These researchers have invented a scheme for finding interesting images. Computers aren't so great at it yet, but humans are. In fact, we're so good at it that we recognize important images before we consciously know it, and this recognition can be measured by EEG. In their setup, a human watches images go by, and the ones that register on the EEG as "of interest" are set aside to be looked at more carefully. In short, it's brain-aided image triage.
Given the current sources of funding for research, the examples given are surveillance camera shots, and the T-word has to be mentioned. This makes the whole project stink of 21st century panopticon. But that's not the important part.
Using a human as a coprocessor, literally as a brain rather than as a person, is new. I imagine it doesn't matter too much which brain you use, aside from some that are very good or very bad at recognizing images. It's also likely that this isn't fun "work." Just looking at rapidly changing images for a long time is tiring, and if you aren't able to do anything else but sit in the chair and let your unconscious processes do something, the boredom would be awful. From my own experience doing EEG biofeedback, the side effects of directly EEG-linked activity can be very unpleasant and unpredictable. I doubt anyone knows yet what the effect would be of long-term work as a rent-a-brain.
A Philip K. Dick dystopia looms, in which "braining" is something the poor do, like plasma donation or prostitution. Maybe it fucks you up pretty bad, but the Wal-Mart hasn't been hiring in a while and you need cash. Too bad about the week-long psychoses a person gets after doing the hookups for a couple of weeks of 12-hour days...