Yes, the military is studying mind reading but not in the way you might think. They aren't looking into ESP, remote viewing, or psychokinesis, though they have done that before to the tune of almost thirty million taxpayer dollars, I'm embarrased to admit, with Project Stargate and a 2005 study on psychic teleportation. But it is every bit as much a project based on pseudoscience.
According to a August 15th CNN article, the United States military has granted four million dollars to researchers from the University of California-Irvine, Carnegie Mellon University, and the University of Maryland so that they can attempt to learn how to translate EEG signals into actual words. If successful, the ability to do so would be incalculably valuabe to patients suffering from severe forms of paralysis, especially the dreaded locked-in syndrome, among many other conditions which hamper or prevent communication. But if I wasn't reading about it with my own eyes, I would swear that this was a joke, or something thought up by a participant in a fifth grade school science fair.
In the study, researchers place electrodes on the scalps of volunteers and analyze the electrical activity that results when certain words are thought of. In the process, they hope to be able to match specific patterns to each individual word. It sounds simple enough, and even plausible if you aren't aware of the nature and limitations of EEG readings, but it is in effect not all that different than trying to learn how to speak baby by analyzing their cries.
The problem in this is that EEG output, which primarily comes only from the most superficial layers of the cortex, is nowhere near sensitive enough to allow differentiation of specific patterns associated with single words. It isn't even known if there would be a pattern specific to individual words. Perhaps such a crude means of measuring cortical activity would only provide patterns specific to types of words, like nouns or verbs, or to groupings such as animals or plants. So naturally this would become even more complex and difficult to interpret when attempting to read thoughts made up of longer word groupings, sentences, or mental images.
EEGs are are helpful in giving a general picture of the electrical activity occuring in large groups of neurons, such as when attempting to diagnose a seizure or as an adjunct to a brain death exam, but they are often full of artifacts caused by muscle contractions, eye movements and the electircal activity of the heart. Even if patterns of thought were recognizable in a test subject as reproducibly unique EEG waveforms, there would likely be a great deal of difference between subjects, rendering the technology very cumbersome if not useless. We all have different thought processes and associations with words based on individual experiences. EEG patterns of the word beach might vary greatly depending on if one thinks of the world positively or negatively. This is far too complex to work out as simply as the researchers hope.
Another concern is that there is a long track record of such endeavors not being properly, as in skeptically, approached. This could easily turn into another "Emperor's New Clothes" phenomenon where people falsely claim to be trained in interpreting EEG signals. I can easily imagine this replacing Facilitated Communication (FC) as the in vogue means for parents of children with autism to reveal their hidden intelligence. Then the phony accusations of sexual abuse will inevitably arise. It's not a slipperly slope fallacy because it has happened exactly that way before with FC. What's worse is that FC was easily shown to be only in the mind of the facilitator while EEG mind reading would be much more difficult to disprove to a parent that has already become a believer.
Four millions dollars is a lot of money, and a lot of good could come from the proper investment of it. The military needs a better means of evaluating which projects deserve funding and which should be rejected.