Thalamocortical network shows self-sustained oscillations in a broad frequency range especially during slow wave sleep when cortical neurons show synchronized transitions between a quiescent down state and an active up state with beta and gamma oscillations. Inconsistent with previous models, thalamocortical spindles are separated into slow spindles (8_12 Hz) and fast spindles (13_17 Hz) with differential properties. We proposed that cortical high frequency (∼ 25 Hz) activity during up states is the key ingredient for the generation of slow spindles. In fact, the nonlinear interaction between cortical high frequency and thalamic oscillations at fast spindle frequency reproduces oscillations in the range of the difference between the two frequencies that lies into the range of slow spindle. The developed simple deterministic thalamocortical model is able to reproduce up and down states with stochastic high-frequency up-state activity as well as both fast and slow spindles. In agreement with the previous experimental observations, the fast and slow spindles are generated at opposing phases of the up state. To further confirm the causal relationship between slow spindles and cortical high frequency oscillations, we next showed that externally applied high frequency stimulation enhanced the slow spindle activity. Moreover, the prediction of the model was validated experimentally by recording EEG from subjects during nap. Both model and experimental results show increase in high frequency activity before slow spindles. Our findings suggest the important role of cortical high frequency activity in the generation of slow spindles.

Keywords