October 10, 2001

10:02 AM

Debate over brain scansOver at the Register, one of their writers has gotten into a fantastic pissing contest with InfoSeek's founder over the issue of brain scans and airport security. What are your thoughts?
posted by xochi (7 comments total)

What are your thoughts?

if you had a brain scanner, you'd know.
posted by hotdoughnutsnow at 10:31 AM on October 10, 2001

My thoughts: Mr Greene is a rude, arrogant SOB. But I think those traits are virtually requirements for employment at the Register.
posted by mw at 11:04 AM on October 10, 2001

Generally, brain research studies involve averaging over a large number of participants in order to describe a pattern of activation. Given the massive individual differences that tend to occur in brain activation patterns, this proposal seems, to me, to be fraught with statistical concerns.

The FBI study involved comparing the brain response to probes (words/phrases related to a certain stituation) to the brain response during word monitoring (subjects had to press a button when they saw certain words [target] or to ignore non target words). The responses to probes tend to statistically match the responses to target words rather than the responses to non target words when you know the probe words and vice versa when you don't. Note that they only record from three sites (Fz, Cz, and Pz, if you care), so it's not clear where the activation is coming from (EEG has lousy spatial resolution).

In addition, there is the constant problem of extra electrical activity from other cognitive processes and muscle movements that adds noise and decreases the chances of finding any sort of wave form. However, averaging could take care of this.

Finally, the P300 reference in the original article (that precipitated the email war) refers to a positive peak in the EEG at about 300 ms following the stimulus onset, and is [usually] derived by averaging over a large number of trials. What does the P300 really indicate? P300 occurs when the brain responds to a significant but low-probability event. The target words in the FBI study were both signifiant (in that they were monitored for) and infrequent (low probability). The nontarget words were insignificant and frequent. The probe words for knowledgable subjects were significant (they new what the acronyms or phrases meant) and infrequent. In addition, the known stimuli are oddball in the sense that they do not fit in with the other stimuli (targets are familiar to everyone, nontargets are unfamiliar to everyone, and probes are very important in another arena - imagine a reaction of "What's that doing here?"). Is it any wonder that they get their effect?

Further P300 reading

So there could be something to this (since knowledge of the phrases is driving the effect to some degree, despite the frequency confound), but 90-days to being ready? I certainly hope not. Brain waves are tricky things to deal with, and we don't know exactly what the underpinning of the waves involved in this 'fingerprinting' is.

(Huh. That was a lot shorter when I started.)
posted by iceberg273 at 11:27 AM on October 10, 2001

Law enforcement use of such a device, even if it was 100% accurate, would constitue a clear violation of the fi fth amendments protections against self incrimination.
posted by NortonDC at 1:05 PM on October 10, 2001

Steve Kirsch is a charlatan. A dangerous, wealthy charlatan, at that.
posted by holgate at 1:44 PM on October 10, 2001

mw: My thoughts: Mr Greene is a rude, arrogant SOB. But I think those traits are virtually requirements for employment at the Register.

Well, certainly a sniping sneering attitude prevails. That's part of what I love about them.

The people who let themselves get dragged into shouting matches with them do so at their peril. And often, for my entertainment.
posted by cps at 4:22 PM on October 10, 2001

I'd love to know how Kirsch proposes to select the images to be used in the test process. According to his site, the stimuli should be images that "only a terrorist should recognize", such as a page from an al Quaeda code book or a map of a terrorist training camp. If we have specialized, classified information of that nature, the government certainly isn't going to make it available for the viewing pleasure of airline passengers the world over. In fact, if we have enough detailed inside information to provide varied and accurate tests (since the images have to provoke "a strong response"), why wouldn't we just use it to deal with the terrorists in a more direct fashion, like cruise missiles?

Of course, what can you expect from a man who thinks planes should be equipped with "auto-land" buttons (see middle of page page) to be used in case of hijack?
posted by atlee at 4:42 PM on October 10, 2001