With a little bit of coaching, people are able to fool fMRI scans by hiding memories and creating false new memories. A Stanford neuroscientist has been working on a computer program which can use brain scan data to test whether or not a person is remembering something-- a narrow version of a lie detector! However, with coaching and concentration, people are able to fool the brain scans.

For example, the brain scans can distinguish between a novel experience-- such as when someone first sees a face-- and a memory -- remembering the face. But if subjects are told to try to conceal their memory state by pretending they are seeing the face for the first time when they are actually remembering-- or vice versa-- they successfully fool the brain scans. In other words, the brain regions that would normally be associated with a novel experience are activated with memory. It is not surprising that people can learn to fool fMRI-- in fact, with training, subjects can consciously get better at up-regulating or down-regulating brain regions by watching real time fMRI scans (see studies here and here). In other words, they watch a picture of their brain with regions lighting up and can purposefully change which areas are lighting up (with some instruction from researchers).

In the study, the very cool part is that people can specifically conceal whether or not a memory is real. 400 faces were shown to subjects who had seen 200 of them the previous day, and the participants were coached to try to dupe the test. (Strategies included focusing on features of a familiar face that they had not noticed before and making up false memories of a novel face). Brain activity reflected these conscious changes! This study illustrates the difficulty of implementing brain scans as evidence in a court of law-- as it stands, fMRIs are far from an objective lie detector. Subjects who are stressed, thinking about something that happened long ago, trying too hard to be honest, or purposely trying to be deceitful all can create false results.