REBUS and the Anarchic Brain: Toward a Unified Model of the Brain Action of Psychedelics.

This paper formulates the action of psychedelics by integrating the free-energy principle and entropic brain hypothesis. We call this formulation relaxed beliefs under psychedelics (REBUS) and the anarchic brain, founded on the principle that-via their entropic effect on spontaneous cortical activity-psychedelics work to relax the precision of high-level priors or beliefs, thereby liberating bottom-up information flow, particularly via intrinsic sources such as the limbic system. We assemble evidence for this model and show how it can explain a broad range of phenomena associated with the psychedelic experience. With regard to their potential therapeutic use, we propose that psychedelics work to relax the precision weighting of pathologically overweighted priors underpinning various expressions of mental illness. We propose that this process entails an increased sensitization of high-level priors to bottom-up signaling (stemming from intrinsic sources), and that this heightened sensitivity enables the potential revision and deweighting of overweighted priors. We end by discussing further implications of the model, such as that psychedelics can bring about the revision of other heavily weighted high-level priors, not directly related to mental health, such as those underlying partisan and/or overly-confident political, religious, and/or philosophical perspectives. SIGNIFICANCE STATEMENT: Psychedelics are capturing interest, with efforts underway to bring psilocybin therapy to marketing authorisation and legal access within a decade, spearheaded by the findings of a series of phase 2 trials. In this climate, a compelling unified model of how psychedelics alter brain function to alter consciousness would have appeal. Towards this end, we have sought to integrate a leading model of global brain function, hierarchical predictive coding, with an often-cited model of the acute action of psychedelics, the entropic brain hypothesis. The resulting synthesis states that psychedelics work to relax high-level priors, sensitising them to liberated bottom-up information flow, which, with the right intention, care provision and context, can help guide and cultivate the revision of entrenched pathological priors.

Dynamic causal modelling revisited.

This paper revisits the dynamic causal modelling of fMRI timeseries by replacing the usual (Taylor) approximation to neuronal dynamics with a neural mass model of the canonical microcircuit. This provides a generative or dynamic causal model of laminar specific responses that can generate haemodynamic and electrophysiological measurements. In principle, this allows the fusion of haemodynamic and (event related or induced) electrophysiological responses. Furthermore, it enables Bayesian model comparison of competing hypotheses about physiologically plausible synaptic effects; for example, does attentional modulation act on superficial or deep pyramidal cells - or both? In this technical note, we describe the resulting dynamic causal model and provide an illustrative application to the attention to visual motion dataset used in previous papers. Our focus here is on how to answer long-standing questions in fMRI; for example, do haemodynamic responses reflect extrinsic (afferent) input from distant cortical regions, or do they reflect intrinsic (recurrent) neuronal activity? To what extent do inhibitory interneurons contribute to neurovascular coupling? What is the relationship between haemodynamic responses and the frequency of induced neuronal activity? This paper does not pretend to answer these questions; rather it shows how they can be addressed using neural mass models of fMRI timeseries.

Caching mechanisms for habit formation in Active Inference.

A popular distinction in the human and animal learning literature is between deliberate (or willed) and habitual (or automatic) modes of control. Extensive evidence indicates that, after sufficient learning, living organisms develop behavioural habits that permit them saving computational resources. Furthermore, humans and other animals are able to transfer control from deliberate to habitual modes (and vice versa), trading off efficiently flexibility and parsimony - an ability that is currently unparalleled by artificial control systems. Here, we discuss a computational implementation of habit formation, and the transfer of control from deliberate to habitual modes (and vice versa) within Active Inference: a computational framework that merges aspects of cybernetic theory and of Bayesian inference. To model habit formation, we endow an Active Inference agent with a mechanism to "cache" (or memorize) policy probabilities from previous trials, and reuse them to skip - in part or in full - the inferential steps of deliberative processing. We exploit the fact that the relative quality of policies, conditioned upon hidden states, is constant over trials; provided that contingencies and prior preferences do not change. This means the only quantity that can change policy selection is the prior distribution over the initial state - where this prior is based upon the posterior beliefs from previous trials. Thus, an agent that caches the quality (or the probability) of policies can safely reuse cached values to save on cognitive and computational resources - unless contingencies change. Our simulations illustrate the computational benefits, but also the limits, of three caching schemes under Active Inference. They suggest that key aspects of habitual behaviour - such as perseveration - can be explained in terms of caching policy probabilities. Furthermore, they suggest that there may be many kinds (or stages) of habitual behaviour, each associated with a different caching scheme; for example, caching associated or not associated with contextual estimation. These schemes are more or less impervious to contextual and contingency changes.

