Ordinary Nothing

Another case of wide publication for a juicy bit of news that  is based on rock-solid science. That is, assuming your definition of rock-solid is rather fluid.

The first news reports made this seem worse than it really is, I admit. It seemed so absurd that I just ignored it the first time I saw it. And the second time. And the… You get the drift. But it keep on coming, and most people seem to take this seriously for some reason.

Six people were asked to shoot toy guns, five people did not shoot the toy guns.
Then they were all placed under fMRI, and under polygraph tests, and told to alternately lie and tell the truth about the shooting.

And lo and behold, there were noticeable differences in brain activity.
Which led the researchers to assume lots of things about how the brain handles truths and lies differently. They want more funds to develop sophisticated lie-detectors based on this.

The original news reports (reporters must not be up to par these days) claimed that the shooters were told to lie, and the non-shooters to tell the truth, which of course turns the whole thing into a complete mess. The original news reports also claimed that there were six shooters and three non shooters, which strikes me as somewhat a lot harder for a reporter to miss.

One problem which is still there is the huge sample size. Eleven people. How do you get statistically significant results from that is beyond me.
Maybe they just need a baseline for these specific eleven people? This could work, get eleven suspected terrorists to do the research with you, ascertain that you know when they’re lying, and then interrogate them about their own terrorist activities… Somehow I have the feeling it’s not quite what they wanted here… Besides, if they were terrorists, they may have lied on the control questions as well.

Another problem is that… well… In this research both the fMRI and the polygraph showed distinct results.
But polygraphs pretty much suck at what they do. So now we know fMRI is as good as polygraph testing? Wow!

Besides, the polygraph testing itself was not even done with the best and newest polygraph techniques. Wrong type of control questions. The measured the wrong thing. They don’t have legitimate comparisons.
And when everybody knows it’s plain research and specifically told not to try and lie, and not to get excited, it can be expected that their brain activity (both for telling truth, and for lying) will be different from that of someone who is being accused of a serious crime.

Show me a larger sample, with cases where the polygraph fails miserably under best practices, and where the fMRI is accurate. Then there will be something worth considering.

December 1st, 2004 at 12:38 EET | Categories: Science | Track comments. You can skip to the end and leave a response. Pinging is currently not allowed.