That is one of the questions that L.F. Abbott brings up in his recent Neuron opinion paper. He tackles the current dogma that it is only through prediction and testing that one can conduct science,
To apply prediction as the ultimate test of a theory is a distortion of history. Many of the most celebrated moments in quantitative science–the gravitational basis of the shape of planetary orbits, the quantum basis of the spectrum of the hydrogen atom, and the relativistic origin of the precession of the orbit of Mercury–involved postdictions of known and well-characterized phenomena… The key test of the value of a theory is not necessarily whether it predicts something new, but whether it makes postdictions that generalize to other systems and provide valuable new ways of thinking.
This is an interesting outlook, and you cannot criticize him for beating a straw man because there seems to be a strong impulse toward prediction, for better or worse. He is right to emphasize the role that theory has played in the history of science, although I would note that a push towards empiricism may be behind the leaps and bounds that science advanced in the 20th century.
He also discusses some of the technical issues facing theoretical neuroscientists. Computationally, the neuron has so many inputs that the information-carrying spikes would theoretically be likely to be drowned out by the large average number of spikes. There are three current proposals to deal with this problem: correlate the inputs, reduce the number of presynaptic afferents, or integrate over the inhibitory and excitatory inputs to reduce the value of the non-informative spikes.
Later in the paper he discusses some of the hurdles theoretical neuroscientists must overcome, and why considering the synapse to be the “slow” learner and the neuron to be the “fast” computational element may be radically wrong. It’s a good, short paper that brings up a number of interesting points without laboring over any one for too long.
Abbott LF. 2008 Theoretical neuroscience rising. Neuron 60:489-495. doi:10.1016/j.neuron.2008.10.019