Humans make decisions based on arbitrary and capricious preferences, and are good at coming up with logical-sounding reasons to justify those decisions (eg: rejecting a candidate because she is a black woman, but saying it was because she didn't communicate well). A neural net makes a decision based on a perfectly reproducible sequence of calculations. There has been a lot of work published about how we can explain the output of neural nets, and IMO this has much better prospects than trying to explain a human’s decisions by scanning the brain or whatnot.