One of the much documented problems of Bayesian approaches to cognitive modeling is that there isn't much of a neural grounding. The NEF can be used to compute probabilistic computations with ease (see Chapter 9 of Neural Engineering), however have this been used in a published model before? Are there drawbacks that make merging the NEF with Bayesian modelling methods unfeasible in terms of resources required?
Asked
Active
Viewed 64 times
1
-
The thesis, Neural Plausibility of Bayesian Inference, has been available for a few years. Haven't managed to write an answer about it yet :( – Seanny123 Jan 11 '21 at 14:48