Conclusion
- We really do not understand how neurons work together
- Neurophysiology helps but is currently lacking measurement tools
- Computational modelling can help
- Neurally plausible models can help very much
- Picking the right model is difficult
- Precise neural models (e.g. Genesis) are too computationally
expensive
- Neurally inspired models (e.g. backpropagation) only help to
show distributed processing
- Size of the model may also be important
- Hopefully, CANT is a good compromise
- CANT can be used to show that spontaneous neural activation
is computaionally useful
