Lately I've been fascinated about learning about AI. I've been reading AIMA (aima.cs.berkeley.edu), the entire book from front. I do this for fun, it's very enjoyable. Sometimes friends ask me how I've been and what I've been up to, and they're often surprised to hear that I enjoy reading an textbook. Well this isn't the first textbook, and those that I do read, I love to get a deeper understanding. Some textbooks I loved reading were CLRS, Operating System Concepts (Silberschatz, et al.), uC++ book on concurrency and parallel programming (Buhr, http://plg.uwaterloo.ca/usystem/pub/uSystem/uC++book.pdf). Currently it's AIMA, and also http://neuralnetworksanddeeplearning.com/chap1.html. They're fascinating stuff.
Oh while on the bed and letting mind wander, I had a thought. Self taught learning is better than following instructional learning. Because self taught is like an informed search in AI whereby you have an idea of what you want to learn about, so in a sense you have a heuristic function.
And although unproven, brain memory is limited. And so to cope with that, we "forget" things over time. This means we need to review things periodically to freshen up, and it's a way of the body deallocating and allocating memory.