[Tyler]

Questioning the Correlation of fMRI and Behavior
     

We've posted numerous articles on behavioural research being done with
the aid of fMRI scans of the human brain. During recent months, quite
a controversy has been brewing in the scientific community over
statistical correlation methods used in these studies. Last month, a
widely circulated paper, Voodoo
Correlations in Social Neuroscience (PDF format), criticizing fMRI
studies was published in Perspective on Psychological Science. The
paper was written by Edward
Vul, a PhD student at MIT. A recent Nature
News article summarizes the controversy. The Neurocritic blog explores
Vul's criticisms while NewScientist
attempts to explain the controversy. The International Cognition and
Culture Institute also published a summary
of the goings on.  Several of the criticized researchers have written a
joint rebuttal (PDF
format) that agrees with some points of Vul's paper while pointing out
several "partially flawed and misleading" statistical arguments. Not
surprisingly Vul has now posted a web page rebutting the
rebuttal. The only sure result of all this is likely to be more
rigorous studies of fMRI data resulting in better research
on the inner workings of the human brain.
   

http://robots.net/article/2732.html