Neurophysiologically-informed markers of individual variability and pharmacological manipulation of human cortical gamma.

The ability to quantify synaptic function at the level of cortical microcircuits from non-invasive data would be enormously useful in the study of neuronal processing in humans and the pathophysiology that attends many neuropsychiatric disorders. Here, we provide proof of principle that one can estimate inter-and intra-laminar interactions among specific neuronal populations using induced gamma responses in the visual cortex of human subjects - using dynamic causal modelling based upon the canonical microcircuit (CMC; a simplistic model of a cortical column). Using variability in induced (spectral) responses over a large cohort of normal subjects, we find that the predominant determinants of gamma responses rest on recurrent and intrinsic connections between superficial pyramidal cells and inhibitory interneurons. Furthermore, variations in beta responses were mediated by inter-subject differences in the intrinsic connections between deep pyramidal cells and inhibitory interneurons. Interestingly, we also show that increasing the self-inhibition of superficial pyramidal cells suppresses the amplitude of gamma activity, while increasing its peak frequency. This systematic and nonlinear relationship was only disclosed by modelling the causes of induced responses. Crucially, we were able to validate this form of neurophysiological phenotyping by showing a selective effect of the GABA re-uptake inhibitor tiagabine on the rate constants of inhibitory interneurons. Remarkably, we were able to recover the pharmacodynamics of this effect over the course of several hours on a per subject basis. These findings speak to the possibility of measuring population specific synaptic function - and its response to pharmacological intervention - to provide subject-specific biomarkers of mesoscopic neuronal processes using non-invasive data. Finally, our results demonstrate that, using the CMC as a proxy, the synaptic mechanisms that underlie the gain control of neuronal message passing within and between different levels of cortical hierarchies may now be amenable to quantitative study using non-invasive (MEG) procedures.

Linking canonical microcircuits and neuronal activity: Dynamic causal modelling of laminar recordings.

Neural models describe brain activity at different scales, ranging from single cells to whole brain networks. Here, we attempt to reconcile models operating at the microscopic (compartmental) and mesoscopic (neural mass) scales to analyse data from microelectrode recordings of intralaminar neural activity. Although these two classes of models operate at different scales, it is relatively straightforward to create neural mass models of ensemble activity that are equipped with priors obtained after fitting data generated by detailed microscopic models. This provides generative (forward) models of measured neuronal responses that retain construct validity in relation to compartmental models. We illustrate our approach using cross spectral responses obtained from V1 during a visual perception paradigm that involved optogenetic manipulation of the basal forebrain. We find that the resulting neural mass model can distinguish between activity in distinct cortical layers - both with and without optogenetic activation - and that cholinergic input appears to enhance (disinhibit) superficial layer activity relative to deep layers. This is particularly interesting from the perspective of predictive coding, where neuromodulators are thought to boost prediction errors that ascend the cortical hierarchy.

Computational neuroimaging strategies for single patient predictions.

Neuroimaging increasingly exploits machine learning techniques in an attempt to achieve clinically relevant single-subject predictions. An alternative to machine learning, which tries to establish predictive links between features of the observed data and clinical variables, is the deployment of computational models for inferring on the (patho)physiological and cognitive mechanisms that generate behavioural and neuroimaging responses. This paper discusses the rationale behind a computational approach to neuroimaging-based single-subject inference, focusing on its potential for characterising disease mechanisms in individual subjects and mapping these characterisations to clinical predictions. Following an overview of two main approaches - Bayesian model selection and generative embedding - which can link computational models to individual predictions, we review how these methods accommodate heterogeneity in psychiatric and neurological spectrum disorders, help avoid erroneous interpretations of neuroimaging data, and establish a link between a mechanistic, model-based approach and the statistical perspectives afforded by machine learning.

