The computer program metaphor

Ever since ENIAC in 1945 at Penn, or before that the writings of von Neumann and others, the computer has been an alluring metaphor for human cognition. I believe it leads us into false assumptions.

The level above sequence control may appear to resemble programs with if/then contingencies when sequences available for a purpose are familiar and well practiced (and especially when they are written down as recipes or instructions), but even then innovation is the norm should the familiar next sequence be thwarted by disturbances.

That norm characterizes what I prefer to call the planning level. What known or imaginable sequences terminate in a given desired outcome? Try them out in imagination, and when one seems to work in imagination, put it, or some part of it, to the test of environmental feedback. In B:CP, Bill’s example of searching for his glasses is not execution of a previously established program, it is a review of sequences which he remembered recently controlling, or which he imagines he might recently have controlled, each of which terminates in setting his glasses down someplace, and in each case going back and resuming at the place in which that sequence terminated. It is too ad hoc and improvisatory to be called a program on the analogy of a computer program.

Rather than being a level at which programs exist and are run, it is more fundamentally that level at which programs are created by cobbling together the final outputs and initiating inputs of sequences at the level below. It is still at the behest of principles at the next level up, but Principle control does not always select from a library of tested and stored programs. In the manner of Captain Picard, it holds a desired outcome as its reference and says to the crew below that command deck, “Make it so.” Yes, of course, every level does that. But the crew cobbling sequences together has a lot more freedom to be creative. if there’s a well-trodden path all well and good; but new commands may come down, and even for familiar ones disturbances happen, so that a volatile search for new ways to connect the input of one sequence to the output of a prior one is pretty ordinary.

If/then contingencies, in particular, are not implemented by some neurological structure specialized to perform Boolean logic. An input perception is available or it is not; a perceptual input function receives a given perceptual input (with adequate amplitude) or it does not; correspondingly, reference input functions and so on. Sequence control is a chaining together of control loops such that the control of one is the condition for controlling the next, but we don’t claim if/then contingency.