I think the definition will change as we learn more, but my working definition of solving the brain is: one, we can model, maybe in a computer, the processes that generate things like thoughts and feelings, and two, we can understand how to cure brain disorders, like Alzheimer's and epilepsy. Those are my two driving goals. One is more human-condition oriented, and one more clinical.

It's actually kind of weird that we can comprehend the law of gravity, or that we can understand quantum mechanics, enough at least to make computers.

A big part of my job is to remember failure, and reboot failure, when the timing is right.

Behavioral economics can explain some things, but it's hard to explain a lot of the underlying processes that generate these decisions, much less some of these unconscious things that we don't have a handle on at all.

Synthesize new ideas constantly. Never read passively. Annotate, model, think, and synthesize while you read, even when you're reading what you conceive to be introductory stuff. That way, you will always aim towards understanding things at a resolution fine enough for you to be creative.

I spend a lot of time going over old conversation summaries. A lot of the old ones are about ideas that ended in failure, the project didn't work. But hey, you know what? That was five years ago, and now computers are faster, or some new information has come along, the world is different. So we're able to reboot the project.

The brain is really hard to see. The whole thing is very large - the human brain is several pounds in weight - but the connections between brain cells, known as synapses, are really tiny. They're nanoscale in dimension. So if you want to see how the cells of the brain are connected in networks, you have to see those connections, those synapses.