis. But a further analysis shows that our subjects may have better access to their
cognitions than did any of the subjects in the studies reviewed by Ericsson and Simon. Consider
Ericsson and Simon's above-quoted "ideal... retrospective... instruction. 'Report everything you
can remember about your thoughts during the last problem' " (p.19). According to Ericsson and
Simon, this is an "ideal case" because the retrieval cues are present in STM when the general
instruction is given. The thoughts themselves which the subjects are to recall are clearly not in
STM, because the problem during which the thoughts occurred may have required several
minutes to complete.
By contrast, we will see that in our own study subjects were trained to use a clear,
unambiguous beep as a signal to report everything they can about their experience that was
occurring at the moment of that same beep. Thus, the entire experience which our subjects
were to describe, not merely the retrieval cues, has just occurred when the subject begins to
report, and so it would appear that our own subjects' reports may be better, perhaps much better
than the "retrospective ideal" described by Ericsson and Simon.
We conclude our review of Ericsson and Simon's analysis with a summary of our two
conclusions: First, Ericsson and Simon were not including studies such as ours in their analysis;
and second, a study such as ours may provide even more accurate data on cognition than
Ericsson and Simon's "ideal" study.
Next Topic | Previous Topic | Beginning
The most accurate verbal reports are about moments
We formed two conclusions regarding Ericsson and Simon's review: they were not including
studies such as ours in their analysis; and second, a study such as ours may provide even more
accurate data on cognition than their "ideal" study.
We reached essentially the same two conclusions when reviewing Nisbett and Wilson's
analysis as we did when reviewing Ericsson and Simon's. That may seem surprising given that
Nisbett and Wilson and Ericsson and Simon are seen to be on opposing sides of the
introspection debate. The explanation for the similarity of conclusion is that both Nisbett and
Wilson and Ericsson and Simon were interested in the same kinds of verbal reports. Ericsson
and Simon were more supportive of introspection because their information-processing model
identifies some cognitions as being more accessible than others. For example, an introspection
of a cognition that demands a search of long-term memory but provides inadequate retrieval
cues is more difficult than an immediate retrospection (the "ideal" discussed above) where the
retrieval cues are already present in short-term memory. Thus Ericsson and Simon supported
the adequacy of some verbal reports because they identify some introspective tasks as easier
than others.
We have seen that neither Ericsson and Simon nor Nisbett and Wilson directly considered
the detailed descriptions of moments of inner experience such as are provided in the present
book, and we have shown that such descriptions may provide a more accurate view of cognition
than any of the studies called "introspective" by either Nisbett and Wilson or Ericsson and
Simon. The present book explores that possibility of accurate reporting of cognition by
instructing our subjects to "describe what you experience at this particular moment." We do not
ask them to explain why they are having that particular experience. We do not ask them to tell
us what particular environmental events are responsible for that experience. In fact, we do not
even ask them to tell us whether that experience is typical for them or not.
In short, we ask them to report only those experiences for which they have direct (not
inferred) access: We simply ask them to describe what they are aware of at some particular
moment in time, and at that particular moment when the experience is as fresh as it can possibly
be. No introspective study has used a method like this, and we have seen that there is reason to
believe that such a method may produce relatively accurate descriptions of inner experience.
Of course, the fact that Nisbett and Wilson and Ericsson and Simon can be interpreted as
supporting the possible existence of a method of accurate introspection does not make the
present method necessarily accurate. The present study can be viewed as a test of the premise
that subjects can give accurate responses to our simple descriptive request. We ask that the
reader suspend for the moment any biases about introspection and decide for him or herself
whether this study provides accurate, useful descriptions of inner experience, evaluating this
study on its own merits rather than on preexisting beliefs about introspection.
Next Topic | Previous Topic | Beginning
References
Bem, D.J., & McConnell, H.K. (1970). Testing the self-perception explanation of dissonance
phenomena: On the salience of premanipulation attitudes. Journal of Personality and Social
Psychology, 14, 23-31.
Ericsson, K.A. & Simon, H.A. (1984). Protocol analysis: Verbal reports as data. Cambridge,
Massachusetts: The MIT Press.
Grover, S.C. (1982). A re-evaluation of the introspection controversy: Additional considerations.
The Journal of General Psychology, 106, 205-212.
Nisbett, R.E., & Wilson, T.D. (1977). Telling more than we know: Verbal reports on mental
processes. Psychological Review, 84, 231-259.
Storms, M.D,. & Nisbett, R.E. (9170). Insomnia and the attribution process. Journal of
Personality and Social Psychology, 2, 319-328.
Previous Topic | Beginning