Commentary on: “Functional Neuroimaging—Can It Contribute to Our Understanding of Processes of Change?”

If psychodynamic concepts are to be tested using functional neuroimaging (fMRI), they must be quantified. This is not a metaphor; rather, it is a literal requirement. To enter a psychological hypothesis into fMRI analysis software is to use the number keys. The general linear model can't read prose.

With a patient, in my office, I may ask “Are you experiencing my silence as though I were against you?” But of the same patient lying in an fMRI scanner, I must ask “010203?” In my office, the patient may respond “Not at all! When you did not speak I worried that you do not care for me, that I bore you.” But in the scanner she must reply “121350.” In my office I may consider her answer and think “My interpretation seems off base; perhaps my countertransference is leading me astray.” But analyzing my fMRI findings, I must think “The correlation coefficient between my a priori model and the BOLD signal is .126 and insignificant; I must reanalyze the data or choose a new task.”

Are my two questions the same? Are her two answers? In short, can analytic ideas be quantified—can words become numbers? Beutel and Huber ask an important question: what is the “potential of functional neuroimaging [to aid] in the understanding and evaluation of psychotherapy?” Their answer, though a qualified “yes,” gives short shrift to three fundamental methodological questions that need attention. They are:

1. Can subjective experience be converted to numbers?

2. Can the thinking brain be converted to numbers?

3. Can subjectivity-numbers and brain-numbers be correlated?

As a function of technological requirements, fMRI must answer “yes” to these questions if it is to proceed. But this does not mean that psychoanalysis must as well. Because they do not fully address the problems of quantification and correlation, I fear that Beutel & Huber have underestimated the likelihood that we are witnessing the emergence of a neuropsychoimagin