Heisenberg, Kandinsky, and the Heteromethod Convergence Problem: Lessons From Within and Beyond Psychology

Abstract

During the past 100 years, advances in personality assessment have paralleled key events in art and physics; but for the most part, these parallels have gone unrecognized. In this article, I discuss the ways in which 2 movements in 20th-century art (cubism and nonrepresentational painting) and 2 principles from 20th-century physics (the uncertainty principle and the observer effect) combined to create an intellectual context for the process dissociation approach to personality assessment, a research strategy wherein naturally occurring influences on test scores are deliberately manipulated to illuminate underlying response processes. I discuss core elements of a process-focused paradigm for 21st-century personality assessment including (a) the need for researchers to explore test score divergences as well as convergences, (b) a view of the assessor as active shaper (rather than dispassionate observer) of testee behavior, and (c) the importance of integrating personality assessment concepts and methods with ideas and findings from disciplines within and outside psychology.

Acknowledgments

Editor's Note: This is an invited article based on a Master Lecture given March 27, 2008 at the Society for Personality Assessment Annual Conference in New Orleans, LA.

Notes

¹ Images of well-known works of art are now easily accessible via the Internet; access information for all paintings cited in this article is summarized in the Appendix. If any of these links become inaccessible, these works can be accessed through the following sequence: (a) go to the Google Web page at www.google.com; (b) click on the Images link in upper left corner; and (c) enter the artist's last name, followed by the name of the painting, in the text box below Google Image Search (e.g., Picasso Man With a Guitar).

² Needless to say, not all testees genuinely attempt to engage and respond accurately to self-report test items (see Gallen & Berry, 1996; Graham, 2000), and those that show evidence of content nonresponsiveness (e.g., random or markedly acquiescent responding) are most likely not engaging in this introspective process (Bornstein, 2007b).

³ Other studies in this area have documented the differential impact of gender role on responses to self-report and free response dependency tests such that gender role moderates scores on self-report dependency scales (with high scores associated with high femininity and low masculinity in both women and men) but is unrelated to free response dependency scores (Bornstein, Bowers, & Bonner, 1996b). Insofar as gender role socialization can be considered a preexisting “manipulation” that differs in different participants, these results represent additional evidence in support of the process dissociation approach (see Bornstein, 2002).

⁴ There is, however, one potential exception to this general rule: When neuroimaging technology reaches the point that the mechanics of obtaining accurate functional magnetic resonance imagery (fMRI) data are less intrusive, it will be possible to simultaneously collect fMRI data while administering one or more traditional personality assessment instruments (e.g., self-report, free response). At present, the physical constraints and distractions inherent in fMRI assessment make simultaneous administration of traditional personality tests difficult if not impossible.

⁵ The uncertainly principle and the observer effect are often confused, but there are subtle differences between the two phenomena. The observer effect refers to the unavoidable influence of observer on observed and the ways in which the act of measurement alters the system being measured. The uncertainty principle, in contrast, describes a more fundamental physical constraint on measurement—the impossibility of assessing more than one feature of a system at a given point in time.

⁶ In this context, it is worth distinguishing test score divergence in the process dissociation model from discriminant validity, as the term has traditionally been used by psychometricians. In contrast to discriminant validity, which describes predicted nonrelationships (i.e., near-zero correlations) between a test score and some external criterion, test score divergence in the process dissociation model refers to a decreased correlation between scores on two tests in response to a manipulation specifically designed to decrease these intercorrelations (e.g., a decreased correlation between scores on self-report and free-response measures of stress tolerance in response to a prime or instructional manipulation).