Dopamine, Salience, and Response Set Shifting in Prefrontal Cortex.

Dopamine is implicated in multiple functions, including motor execution, action learning for hedonically salient outcomes, maintenance, and switching of behavioral response set. Here, we used a novel within-subject psychopharmacological and combined functional neuroimaging paradigm, investigating the interaction between hedonic salience, dopamine, and response set shifting, distinct from effects on action learning or motor execution. We asked whether behavioral performance in response set shifting depends on the hedonic salience of reversal cues, by presenting these as null (neutral) or salient (monetary loss) outcomes. We observed marked effects of reversal cue salience on set-switching, with more efficient reversals following salient loss outcomes. L-Dopa degraded this discrimination, leading to inappropriate perseveration. Generic activation in thalamus, insula, and striatum preceded response set switches, with an opposite pattern in ventromedial prefrontal cortex (vmPFC). However, the behavioral effect of hedonic salience was reflected in differential vmPFC deactivation following salient relative to null reversal cues. l-Dopa reversed this pattern in vmPFC, suggesting that its behavioral effects are due to disruption of the stability and switching of firing patterns in prefrontal cortex. Our findings provide a potential neurobiological explanation for paradoxical phenomena, including maintenance of behavioral set despite negative outcomes, seen in impulse control disorders in Parkinson's disease.

Estimating anatomical trajectories with Bayesian mixed-effects modeling.

We introduce a mass-univariate framework for the analysis of whole-brain structural trajectories using longitudinal Voxel-Based Morphometry data and Bayesian inference. Our approach to developmental and aging longitudinal studies characterizes heterogeneous structural growth/decline between and within groups. In particular, we propose a probabilistic generative model that parameterizes individual and ensemble average changes in brain structure using linear mixed-effects models of age and subject-specific covariates. Model inversion uses Expectation Maximization (EM), while voxelwise (empirical) priors on the size of individual differences are estimated from the data. Bayesian inference on individual and group trajectories is realized using Posterior Probability Maps (PPM). In addition to parameter inference, the framework affords comparisons of models with varying combinations of model order for fixed and random effects using model evidence. We validate the model in simulations and real MRI data from the Alzheimer's Disease Neuroimaging Initiative (ADNI) project. We further demonstrate how subject specific characteristics contribute to individual differences in longitudinal volume changes in healthy subjects, Mild Cognitive Impairment (MCI), and Alzheimer's Disease (AD).

A DCM study of spectral asymmetries in feedforward and feedback connections between visual areas V1 and V4 in the monkey.

This paper reports a dynamic causal modeling study of electrocorticographic (ECoG) data that addresses functional asymmetries between forward and backward connections in the visual cortical hierarchy. Specifically, we ask whether forward connections employ gamma-band frequencies, while backward connections preferentially use lower (beta-band) frequencies. We addressed this question by modeling empirical cross spectra using a neural mass model equipped with superficial and deep pyramidal cell populations-that model the source of forward and backward connections, respectively. This enabled us to reconstruct the transfer functions and associated spectra of specific subpopulations within cortical sources. We first established that Bayesian model comparison was able to discriminate between forward and backward connections, defined in terms of their cells of origin. We then confirmed that model selection was able to identify extrastriate (V4) sources as being hierarchically higher than early visual (V1) sources. Finally, an examination of the auto spectra and transfer functions associated with superficial and deep pyramidal cells confirmed that forward connections employed predominantly higher (gamma) frequencies, while backward connections were mediated by lower (alpha/beta) frequencies. We discuss these findings in relation to current views about alpha, beta, and gamma oscillations and predictive coding in the brain.

Dynamic causal modelling for functional near-infrared spectroscopy.

