I’ve always believed that we’re too overawed by our own technological prowess, but this poster presented last June at the Human Brain Mapping Conference should really serve as a cautionary tale for those of us who glance at fMRI research and figure that the findings are wholly trustworthy.

A graduate student at Dartmouth was busily engaged in two things that graduate students do best: research, and general goofiness.  The research, in the area of decision-making, involved scanning people’s brains while they observed photographs and attempted to guess the emotional valence of the subjects depicted.  The goofiness involved using various objects bought at the supermarket as scanning targets while they were developing their protocols.  A pumpkin went in first, and then a cornish game hen.  Finally, the researcher purchased a whole Atlantic Salmon and strapped him into the fMRI machine.  Ever the scientist, he ran the protocols properly, showing the deceased fishie the photographs and asking it to guess at the emotions being displayed.

A certain amount of false positives are expected, but it was where the false positives were that alarmed the grad student – right in the salmon’s dead little brain cavity. 

Rather than worrying, however, that deceased ectotherms have the ability to successfully read human emotions, we should take this as a warning to have a healthy skepticism regarding the interpretation of fMRI data, especially since multiple comparisons, though recommended, are hardly ever done.  I’m always hearing about how fMRI research has pinpointed the part of our brain that may be responsible for (divorce, personality traits, wristwatch preference)… brain imaging research is amazing and we’ve advanced by leaps and bounds – but in a lot of ways, we’re still swimming upstream.