Computationalism

  • computational theory of mind
  • mind is an information processing system
    • cognition and consciousness are forms of computation
    • the system "runs on" neural activity in the brain
  • various thoughts from Josha Bach:
    • Turing test is a test if people they are generally intelligent enough to understand themselves to pass the Turing test
    • children learn that they might do something they do not want to do - to do it they have to model themselves in the world (so they have to run artificial simulations)
    • general intelligence is being able to create generalizations (models) from patterns - some of them are so complex that the thing needs to be able to model itself in the environment (emergent self awareness)
    • mind is creating both the universe and the (story of) self - based on the inputs it's receiving (from retina for example) - you are the machine that thinks it's you
      • it simulates everything that it knows how to, including simulating a person
      • identity is a story that brain tells itself, there's no real me, each person has different simulation of me and I have one (or more) as well - self is an illusion made through Sequences of Static Snapshots
      • we suffer in our minds, and we can change our definition of suffering as well
  • George Lakoff is an interesting connection here - he talks about the symbols in the computer being formally manipulated, but(and?) not having any meaning - syntax is not semantics:

    There is no such thing as a computational person, whose mind is like computer software, able to work on any suitable computer or neural hardware whose mind somehow derives meaning from taking meaningless symbols as input, manipulating them by rule, and giving meaningless symbols as output. Real people have embodied minds whose conceptual systems arise from, are shaped by, and are given meaning through living human bodies. The neural structures of our brains produce conceptual systems and linguistic structures that cannot be adequately accounted for by formal systems that only manipulate symbols.

    — Philosophy in the Flesh -- George Lakoff

  • tensions with computationalism
    • connectionism models at "low-level" (neurons), where computationalism models at "symbolic level"
    • connectionism focused on learning from stimuli (back-propagation in neural networks), where computationalism focuses on learning as "elaborating possibilities through symbolic processing"
  • also, brain writing the story (theory) of itself reminds me of Computationalism - to model complex situations it has to model itself inside of them