How serious is this challenge to the validity of fMRI research?

July 20, 2016 9:02 AM

A recent study claims to have found problems with fMRI software that cast a huge number of past studies into doubt. Is this over-inflated hype (debunking for sake of debunking) or is this a genuine challenge that actually invalidates past research? the report / article in the Economist.
posted by aunt_winnifred to Science & Nature (4 answers total)

Best answer: as someone with a 'cognitive psychology phd dropout' familiarity with both fMRI and statistics...

my initial concern is the number of times they divided and re-ran the experiment drastically increases the potential for false positives- which is exactly what they're complaining about. 500 subjects is a fairly small sample when you're running three million comparisons.

this article has a discussion about a much simpler p-test that speaks to my question about the article.
posted by noloveforned at 9:22 AM on July 20, 2016

Is the bulk of fMRI data questionable? (Retraction Watch)
posted by caek at 9:52 AM on July 20, 2016 [1 favorite]

A lot of fMRI studies are complete voodoo. Generally, fMRI looks at BOLD - blood oxygen level dependent contrast imaging - spots with increased blood flow is used as a correlate of neural activity, which is kinda sketchy but it's a tool that's available. Researchers have always acknowledged this and take fMRI studies with a grain of salt.

Dead Salmon fMRI
posted by porpoise at 11:09 AM on July 20, 2016 [3 favorites]

finally able to find a few followup articles that will undoubtedly be better informed on the subject than i will-

Donders Institute reacts to concerns about fMRI validity

Read the official response on the Eklund et al. paper from the Organization of Human Brain Mapping (OHBM) in the Huffington Post.

Read more technical comments on the Eklund et al. paper from the Organization of Human Brain Mapping (OHBM).
posted by noloveforned at 5:41 AM on July 26, 2016 [1 favorite]