A rich repertoire of oscillatory signals is detected from human brains with electro- and magnetoencephalography (EEG/MEG). However, the principles underwriting coherent oscillations and their link with neural activity remain under debate. Here, we revisit the mechanistic hypothesis that transient brain rhythms are a signature of metastable synchronization, occurring at reduced collective frequencies due to delays between brain areas. We consider a system of damped oscillators in the presence of background noise – approximating the short-lived gamma-frequency oscillations generated within neuronal circuits – coupled according to the diffusion weighted tractography between brain areas. Varying the global coupling strength and conduction speed, we identify a critical regime where spatially and spectrally resolved metastable oscillatory modes (MOMs) emerge at sub-gamma frequencies, approximating the MEG power spectra from 89 healthy individuals at rest. Further, we demonstrate that the frequency, duration, and scale of MOMs – as well as the frequency-specific envelope functional connectivity – can be controlled by global parameters, while the connectome structure remains unchanged. Grounded in the physics of delay-coupled oscillators, these numerical analyses demonstrate how interactions between locally generated fast oscillations in the connectome spacetime structure can lead to the emergence of collective brain rhythms organized in space and time.

Existing whole-brain models are generally tailored to the modelling of a particular data modality (e.g., fMRI or MEG/EEG). We propose that despite the differing aspects of neural activity each modality captures, they originate from shared network dynamics. Building on the universal principles of self-organising delay-coupled nonlinear systems, we aim to link distinct features of brain activity - captured across modalities - to the dynamics unfolding on a macroscopic structural connectome. To jointly predict connectivity, spatiotemporal and transient features of distinct signal modalities, we consider two large-scale models - the Stuart Landau and Wilson and Cowan models - which generate short-lived 40 Hz oscillations with varying levels of realism. To this end, we measure features of functional connectivity and metastable oscillatory modes (MOMs) in fMRI and MEG signals - and compare them against simulated data. We show that both models can represent MEG functional connectivity (FC), functional connectivity dynamics (FCD) and generate MOMs to a comparable degree. This is achieved by adjusting the global coupling and mean conduction time delay and, in the WC model, through the inclusion of balance between excitation and inhibition. For both models, the omission of delays dramatically decreased the performance. For fMRI, the SL model performed worse for FCD and MOMs, highlighting the importance of balanced dynamics for the emergence of spatiotemporal and transient patterns of ultra-slow dynamics. Notably, optimal working points varied across modalities and no model was able to achieve a correlation with empirical FC higher than 0.4 across modalities for the same set of parameters. Nonetheless, both displayed the emergence of FC patterns that extended beyond the constraints of the anatomical structure. Finally, we show that both models can generate MOMs with empirical-like properties such as size (number of brain regions engaging in a mode) and duration (continuous time interval during which a mode appears). Our results demonstrate the emergence of static and dynamic properties of neural activity at different timescales from networks of delay-coupled oscillators at 40 Hz. Given the higher dependence of simulated FC on the underlying structural connectivity, we suggest that mesoscale heterogeneities in neural circuitry may be critical for the emergence of parallel cross-modal functional networks and should be accounted for in future modelling endeavours.

Major Depressive Disorder (MDD) is a complex, heterogeneous condition affecting millions worldwide. Computational neuropsychiatry offers potential breakthroughs through the mechanistic modeling of this disorder. Using the Kolmogorov theory (KT) of consciousness, we developed a foundational model where algorithmic agents interact with the world to maximize an Objective Function evaluating affective valence. Depression, defined in this context by a state of persistently low valence, may arise from various factors—including inaccurate world models (cognitive biases), a dysfunctional Objective Function (anhedonia, anxiety), deficient planning (executive deficits), or unfavorable environments. Integrating algorithmic, dynamical systems, and neurobiological concepts, we map the agent model to brain circuits and functional networks, framing potential etiological routes and linking with depression biotypes. Finally, we explore how brain stimulation, psychotherapy, and plasticity-enhancing compounds such as psychedelics can synergistically repair neural circuits and optimize therapies using personalized computational models.

In the Kolmogorov Theory of Consciousness, algorithmic agents utilize inferred compressive models to track coarse-grained data produced by simplified world models, capturing regularities that structure subjective experience and guide action planning. Here, we study the dynamical aspects of this framework by examining how the requirement of tracking natural data drives the structural and dynamical properties of the agent. We first formalize the notion of a generative model using the language of symmetry from group theory, specifically employing Lie pseudogroups to describe the continuous transformations that characterize invariance in natural data. Then, adopting a generic neural network as a proxy for the agent dynamical system and drawing parallels to Noether’s theorem in physics, we demonstrate that data tracking forces the agent to mirror the symmetry properties of the generative world model. This dual constraint on the agent’s constitutive parameters and dynamical repertoire enforces a hierarchical organization consistent with the manifold hypothesis in the neural network. Our findings bridge perspectives from algorithmic information theory (Kolmogorov complexity, compressive modeling), symmetry (group theory), and dynamics (conservation laws, reduced manifolds), offering insights into the neural correlates of agenthood and structured experience in natural systems, as well as the design of artificial intelligence and computational models of the brain.

