Neural networks (biological ones) engage in learning too!
Byrne proposes that imitation in great apes is actually composed of two stages of processing. The first is the absorption by an observer, an infant chimpanzee, for example, of simple goal-directed acts. This process would be enabled by way of mirror neurons. The second is a process of learning more complex behaviours by observing hundreds of instances of such complex behaviours. Each complex behaviour is composed of sequences of the simple acts in the observer’s brain from the first process. This learning process, like machine learning, would detect statistical regularities in the hundreds of observed complex behaviours.
Link to paper: http://rstb.royalsocietypublishing.org/content/358/1431/529
You may also like to browse other academic papers: http://www.thesentientrobot.com/category/tsr-academic-papers/