Engineering News

So, the machine alone generates 810 false positives per hour, while the "cyborg" approach reduces this to 5 false positives per hour on the basis of the highly-touted P300 response. They don't tell how many true positives are identified per hour by the machine, so the actual error rate per total positives identified is not given.

They also forgot to include a human operator without the EEG as a control, or the article didn't bother to report those results. How many false positives are generated using ordinary methods, I wonder?

Even supposing that a single human operator without EEG cannot monitor ten images per second, and therefore several operators might be required for the monitoring job, effective monitoring which generates no false positives, or at minimum, very low false positive rates, would constitute a saving if resources are deployed to deal with any percentage of false positives.

I'm also curious about the fatigue factor for personnel reviewing ten images per second by EEG. P-300 is a marker for recognition. The success rate for individuals would be expected to vary, depending on the individual's capacity to be attentive and recognize true positives, and capacity would be expected to decline as the individual becomes fatigued. I wonder what the average fatigue rate is for the task of reviewing 600 images per minute, cw other tasks. Or perhaps "subconscious" processing is not fatiguing at all? Perhaps the operator should be encouraged to have a beer before work?

One good lookin' chick walking by would screw up the whole system....lol

On the other hand this would be an excellent sleep aid....