Subthreshold oscillations and neuronal input–output relationships

The difference between resonator and integrator neurons relies upon a frequency preference for the former, while the latter monotonically increase their spiking probability with the frequency of a periodic input. In model neurons where dynamics on multiple time scales are present (hence ubiquitously in real neurons) the concept of frequency preference can naturally be broaden to take into account preference towards temporally specific n-uples of presynaptic inputs. On the other hand, single neurons with dynamics on multiple time scales can detect specific temporal patterns as we discuss in this paper.