ABSTRACT
Phase transitions abound in nature and society, and, from species extinction to stock market collapse, their prediction is of widespread importance. In earlier work we showed that Global Transfer Entropy, a general measure of information flow, was found to peaks away from the transition on the disordered side for the Ising model, a canonical second order transition. Here we show that (a) global transfer entropy also peaks on the disordered side of the transition of finite first order transitions, such as ecology dynamics on coral reefs, which have latent heat and no correlation length divergence, and (b) analysis of information flow across state boundaries unifies both transition orders. We obtain the first information-theoretic result for the high-order Potts model and the first demonstration of early warning of a first order transition. The unexpected earlier finding that global transfer entropy peaks on the disordered side of a transition is also found for finite first order systems, albeit not in the thermodynamic limit. By noting that the interface length of clusters in each phase is the dominant region of information flow, we unify the information theoretic behaviour of first and second order transitions.
Subject(s)
Hot Temperature , Entropy , Phase Transition , ThermodynamicsABSTRACT
Neuroimaging studies of the psychedelic state offer a unique window onto the neural basis of conscious perception and selfhood. Despite well understood pharmacological mechanisms of action, the large-scale changes in neural dynamics induced by psychedelic compounds remain poorly understood. Using source-localised, steady-state MEG recordings, we describe changes in functional connectivity following the controlled administration of LSD, psilocybin and low-dose ketamine, as well as, for comparison, the (non-psychedelic) anticonvulsant drug tiagabine. We compare both undirected and directed measures of functional connectivity between placebo and drug conditions. We observe a general decrease in directed functional connectivity for all three psychedelics, as measured by Granger causality, throughout the brain. These data support the view that the psychedelic state involves a breakdown in patterns of functional organisation or information flow in the brain. In the case of LSD, the decrease in directed functional connectivity is coupled with an increase in undirected functional connectivity, which we measure using correlation and coherence. This surprising opposite movement of directed and undirected measures is of more general interest for functional connectivity analyses, which we interpret using analytical modelling. Overall, our results uncover the neural dynamics of information flow in the psychedelic state, and highlight the importance of comparing multiple measures of functional connectivity when analysing time-resolved neuroimaging data.
Subject(s)
Cerebral Cortex/drug effects , Connectome , Hallucinogens/pharmacology , Ketamine/pharmacology , Lysergic Acid Diethylamide/pharmacology , Magnetoencephalography/drug effects , Nerve Net/drug effects , Psilocybin/pharmacology , Adult , Anticonvulsants/pharmacology , Cerebral Cortex/physiology , Female , Hallucinogens/administration & dosage , Humans , Ketamine/administration & dosage , Lysergic Acid Diethylamide/administration & dosage , Male , Nerve Net/physiology , Psilocybin/administration & dosage , Tiagabine/pharmacology , Young AdultABSTRACT
Granger-Geweke causality (GGC) is a powerful and popular method for identifying directed functional ('causal') connectivity in neuroscience. In a recent paper, Stokes and Purdon (2017b) raise several concerns about its use. They make two primary claims: (1) that GGC estimates may be severely biased or of high variance, and (2) that GGC fails to reveal the full structural/causal mechanisms of a system. However, these claims rest, respectively, on an incomplete evaluation of the literature, and a misconception about what GGC can be said to measure. Here we explain how existing approaches resolve the first issue, and discuss the frequently-misunderstood distinction between functional and effective neural connectivity which underlies Stokes and Purdon's second claim.
Subject(s)
Neurosciences , RestABSTRACT
BACKGROUND: Granger causality is well established within the neurosciences for inference of directed functional connectivity from neurophysiological data. These data usually consist of time series which subsample a continuous-time biophysiological process. While it is well known that subsampling can lead to imputation of spurious causal connections where none exist, less is known about the effects of subsampling on the ability to reliably detect causal connections which do exist. NEW METHOD: We present a theoretical analysis of the effects of subsampling on Granger-causal inference. Neurophysiological processes typically feature signal propagation delays on multiple time scales; accordingly, we base our analysis on a distributed-lag, continuous-time stochastic model, and consider Granger causality in continuous time at finite prediction horizons. Via exact analytical solutions, we identify relationships among sampling frequency, underlying causal time scales and detectability of causalities. RESULTS: We reveal complex interactions between the time scale(s) of neural signal propagation and sampling frequency. We demonstrate that detectability decays exponentially as the sample time interval increases beyond causal delay times, identify detectability "black spots" and "sweet spots", and show that downsampling may potentially improve detectability. We also demonstrate that the invariance of Granger causality under causal, invertible filtering fails at finite prediction horizons, with particular implications for inference of Granger causality from fMRI data. COMPARISON WITH EXISTING METHODS: Our analysis emphasises that sampling rates for causal analysis of neurophysiological time series should be informed by domain-specific time scales, and that state-space modelling should be preferred to purely autoregressive modelling. CONCLUSIONS: On the basis of a very general model that captures the structure of neurophysiological processes, we are able to help identify confounds, and offer practical insights, for successful detection of causal connectivity from neurophysiological recordings.