Functional near-infrared spectroscopy (fNIRS) is an emerging technique for measuring changes in cerebral hemoglobin concentration via optical absorption changes. Although there is great interest in using fNIRS to study brain connectivity, current methods are unable to infer the directionality of neuronal connections. In this paper, we apply Dynamic Causal Modelling (DCM) to fNIRS data. Specifically, we present a generative model of how observed fNIRS data are caused by interactions among hidden neuronal states. Inversion of this generative model, using an established Bayesian framework (variational Laplace), then enables inference about changes in directed connectivity at the neuronal level. Using experimental data acquired during motor imagery and motor execution tasks, we show that directed (i.e., effective) connectivity from the supplementary motor area to the primary motor cortex is negatively modulated by motor imagery, and this suppressive influence causes reduced activity in the primary motor cortex during motor imagery. These results are consistent with findings of previous functional magnetic resonance imaging (fMRI) studies, suggesting that the proposed method enables one to infer directed interactions in the brain mediated by neuronal dynamics from measurements of optical density changes.

Bayesian model selection for group studies - revisited.

In this paper, we revisit the problem of Bayesian model selection (BMS) at the group level. We originally addressed this issue in Stephan et al. (2009), where models are treated as random effects that could differ between subjects, with an unknown population distribution. Here, we extend this work, by (i) introducing the Bayesian omnibus risk (BOR) as a measure of the statistical risk incurred when performing group BMS, (ii) highlighting the difference between random effects BMS and classical random effects analyses of parameter estimates, and (iii) addressing the problem of between group or condition model comparisons. We address the first issue by quantifying the chance likelihood of apparent differences in model frequencies. This leads to the notion of protected exceedance probabilities. The second issue arises when people want to ask "whether a model parameter is zero or not" at the group level. Here, we provide guidance as to whether to use a classical second-level analysis of parameter estimates, or random effects BMS. The third issue rests on the evidence for a difference in model labels or frequencies across groups or conditions. Overall, we hope that the material presented in this paper finesses the problems of group-level BMS in the analysis of neuroimaging and behavioural data.

Efficient gradient computation for dynamical models.

Data assimilation is a fundamental issue that arises across many scales in neuroscience - ranging from the study of single neurons using single electrode recordings to the interaction of thousands of neurons using fMRI. Data assimilation involves inverting a generative model that can not only explain observed data but also generate predictions. Typically, the model is inverted or fitted using conventional tools of (convex) optimization that invariably extremise some functional - norms, minimum descriptive length, variational free energy, etc. Generally, optimisation rests on evaluating the local gradients of the functional to be optimized. In this paper, we compare three different gradient estimation techniques that could be used for extremising any functional in time - (i) finite differences, (ii) forward sensitivities and a method based on (iii) the adjoint of the dynamical system. We demonstrate that the first-order gradients of a dynamical system, linear or non-linear, can be computed most efficiently using the adjoint method. This is particularly true for systems where the number of parameters is greater than the number of states. For such systems, integrating several sensitivity equations - as required with forward sensitivities - proves to be most expensive, while finite-difference approximations have an intermediate efficiency. In the context of neuroimaging, adjoint based inversion of dynamical causal models (DCMs) can, in principle, enable the study of models with large numbers of nodes and parameters.

Algorithmic procedures for Bayesian MEG/EEG source reconstruction in SPM.

The MEG/EEG inverse problem is ill-posed, giving different source reconstructions depending on the initial assumption sets. Parametric Empirical Bayes allows one to implement most popular MEG/EEG inversion schemes (Minimum Norm, LORETA, etc.) within the same generic Bayesian framework. It also provides a cost-function in terms of the variational Free energy-an approximation to the marginal likelihood or evidence of the solution. In this manuscript, we revisit the algorithm for MEG/EEG source reconstruction with a view to providing a didactic and practical guide. The aim is to promote and help standardise the development and consolidation of other schemes within the same framework. We describe the implementation in the Statistical Parametric Mapping (SPM) software package, carefully explaining each of its stages with the help of a simple simulated data example. We focus on the Multiple Sparse Priors (MSP) model, which we compare with the well-known Minimum Norm and LORETA models, using the negative variational Free energy for model comparison. The manuscript is accompanied by Matlab scripts to allow the reader to test and explore the underlying algorithm.

