What you can do is see which neurons were activated by a certain event, then figure out what other situations they are activated by and the patterns that they m

What you can do is see which neurons were activated by a certain event, then figure out what other situations they are activated by and the patterns that they match. That's won't get you a coherent explanation straight away, especially for a very large network, but it's a start.