Then I apologize for my largely off-topic response (though points 3 and 4 still apply to some extent).

However, there are two other things:

1. I would imagine the specific anatomical pattern of the cortex would be enough to classify on, let alone the functional activation pattern produced by specific tasks, especially if significant spatial blurring is being used.

2. There's plenty of literature out there that shows how our pattern of brain activity evolves as we learn new skills. For example, piano players tend to use smaller volumes of motor cortex (that activate more intensely) when performing finger movement tasks. Broca's and Wernicke's area activation changes as one learns a new language. If my brain pattern were to be stored in some database, I doubt it would be very useful to compare against if I were to learn a new skill.

As a fellow researcher in fMRI with a doctorate myself, I have to take exception with how easily you think you can "classify" people with terrorist (or just even simply dishonest) thoughts.

1. Your study most likely stimulated the subject several times, of which an average was taken and then classified. This would not work in the real world, as repeating the same question over and over ("Are you a terrorist? Are you a terrorist? ...") would not work similarly to having a subject push a button repeatedly. For brain imaging technology to work, you have to be able to image individual thoughts, not averages over many similar thoughts. So called time-resolved fMRI tries to do this, but makes several compromises to do so.

2. Motor tasks activating the primary motor cortex are very simple from a cognitive processing standpoint. That's why it so easy to classify motor tasks. Studies of other primary sensory areas, such as the visual and auditory cortices, are also relatively easy areas to classify. In the real world, you have to classify abstract thinking, which is much more diffuse and complex in nature, involving several areas of the brain, at significantly less activation strength. Yes, there have been fMRI studies that demonstrated some success at discerning people who lie from those who tell the truth, or even democrat from republican, but the success was significantly worse than your theoretical 96%, and additionally required a battery of several questions on highly cooperative subjects averaged together.

3. You listed several reasons why fMRI is impractical, but you left out the most pressing one: motion. There are motion correction algorithms, but they sacrifice significant spatial resolution to work, and they often fail for large movements. There is no way to ensure that a subject is cooperative enough to stay still throughout the scan. If I were a terrorist, I would simply keep moving my head during the scan, rendering any result inconclusive.

4. As was hinted at in a previous post, the statistics of the situation make brain imaging an even bleaker prospect. In the general population (say, those who fly on planes), terrorists represent and extremely low prevalence, or prior probability. You would need a test with incredibly high sensitivity to be of any practical use in this situation. For example, if the prevalence was 1:100000, a 96% sensitive test has a positive predictive value of 0.02%--that is, you would have ~5000 falsely positive terrorists before finding a true terrorist.

The cheapest solution for an external ATSC tuner is a used Motorola DSR-550, also known as the cable box issued with the now defunt VOOM service. You can get one on ebay for about $75. I own one, and it works perfectly. The only thing you need to be careful of is that the box you bid on must have already been activated and not "killed" (i.e., not active on the day VOOM when dark). You also need to make sure that the seller includes the remote. There are some universals (like a Harmony) that can control it, but I believe the setup requires the original remote. There's a FAQ all about this cheap solution @ http://www.satelliteguys.us/showthread.php?s=f5cd1 50bdce94a332dd31deb46b954b3&t=7178