Arousal results in widespread activation of brain areas to increase their response in task and behavior relevant ways. Mediated by the Ascending Reticular Arousal System (ARAS), arousal-dependent inputs interact with neural circuitry to shape their dynamics. In the occipital cortex, such inputs may trigger shifts between dominant oscillations, where α activity is replaced by γ activity, or vice versa. A salient example of this are spectral power alternations observed while eyes are opened and/or closed. These transitions closely follow fluctuations in arousal, suggesting a common origin.To better understand the mechanisms at play, we developed and analyzed a computational model composed of two modules: a thalamocortical feedback circuit coupled with a superficial cortical network. Upon activation by noise-like inputs originating from the ARAS, our model is able to demonstrate that noise-driven nonlinear interactions mediate transitions in dominant peak frequency, resulting in the simultaneous suppression of α limit cycle activity and the emergence of γ oscillations through coherence resonance. Reduction in input provoked the reverse effect-leading to anticorrelated transitions between α and γ power. Taken together, these results shed a new light on how arousal shapes oscillatory brain activity.