Contrast gain control and horizontal interactions in V1: a DCM study.

Using high-density electrocorticographic recordings - from awake-behaving monkeys - and dynamic causal modelling, we characterised contrast dependent gain control in visual cortex, in terms of synaptic rate constants and intrinsic connectivity. Specifically, we used neural field models to quantify the balance of excitatory and inhibitory influences; both in terms of the strength and spatial dispersion of horizontal intrinsic connections. Our results allow us to infer that increasing contrast increases the sensitivity or gain of superficial pyramidal cells to inputs from spiny stellate populations. Furthermore, changes in the effective spatial extent of horizontal coupling nuance the spatiotemporal filtering properties of cortical laminae in V1 - effectively preserving higher spatial frequencies. These results are consistent with recent non-invasive human studies of contrast dependent changes in the gain of pyramidal cells elaborating forward connections - studies designed to test specific hypotheses about precision and gain control based on predictive coding. Furthermore, they are consistent with established results showing that the receptive fields of V1 units shrink with increasing visual contrast.

Anatomical connectivity and the resting state activity of large cortical networks.

This paper uses mathematical modelling and simulations to explore the dynamics that emerge in large scale cortical networks, with a particular focus on the topological properties of the structural connectivity and its relationship to functional connectivity. We exploit realistic anatomical connectivity matrices (from diffusion spectrum imaging) and investigate their capacity to generate various types of resting state activity. In particular, we study emergent patterns of activity for realistic connectivity configurations together with approximations formulated in terms of neural mass or field models. We find that homogenous connectivity matrices, of the sort of assumed in certain neural field models give rise to damped spatially periodic modes, while more localised modes reflect heterogeneous coupling topologies. When simulating resting state fluctuations under realistic connectivity, we find no evidence for a spectrum of spatially periodic patterns, even when grouping together cortical nodes into communities, using graph theory. We conclude that neural field models with translationally invariant connectivity may be best applied at the mesoscopic scale and that more general models of cortical networks that embed local neural fields, may provide appropriate models of macroscopic cortical dynamics over the whole brain.

Nonlinear coupling between occipital and motor cortex during motor imagery: a dynamic causal modeling study.

We demonstrate the capacity of dynamic causal modeling to characterize the nonlinear coupling among cortical sources that underlie time-frequency modulations in MEG data. Our experimental task involved the mental rotation of hand drawings that ten subjects used to decide if it was a right or left hand. Reaction times were shorter when the stimuli were presented with a small rotation angle (fast responses) compared to a large rotation angle (slow responses). The grand-averaged data showed that in both cases performance was accompanied by a marked increase in gamma activity in occipital areas and a concomitant decrease in alpha and beta power in occipital and motor regions. Modeling directed (cross) frequency interactions between the two regions revealed that after the stimulus induced a gamma increase and beta decrease in occipital regions, interactions with the motor area served to attenuate these modulations. The difference between fast and slow behavioral responses was manifest as an altered coupling strength in both forward and backward connections, which led to a less pronounced attenuation for more difficult (slow reaction time) trials. This was mediated by a (backwards) beta to gamma coupling from motor till occipital sources, whereas other interactions were mainly within the same frequency. Results are consistent with the theory of predictive coding and suggest that during motor imagery, the influence of motor areas on activity in occipital cortex co-determines performance. Our study illustrates the benefit of modeling experimental responses in terms of a generative model that can disentangle the contributions of intra-areal vis-à-vis inter-areal connections to time-frequency modulations during task performance.

Dynamic causal modelling of lateral interactions in the visual cortex.

