The Temporal Dimension of Consciousness: exploring Intrinsic Neural Timescales (INTs) and their role as a facilitator of consciousness

Recently, it has been proposed that the temporal structure of the brain’s spontaneous activity might be key for consciousness, despite having been overlooked by most current theoretical frameworks. We refer to this inquiry as the investigation of the “temporal dimension of consciousness”. On the neuronal side, this temporal dimension can be approached by probing Intrinsic Neural Timescales (INTs) – defined as the intrinsic temporal durations of neural activity – and their properties, such as their spatial organization across the scalp. In this work, we aimed to deepen the current understanding of the role of INTs in the emergence of consciousness. Our work was carried out especially within the context of the clinical challenges posed by the differential diagnosis of disorders of consciousness (DoCs), which currently suffers from very high rates of error. We here present three studies that we have led as part of our research program. In the first study, we investigated the spatial relation between INTs and the average oscillatory speed in the alpha frequency range (7-13 Hz). Both, in fact, are measures that have been related to the temporal resolution of sensory processing, but their exact relation is far from clear. We hypothesized a clear correlative pattern between the two measures in the fully conscious state; further, we hypothesized that this relation would be disrupted together with loss of consciousness. We showed a significantly negative correlation, at the channel level, between INT lengths probed with the Autocorrelation Window – 0 (ACW – 0) and Alpha Peak Frequencies (APF) in the resting state hd-EEG recordings of our conscious population; additionally, we observed a total disruption of this correlation across different unconscious states – e.g. anesthetic induction (with two different agents) and in individuals with DoCs. These first findings indicate that the relation at different time scales between different measures of temporal processing are key for consciousness. In the second study, we advanced our current methodological arsenal to infer the duration of INTs, by validating a new tool based on Permutation Entropy (PE), which is specifically developed to avoid the confounding effects of the nonstationary and nonlinear nature of the neural signal. We first show, in simulated data, that this measure (which we named Permutation Entropy Time Delay estimation – PE-TD) is indeed less sensitive to nonstationarities and nonlinearities in a neural signal. Second, we observe a high topographic similarity between PE-TD and ACW-0, validating this approach in healthy awake volunteers; surprisingly, this spatial similarity was less evident in DoC data, which might hint at the differential effects of nonstationarity and nonlinearity in these two different states of consciousness. In the third and last study, the objective was to test for the presence of non-random dynamic transitions between different topographies of INTs, supporting claims of a dynamic repertoire of INTs. Leveraging two different datasets (MEG and hd-EEG), we employ a data-driven approach to identify "dynamic INT states." In MEG, we found 10 dynamic INT topographic states, with four maps displaying a significant correlation with myelination maps, indicating a cortical hierarchy. The dynamic transition time series showed intermediate randomness. In EEG, with 7 recurrent topographic states, we show higher randomness in unconscious individuals. Further, when consciousness is lost, dynamic INT transitions display less memory, indicating a less complex and thus a poorer dynamic INT repertoire. These findings extend our understanding of the temporal organization of the resting brain, linking it to consciousness as proposed by the Temporospatial Theory of Consciousness (TTC). Taken together, our results