Does computational neuroscience need new synaptic learning paradigms?

- Models of synaptic plasticity and learning are inspired by few classical paradigms.
- The models explain sensory development, conditioning and associative memory.
- So far they do not satisfactorily explain one-shot learning and flexible planning.
- Food caching animals show impressive fast learning and flexible planning.
- Behavioural and physiological data from these animals could constrain new models.

Computational neuroscience is dominated by a few paradigmatic models, but it remains an open question whether the existing modelling frameworks are sufficient to explain observed behavioural phenomena in terms of neural implementation. We take learning and synaptic plasticity as an example and point to open questions, such as one-shot learning and acquiring internal representations of the world for flexible planning.