>an analogy to machine learning neural nets, so the analogy is not biological. Except that "neural nets" are only "neural" by a (biological) analogy to neurons

>an analogy to machine learning neural nets, so the analogy is not biological. Except that "neural nets" are only "neural" by a (biological) analogy to neurons. So he is describing a biological phenomemon by comparison to a a type of machine learning which is itself described by the very same biological phenomenon he's trying to describe.