There are two broad categories of Neural Nets. The first, which is well known, are based on back-propagation or clevver variants thereof. The real brain does not use back-propagation for learning, not even close, and this has always bothered me.