This paper presents a dynamic causal model based upon neural field models of the Amari type. We consider the application of these models to non-invasive data, with a special focus on the mapping from source activity on the cortical surface to a single channel. We introduce a neural field model based upon the canonical microcircuit (CMC), in which neuronal populations are assigned to different cortical layers. We show that DCM can disambiguate between alternative (neural mass and field) models of cortical activity. However, unlike neural mass models, DCM with neural fields can address questions about neuronal microcircuitry and lateral interactions. This is because they are equipped with interlaminar connections and horizontal intra-laminar connections that are patchy in nature. These horizontal or lateral connections can be regarded as connecting macrocolumns with similar feature selectivity. Crucially, the spatial parameters governing horizontal connectivity determine the separation (width) of cortical macrocolumns. Thus we can estimate the width of macro columns, using non-invasive electromagnetic signals. We illustrate this estimation using dynamic causal models of steady-state or ongoing spectral activity measured using magnetoencephalography (MEG) in human visual cortex. Specifically, we revisit the hypothesis that the size of a macrocolumn is a key determinant of neuronal dynamics, particularly the peak gamma frequency. We are able to show a correlation, over subjects, between columnar size and peak gamma frequency - that fits comfortably with established correlations between peak gamma frequency and the size of visual cortex defined retinotopically. We also considered cortical excitability and assessed its relative influence on observed gamma activity. This example highlights the potential utility of dynamic causal modelling and neural fields in providing quantitative characterisations of spatially extended dynamics on the cortical surface - that are parameterised in terms of horizontal connections, implicit in the cortical micro-architecture and its synaptic parameters.

Learning and generalization under ambiguity: an fMRI study.

Adaptive behavior often exploits generalizations from past experience by applying them judiciously in new situations. This requires a means of quantifying the relative importance of prior experience and current information, so they can be balanced optimally. In this study, we ask whether the brain generalizes in an optimal way. Specifically, we used Bayesian learning theory and fMRI to test whether neuronal responses reflect context-sensitive changes in ambiguity or uncertainty about experience-dependent beliefs. We found that the hippocampus expresses clear ambiguity-dependent responses that are associated with an augmented rate of learning. These findings suggest candidate neuronal systems that may be involved in aberrations of generalization, such as over-confidence.

Canalization and plasticity in psychopathology.

This theoretical article revives a classical bridging construct, canalization, to describe a new model of a general factor of psychopathology. To achieve this, we have distinguished between two types of plasticity, an early one that we call 'TEMP' for 'Temperature or Entropy Mediated Plasticity', and another, we call 'canalization', which is close to Hebbian plasticity. These two forms of plasticity can be most easily distinguished by their relationship to 'precision' or inverse variance; TEMP relates to increased model variance or decreased precision, whereas the opposite is true for canalization. TEMP also subsumes increased learning rate, (Ising) temperature and entropy. Dictionary definitions of 'plasticity' describe it as the property of being easily shaped or molded; TEMP is the better match for this. Importantly, we propose that 'pathological' phenotypes develop via mechanisms of canalization or increased model precision, as a defensive response to adversity and associated distress or dysphoria. Our model states that canalization entrenches in psychopathology, narrowing the phenotypic state-space as the agent develops expertise in their pathology. We suggest that TEMP - combined with gently guiding psychological support - can counter canalization. We address questions of whether and when canalization is adaptive versus maladaptive, furnish our model with references to basic and human neuroscience, and offer concrete experiments and measures to test its main hypotheses and implications. This article is part of the Special Issue on "National Institutes of Health Psilocybin Research Speaker Series".

The influence of spontaneous activity on stimulus processing in primary visual cortex.

Spontaneous activity in the resting human brain has been studied extensively; however, how such activity affects the local processing of a sensory stimulus is relatively unknown. Here, we examined the impact of spontaneous activity in primary visual cortex on neuronal and behavioural responses to a simple visual stimulus, using functional MRI. Stimulus-evoked responses remained essentially unchanged by spontaneous fluctuations, combining with them in a largely linear fashion (i.e., with little evidence for an interaction). However, interactions between spontaneous fluctuations and stimulus-evoked responses were evident behaviourally; high levels of spontaneous activity tended to be associated with increased stimulus detection at perceptual threshold. Our results extend those found in studies of spontaneous fluctuations in motor cortex and higher order visual areas, and suggest a fundamental role for spontaneous activity in stimulus processing.