There is a question on this site that asks a somewhat related question, whether there are non-physical models for cognition. However, that question still assumes a computational paradigm for the non-physical cognition.
Another question asks whether computational learning theory has been applied to cognitive science, which would then allow us to say what non-computational learning would look like. However, there is only one answer that provides a non-computational alternative.
Are there any accepted models that do away with computationalism altogether? Such a model would depend on cognition performing incomputable feats, such as being a Turing oracle, violating the No Free Lunch Theorem, or reducing NP to P.