Well, you could analyze how the human brain performs tasks that computers can't seem to do very nicely (either because our algorithms to do those tasks aren't very good, or because of some sort of inherent limitation). Among these tasks are image recognition, natural language processing, speech recognition, etc. Imagine the possibilities: you could get up in the morning, step inside of your autonomous vehicle, ask it to "drive me to work, but stop at a good coffee shop first", and have it drive you to your destination in the middle of a hurricane. Or alternatively, you could datamine the shit out of everything that ever existed.

The last part is probably just wishful thinking. I ended up writing a research paper on this topic last year, and those issues I listed are actually the ones I saw most frequently listed as the possible applications of advances in neural and cognitive sciences.