MRI can decode subjective, but not objective, memories

In experiments that cast further doubt on claims of lie detection by fMRI measurements, Rissman et al. find that subjective memory states can be decoded accurately under controlled experimental conditions, but that fMRI has uncertain utility for objectively detecting an individual's past experiences. Here is a nice summary of their work from Gilbert Chin:

As a consequence of recent investigations that have used sophisticated methods of analyzing brain activity to propose that objective lie detection may be feasible, it has become apparent that designing a task in which subjects lie whole-heartedly and voluntarily (as opposed to being instructed to do so every fifth answer, for instance) is a nontrivial undertaking. Rissman et al. have approached this challenge by adapting a well-established laboratory paradigm—that of face recognition—to conditions that approximate those of quotidian experience. They asked subjects to study 200 faces and then interrogated them 1 hour later, using a mix of new and old test faces. The menu of responses offered a choice of (i) definitely remembered; (ii–iii) high and low confidence that the face was familiar; and (iv–v) high and low confidence that the face was new.

An analysis of brain activity during the response phase revealed distinctive patterns when old (that is, previously studied) faces were rated by the subject as definitely remembered versus strongly familiar, and also when they were rated as being strongly versus weakly familiar. In contrast, for faces rated as being weakly unfamiliar, it was not possible to tell from the neural activity patterns which were actually new and which had been seen during the study phase, and for weakly familiar faces, the new/old distinction was achievable only some of the time. Furthermore, if subjects were instead told to rate attractiveness during the study phase and then asked to categorize faces by gender during the response phase, it was not possible to diagnose which faces were new and which were not. Taken together, these findings suggest that brain activity reflects subjective, rather than objective, face recognition.