Perturbational complexity analysis predicts the presence of consciousness in humans by stimulating the brain with a brief pulse, recording cortical responses with electroencephalography (EEG), and measuring their spatiotemporal complexity. Measures reflecting perturbational complexity are low when consciousness vanishes in deep sleep, general anesthesia, or after widespread brain damage, and are high during wake, dreaming sleep and after recovery from brain injury. Here we examine the neural circuits underlying perturbational complexity in a mouse model by directly stimulating the cortex and recording high-density EEG, as well as spiking activity throughout the cortico-thalamo-cortical system with up to three Neuropixels probes, while subjects are awake or anesthetized via isoflurane. We find that when the mouse is awake, stimulation of deep cortical layers reliably evokes locally a brief pulse of excitation, followed by a bi-phasic sequence of a 120 ms profound off period and a rebound excitation. A similar pattern of spiking activity, some of which can be attributed to burst spiking, is seen in the connected thalamic nuclei. These spiking patterns are associated with a pronounced late component in the evoked EEG. We infer that cortico-thalamo-cortical interactions drive the long-lasting evoked EEG signals elicited by deep cortical stimulation during the awake state. The cortical and thalamic off period and rebound excitation, as well as the late component in the EEG, are reduced in running mice and are absent in animals anesthetized with isoflurane. This is also mirrored in the perturbational complexity indices computed during quiet wakefulness, running, and anesthetized states. Taken together, these results suggest that cortico-thalamo-cortical interactions contribute to ongoing activity in the conscious state that leads to highly complex global responses.