Lie detection and eye-witness testimony is a messy business. There are a myriad of issues that can arise, from outright lying to false memories created by after-the-fact suggestions and rationalizations. It is the latter effect that makes the proposed use of fMRI in a civil case so troublesome, because even if fMRI were a perfectly accurate technology for identifying lies and truths (which it is not) there would still be the problem of knowing whether the person telling the truth was telling an objective truth or a falsehood he believed to be true. Although the argument could be made that fMRI would function simply as a method for screening outright liars from sincere witnesses (at which point other evidence must be relied upon to determine the veracity of the sincere witnesses' statements), there remain two major problems:
1.) Outright liars can turn into sincere but misled witnesses given enough time and repetition. Since access to MRI scans is not always available in a timely fashion, this would likely be a major issue (for example, the scan in this case comes four years after the incident).
2.) fMRI scans look so damn impressive that any jury (and even most judges) are likely to give them far more weight than they deserve. Considering the degree to which neuroscience peer reviewers can be dazzled by the colourful pictures (and these are people who spend much of their careers dealing with the nuances and limitations of fMRI) it is hard to expect a judge and jury, no matter how intelligent, to give fMRI evidence the appropriate level of ambivalence.
While I am all for the the use of science and technology to improve the justice system, new techniques must be introduced very carefully. The ease with which people can be dazzled by fancy technical words and graphics must be acknowledged along with just what exactly an fMRI 'truth scan' is showing.
Edit: According to alexismadrigal on Twitter, the use of fMRI has been rejected by the court. I found that out about five minutes after posting this... oh the magic of Twitter.