Skip to content


  1. davidlduffy davidlduffy

    Schmidhuber (LSTM etc); Friston etc

    Where this might have a difficulty is when one discusses mathematical understanding. This comes up with computer generated proofs, and in mathematical education. It is objective, but a property of a learner (one can explain a proof, one can generate a new theorem, one can apply a particular method to a new area or recognise a dual). This obviously overlaps a lot of modern physics, and partly addresses de Regt’s problem about simplified “false” models that are actually deeper mathematically, though it does ring with his description of understanding as a skill. But no equivalent to causality, ISTM, especially if one is a constructivist in these matters.

    In the example of population genetics, I think it is the “higher” level where selection etc is happening – the molecular genetics is too “low”. It is the coarsening of the description that allows one to see what is really going on (sim thermodynamics, QM) in a complex system.

  2. Jeb Jeb

    I don’t know Hempel, no idea who he is or the context. I don’t think i want to know any more about him.

    It conjures the image of a man sitting on a spike or a Dalek.

Comments are closed.