Challenges for Psychiatric Neuroimaging: The Case of Conversion Disorder

We are grateful for the two commentaries on our article, for contrasting reasons. Sprevak points to a further way in which the neuroimaging window on the brain may be distorted, whereas Burklund and Lieberman paint us a compelling picture of what we might hope to see on the other side. Our sympathies, however, remain with those, like Sprevak, who would point to flaws in the glass, for all that we wish to see what lies beyond.

Sprevak highlights a limitation of many cognitive psychological models and the tasks used to capture them that arise with what is known as extended cognition. He suggests that this is likely to be a problem for modeling many psychiatric disorders, and we are inclined to agree. At the very least, the need to model cognition beyond what happens in the head means that the scanner environment and task will need to be included in the model; but whether even this limited extension would be sufficient for most psychiatric disorders is doubtful. Although he does not offer a specific psychiatric example, we shall: hysteria, or conversion disorder, as it is now known—a disorder in which life stresses seem to be somehow 'converted' into pseudoneurological deficits. Central to all contemporary models of hysteria is that the patient cannot or will not tell us about what is going on; their relevant experiences are repressed, dissociated or dissembled (Kozlowska 2005). It is thus ripe for exploration with fMRI, which might hope to tell us what the subject cannot tell us themselves, and the research community has not been slow to take up the challenge (Aybek et al. 2008). There are two great difficulties in modeling hysteria in any kind of scanner-friendly way, however. First, the provoking problem is thought to be deeply idiosyncratic and perhaps deeply obscured from the investigator; second, the provoked response is not thought to be static, or fixed, but adaptive—to both the provoking situation and the patient's own illness model. Modeling this process thus requires the investigator to adapt the task to each subject, with obvious implications for group averaging and generalizability. To model the provoking problem, the investigator will have to find some way of recreating the key features of the situation—yet this is a clear example of the extended cognition challenge: what may have been problematic about the situation may not have been that it involved, for example, the appearance of a mouse, but that the patient's thoughts, recollections, guilts, desires, and behavioral responses incorporated the mouse in that distressing way at that moment—a mélange that can never be recreated. Modeling the provoked response is equally challenging. The limb paralysis that arose that day is not there at all times (not, for example, when the patient is asleep, or when they are distracted by something else), and there may not be a detectable difficulty until the limb is put to use (Maruff and Velakoulis 2000). The paralysis depends on environmental and personal cues that are realistically, again, cognitively extended and idiosyncratic.

But it is too tempting, after a hundred years of conceptual stagnation, to ignore the potential that fMRI offers in this case. A limited, even hamstrung, exploration will be better than the experimental void that largely preceded it. One of the authors of this paper has attempted a study of a single subject with conversion disorder, attempting to recreate the experience of the original stressor and provoke the ensuing symptoms through the recall of objectively rated, temporally relevant events (Kanaan et al. 2007). This is clearly a grossly inadequate study, and bears all the hallmark problems we described—miniscule sample size (a single subject, as many fMRI studies in hysteria are), a task that can only gesture toward the illness, and an interpretation (support for a Freudian model of hysteria) that clearly goes well beyond what the current state of knowledge allows us to claim. And it is inadequate for all the reasons we mentioned—that the subject group is difficult to recruit, that substantial...