“… as of this writing, neuroscience knows of no plausible mechanism for carrying a large amount of information forward over indefinitely long intervals in a form that makes it accessible to the mechanisms that effect the primitive two-argument functions that are at the heart of computation.

Until the day comes when neuroscientists are able to specify the neurobiological machinery that performs this key function, all talk about how brains compute is premature. It is premature in the same way that talk of how genes govern epigenesis was premature prior to an understanding of what it was about the structure of a gene that enabled it to carry inherited information forward in time, making it both copyable and available to direct the creation of organic structures. Symbolic memory is as central to computation as DNA is to life.”

— C. Randy Gallistel and Adam Philip King, Memory and the Computational Brain