Gallagher,
S. 2002. Experimenting with introspection (Comment). Trends in Cognitive
Sciences,
6 (9): 374-375.<p>
Experimenting with introspection
Comment from Gallagher
I would like to endorse both the possibilities of using introspection in experimental science, as proposed by Jack and Roepstorff [1] and others, and the various cautions about its use, as indicated by Schooler [2] and others. To elaborate further on their arguments, let me suggest that introspection itself, even as it is used in experimental science, is not a unified concept.
Schooler lists various terms used to signify introspection, or what he calls meta-consciousness. In most cases, what is indicated is something that is Œsecond-order¹ relative to first-order phenomenal experience. This kind of second-order reflexive activity can be very simple, and as such can be found in a large variety of experiments designed to minimize dependency on introspection. In such cases, experimenters might ask their subjects for quick reports about what they experience. ŒDo you experience (see, hear, feel, etc.) X or not¹? In some cases, to avoid the effects of verbal misinterpretation the subject is asked to simply push a button once she experiences X. This still depends on a quick and minimal introspection of the first-order experience (seeing, hearing, feeling, etc.) to be reported. Marcel has demonstrated good reason to be cautious even about this kind of minimal procedure [3]. Specifically, across different report modes (button push, eye blink, verbal ŒYes¹) it is possible for different reports to be generated for the very same trial.
More prolonged forms of introspective self-observation may be called for in some experiments. For instance, if asked to report on an emotion, a subject is required to make considered judgments about her own first-order phenomenal experience. Rather than asking whether or not the subject experiences X, an experimenter might ask the subject what exactly she does experience. In such cases the subject is asked for a description, which is, in fact, an interpretation. The important question here concerns that on which the interpretation is based. If no instruction is given, a naïve subject is likely to give their report in folk psychological terms. To avoid the subjective aspects of such interpretations, the scientist often instructs the subject, or provides a pre-established set of categories from which the subject chooses the relevant interpretation. A significant bias, however, can enter into the experimental procedure just here in this attempt to be objective. The scientist might feel that some degree of objectivity is met because there is a set of scientific categories in play. But what is it that makes them scientific rather than folk psychological? What is the source of these categories? In cases where such categories are meant to be descriptions or interpretations of first-order experience, then either the experimenting scientists must draw the categories from their own phenomenology (simply reinstating subjectivity), or from a more anonymous phenomenology devised by other scientists in other experiments. But under threat of infinite regress, the phenomenological buck has to stop somewhere, and someone¹s introspective self-observation has to answer to the possibilities of temporal and translation dissociations raised by Schooler and many of the traditional critics of introspection.
There are two good responses to such issues one internal and the other external to the experiment itself. The first takes a more systematic approach to introspection by using procedures that allow the subjective experience of experimental subjects to inform the experimental analysis. This is what happens in the study by Lutz et al. [4], cited by Jack and Roepstorff. By instructing subjects to set aside standard (folk psychological) conceptions and theories, instructing them to focus on the first-order experience itself, and asking Œopen¹ questions, Lutz and his colleagues employed a version of systematic phenomenology that allowed subjects themselves to define the proper analytic categories. That is, the categories were generated in the very first-order experience that the experimenter wants to know about, rather than in some other, often anonymous, first-order experience, the relevance of which is a matter of interpretation generated outside of the experiment itself.
This kind of procedure will not work in every type of experiment, nor will it necessarily deliver the best results even in those where it is possible. As Marcel notes [5], in some cases (for example, in attempts to measure the effect of unconscious processing, as in priming) tasks that use introspective procedures, no matter how well they are defined procedurally, might not reveal the degree of effect as clearly as procedures that measure the effect indirectly, without introspection. For such reasons, the second good response to the problems of introspection is the one recommended by Jack and Roepstorff: triangulation. That is, interpret introspective results in an adjudicative mix with results generated from other kinds of experiments and observations. In the experiment by Lutz et al., for example, the interpretation of introspective findings was reinforced by consistent EEG recordings.
These considerations suggest three rules for keeping introspective techniques scientific. Although the first two might seem on first glance to be in opposition to each other, they actually raise an important question about what counts as replication in experiments involving introspection.
Rule 1: keep the introspective procedure explicit and systematic. This will assist in possible replication by other researchers.
Rule 2: minimize dependency on extra-experimental, pre-established categories. On the face of it, this seems to prohibit replication. That is, if one cannot use the categories developed in the experiment one is trying to replicate, how is replication possible? It is important to realize that a prohibition against the use of pre-established categories is, in the case of studies of first-order experience, a prohibition against a certain kind of bias, but it is not a prohibition against using the same categories used in the previous experiment. The very same (or very close) categories might in fact be generated within each experiment by following the same introspective procedure. And if very different categories are generated from one experiment to the next, then something is still in need of explanation. This very possibility motivates the final rule.
Rule 3: triangulate.
Philosophy and Cognitive Sciences, Canisius College, Buffalo, NY 14208, USA.
e-mail: gallaghr@canisius.edu
References
1 Jack, A.I. and Roepstorff, A. (2002) Introspection and cognitive brain mapping: from stimulusrepsonse to scriptreport. Trends Cogn. Sci. 6, 333339
2 Schooler, J. (2002) Re-representing consciousness: dissociations between experience and meta-consciousness. Trends Cogn. Sci. 6, 339344
3 Marcel, A.J. (1993) Slippage in the unity of consciousness. In Experimental and Theoretical Studies of Consciousness (Ciba Foundation Symposium 174) (Bock, G.R. and Marsh, J., eds), pp. 168180, John Wiley & Sons
4 Lutz, A. et al. (2002) Guiding the study of brain dynamics by using first-person data: synchrony patterns correlate with ongoing conscious states during a simple visual task. Proc. Natl. Acad. Sci. U. S. A. 99, 15861591
5 Marcel, A. J. (1998). Blindsight and shape perception: deficit of visual consciousness or of visual function? Brain 121, 15651588