Predictive coding frameworks suggest that neural computations rely on hierarchical error minimization, where sensory signals are evaluated against internal model predictions. However, the neural implementation of this inference process remains unclear. We propose that cross-frequency coupling (CFC) furnishes a fundamental mechanism for this form of inference. We first demonstrate that our previously described Laminar Neural Mass Model (LaNMM) supports two key forms of CFC: (i) Signal-Envelope Coupling (SEC), where lowfrequency rhythms modulate the amplitude envelope of higher-frequency oscillations and (ii) Envelope-Envelope Coupling (EEC), where the envelopes of slower oscillations modulate the envelopes of higher-frequency rhythms. Then, we propose that, by encoding information in signals and their envelopes, these processes instantiate a hierarchical “Comparator” mechanism at the columnar level. Specifically, SEC generates fast prediction-error signals by subtracting top-down predictions from bottom-up oscillatory envelopes, while EEC operates at slower timescales to instantiate gating—a critical computational mechanism for precision-weighting and selective information routing. To establish the face validity and clinical implications of this proposal, we model perturbations of these CFC mechanisms to investigate their roles in pathophysiological and altered neuronal function. We illustrate how, in disorders such as Alzheimer’s disease, disruptions in gamma oscillations following dysfunction in fast-spiking inhibitory interneurons impact Comparator function with an aberrant amplification of prediction errors in the early stages and a drastic attenuation in late phases of the disease. In contrast, by increasing excitatory gain, serotonergic psychedelics diminish the modulatory effect of predictions, resulting in a failure to attenuate prediction error signals (c.f., a failure of sensory attenuation). Collectively, these findings implicate cross-frequency coupling across multiple temporal scales as a key computational mechanism supporting predictive coding and suggest that disruptions in these processes play a central role in disease. Highlights: Using an encoding scheme where information is encoded in signals, their envelopes, and envelopes of envelopes, we show how to implement prediction error and precision modulation in a neural mass model through cross-frequency coupling (CFC). We use the laminar neural mass model (LaNMM), which integrates Jansen-Rit and pyramidal interneuron gamma (PING) submodels to display fast and slow rhythms and provides mechanisms for a) Signal-Envelope Coupling (SEC) , where slow-wave activity modulates the amplitude envelope of fast oscillations (analogous to phase-amplitude coupling), and b) Envelope-Envelope Coupling (EEC) , where the envelopes of slower oscillations modulate the envelopes of higher-frequency rhythms. We show how to use the LaNMM to implement information-based prediction-error evaluation (as used in Active Inference and Kolmogorov Theory), computing the approximate precision-weighted difference between incoming sensory data (envelopes) and internal model predictions (signals or envelopes). We show that using these mechanisms, the Comparator mechanism can operate at multiple levels and timescales, generating fast prediction-error signals (via SEC) and slower gating signals that encode context (e.g., precision) (via EEC). Our model provides insights into the physiological and cognitive consequences of mesoscale circuital alterations in the context of predictive coding. First, we study disorders of fast-spiking interneurons, such as Alzheimer’s Disease (AD). In the early stages of AD, error evaluation and precision are disrupted (inflated error and reduced gating/weight of predictions), leading to higher prediction errors. In later stages, prediction errors are suppressed regardless of predictions or their precision. Then, we show how serotonergic psychedelics increase the effective weight of inputs and diminish that of predictions, resulting in higher prediction error signals. These observations link oscillatory mechanisms and predictive coding alterations, and potentially with the subjective phenomena in each condition—including cognitive decline in AD and hallucinatory states under psychedelics.

Classical serotonergic psychedelics show promise in addressing neurodegenerative disorders such as Alzheimer’s disease by modulating pathological brain dynamics. However, the precise neurobiological mechanisms underlying their effects remain elusive. This study introduces a personalized whole-brain model built upon a laminar neural mass framework to elucidate these effects. Using multimodal neuroimaging data from thirty subjects diagnosed with mild to moderate Alzheimer’s disease, we simulate the impact of serotonin 2A receptor activation, characteristic of psychedelics, on cortical dynamics. By modulating the excitability of layer 5 pyramidal neurons, our models reproduce hallmark changes in EEG power spectra observed under psychedelics, including alpha power suppression and gamma power enhancement. These spectral shifts are shown to correlate strongly with the regional distribution of serotonin 2A receptors. Furthermore, simulated EEG reveals increased complexity and entropy, suggesting restored network function. These findings underscore the potential of serotonergic psychedelics to reestablish healthy oscillatory dynamics in the prodromal and early phases of Alzheimer’s disease and offer mechanistic insights into their potential therapeutic effects in neurodegenerative disorders.