Researchers at Princeton University have figured out how to use functional MRI scans to tell what's on your mind. They can't yet tell exactly what you're thinking about, but they can get close enough to distinguish whether you're thinking about (say) an animal as opposed to a vegetable.
Go ahead, think about a carrot. Picturing a carrot? Good. Okay, now think about a cow. Mooooo. Nicely done. Those two different thoughts each generated a distinct pattern in your brain, and with the help of an fMRI machine, Princeton scientists have narrowed down those patterns enough that they'd have been able to tell you that you went from thinking about a vegetable to thinking about an animal. Not what vegetable or what animal, not yet, but the fact that they're able to get even a general level of detail from just pointing a scanner at your head is fairly impressive, to say the least.
When people think of concrete objects, like carrots, their brains tend to make all kinds of random associations very quickly. Try it: think about a carrot again, and pay attention to where your brain goes. For me, I picture carrots, a farm, a tractor, water, rabbits, dirt, sun, the color orange, and Justin Bieber (but he's always in there somewhere). This group of associations has a specific fMRI signature that can't necessarily be used to identify "carrot," but the signature is common to a more generalized category like "vegetables." If you think of something entirely different, like cows, your brain will probably wander off into generalized "animal" associations that the fMRI can pick up and identify as a different, distinct signature.
So once you've got some baseline fMRI scans from lots of people thinking about items in these generalized categories, you can go back and use a new fMRI scan to guess what category an unknown pattern represents. And it works, too: "researchers found that they could confidently determine from an fMRI image the general topic on a participant's mind."
The next step is to fine-tune this method to be sensitive enough to start being able to discern more and more specific categories, like "root vegetables" instead of just "vegetables" or "ungulates" instead of just animals. As for the future, the researchers have this to say:
"If we give way to unbridled speculation, one can imagine years from now being able to 'translate' brain activity into written output for people who are unable to communicate otherwise, which is an exciting thing to consider."
We love unbridled speculation around here, but if you use even less bridlement, there's also the possibility that this sort of tech could be used to read people's freakin' minds! Maybe for good, maybe for evil, but either way, it's no longer science fiction: just science.