[mc32]

Why don't agencies use fMRI instead of something as unreliable and questionable and discredited as polygraphs?

I'm not sure you'd use fmri as widely as polygraphs are today, but maybe for targeted investigation they'd be a useful tool?

There are claims of up to 90% accuracy[1] while not perfect, are better than polys.

[1]http://www.scientificamerican.com/article/new-lie-detector/

[Balgair]

Because they are no more or less useful. Here's a link to one of the best headlines Wired ever ran:

http://www.wired.com/2009/09/fmrisalmon/

Essentially, even a stinking dead fish will show positives in an fMRI. And yes, these studies have really rocked the fMRI field and called into doubt many experiments.

[dewarrn1]

fMRI as a method is not bankrupt or ill-founded. However, as the study you linked shows, it is susceptible to statistical artifacts that must be controlled to support rigorous conclusions.

[JackFr]

fMRI is not bankrupt or ill founded, but it shows what it shows, that is more or less the geography of brain oxygen consumption.

The ill-founded and bankrupt part comes from all of weak and tenuous interpretation that comes after that. fMRI can teach us much about brain physiology, but when blood flow is linked to psychology we should exhibit heightened skepticism.

[dewarrn1]

Don't believe the hype. Applying fMRI methods to lie detection is at least as problematic as the polygraph. Here's a good review:

http://www.tandfonline.com/doi/abs/10.1080/15265160701842007