Turing Completeness
- We suspected before, but we have shown during the project
that the primitives are Turing Complete.
- Turing completeness means that anything that can be programmed
can be programmed with the system.
- We have a specific result that fLIF neurons are Turing complete.
- Given enough neurons, anything that can be programmed can be
programmed with fLIF neurons.
- More importantly, we've shown that CAs, no matter how they're
implemented, are Turing complete.
- This is an important theoretical result, but it also shows that
the problem is underconstrained.
- For instance, we can implement a psychological model that fits
data by programming the function (post-hoc).
- However, it's kind of tricky to program with neurons, and
we've developed some good skills with it.
