Although the preceding discussions have focused on the ability of neuroimaging data to provide information on psychological constructs, a multimethod approach can also prove extremely useful in directing the interpretation of neuroimaging data. Group statistical analyses often proceed on the assumption that all subjects performed a cognitive task in a similar manner or responded similarly to procedures aimed at inducing a specific psychological state.
Unfortunately, verification of this assumption is often difficult. For instance, if we wish to study fear, it is important that we verify that we indeed induced fear and not disgust or other negative emotions. If we lack certainty that the intended state was provoked, then we cannot confidently assume that the brain responses occurred in relationship to the cognitive process or psychological state in question. The solution to this problem is to triangulate on the desired response using multiple methods, including, for example, measurement of task performance, self-report, and psychophysiological recording. As convergent evidence verifies the induction of the intended process or state (and not an unintended state), confidence in interpreting brain responses increases. This triangulation strategy is an example of the multilevel analytic approach described in the preceding chapter by Berntson and Cacioppo, in which information from different levels is used to mutually tune and calibrate data or concepts across different levels of analysis.
Unfortunately, a problem arises in trying to integrate fMRI data with simultaneous collection of other types of data. Specifically, fMRI is both sensitive to artifacts caused by psychophysiological recording devices and causes interference in those same devices. Nevertheless, it is possible to implement psychophysiological recordings such as galvanic skin response, heart rate, blood pressure, and eye tracking within the fMRI environment (Savoy, Ravicz, & Gollub, 1999). These measures are all easily implemented in the PET environment as well. Similarly, measures of hormonal responses such as Cortisol can be collected in the scanner environment. The large differences in time scales of these various measures can cause interpretational issues when moving across levels. Nevertheless, the benefit of collecting such measures should be increasingly apparent.
Was this article helpful?