Subject(s)
Algorithms , Models, Statistical , Nervous System Physiological Phenomena , Humans , Models, Neurological , Signal Processing, Computer-AssistedABSTRACT
Key to understanding the neuronal basis of consciousness is the characterization of the neural signatures of changes in level of consciousness during sleep. Here we analysed three measures of dynamical complexity on spontaneous depth electrode recordings from 10 epilepsy patients during wakeful rest (WR) and different stages of sleep: (i) Lempel-Ziv complexity, which is derived from how compressible the data are; (ii) amplitude coalition entropy, which measures the variability over time of the set of channels active above a threshold; (iii) synchrony coalition entropy, which measures the variability over time of the set of synchronous channels. When computed across sets of channels that are broadly distributed across multiple brain regions, all three measures decreased substantially in all participants during early-night non-rapid eye movement (NREM) sleep. This decrease was partially reversed during late-night NREM sleep, while the measures scored similar to WR during rapid eye movement (REM) sleep. This global pattern was in almost all cases mirrored at the local level by groups of channels located in a single region. In testing for differences between regions, we found elevated signal complexity in the frontal lobe. These differences could not be attributed solely to changes in spectral power between conditions. Our results provide further evidence that the level of consciousness correlates with neural dynamical complexity.
ABSTRACT
Granger causality has long been a prominent method for inferring causal interactions between stochastic variables for a broad range of complex physical systems. However, it has been recognized that a moving average (MA) component in the data presents a serious confound to Granger causal analysis, as routinely performed via autoregressive (AR) modeling. We solve this problem by demonstrating that Granger causality may be calculated simply and efficiently from the parameters of a state-space (SS) model. Since SS models are equivalent to autoregressive moving average models, Granger causality estimated in this fashion is not degraded by the presence of a MA component. This is of particular significance when the data has been filtered, downsampled, observed with noise, or is a subprocess of a higher dimensional process, since all of these operations-commonplace in application domains as diverse as climate science, econometrics, and the neurosciences-induce a MA component. We show how Granger causality, conditional and unconditional, in both time and frequency domains, may be calculated directly from SS model parameters via solution of a discrete algebraic Riccati equation. Numerical simulations demonstrate that Granger causality estimators thus derived have greater statistical power and smaller bias than AR estimators. We also discuss how the SS approach facilitates relaxation of the assumptions of linearity, stationarity, and homoscedasticity underlying current AR methods, thus opening up potentially significant new areas of research in Granger causal analysis.
ABSTRACT
BACKGROUND: Wiener-Granger causality ("G-causality") is a statistical notion of causality applicable to time series data, whereby cause precedes, and helps predict, effect. It is defined in both time and frequency domains, and allows for the conditioning out of common causal influences. Originally developed in the context of econometric theory, it has since achieved broad application in the neurosciences and beyond. Prediction in the G-causality formalism is based on VAR (vector autoregressive) modelling. NEW METHOD: The MVGC Matlab© Toolbox approach to G-causal inference is based on multiple equivalent representations of a VAR model by (i) regression parameters, (ii) the autocovariance sequence and (iii) the cross-power spectral density of the underlying process. It features a variety of algorithms for moving between these representations, enabling selection of the most suitable algorithms with regard to computational efficiency and numerical accuracy. RESULTS: In this paper we explain the theoretical basis, computational strategy and application to empirical G-causal inference of the MVGC Toolbox. We also show via numerical simulations the advantages of our Toolbox over previous methods in terms of computational accuracy and statistical inference. COMPARISON WITH EXISTING METHOD(S): The standard method of computing G-causality involves estimation of parameters for both a full and a nested (reduced) VAR model. The MVGC approach, by contrast, avoids explicit estimation of the reduced model, thus eliminating a source of estimation error and improving statistical power, and in addition facilitates fast and accurate estimation of the computationally awkward case of conditional G-causality in the frequency domain. CONCLUSIONS: The MVGC Toolbox implements a flexible, powerful and efficient approach to G-causal inference.
Subject(s)
Algorithms , Computer Simulation , Models, Theoretical , Signal Processing, Computer-Assisted , Software , Humans , Time FactorsABSTRACT
There is growing evidence that for a range of dynamical systems featuring complex interactions between large ensembles of interacting elements, mutual information peaks at order-disorder phase transitions. We conjecture that, by contrast, information flow in such systems will generally peak strictly on the disordered side of a phase transition. This conjecture is verified for a ferromagnetic 2D lattice Ising model with Glauber dynamics and a transfer entropy-based measure of systemwide information flow. Implications of the conjecture are considered, in particular, that for a complex dynamical system in the process of transitioning from disordered to ordered dynamics (a mechanism implicated, for example, in financial market crashes and the onset of some types of epileptic seizures); information dynamics may be able to predict an imminent transition.
ABSTRACT
Granger causality is a method for identifying directed functional connectivity based on time series analysis of precedence and predictability. The method has been applied widely in neuroscience, however its application to functional MRI data has been particularly controversial, largely because of the suspicion that Granger causal inferences might be easily confounded by inter-regional differences in the hemodynamic response function. Here, we show both theoretically and in a range of simulations, that Granger causal inferences are in fact robust to a wide variety of changes in hemodynamic response properties, including notably their time-to-peak. However, when these changes are accompanied by severe downsampling, and/or excessive measurement noise, as is typical for current fMRI data, incorrect inferences can still be drawn. Our results have important implications for the ongoing debate about lag-based analyses of functional connectivity. Our methods, which include detailed spiking neuronal models coupled to biophysically realistic hemodynamic observation models, provide an important 'analysis-agnostic' platform for evaluating functional and effective connectivity methods.
Subject(s)
Brain/physiology , Computer Simulation , Hemodynamics/physiology , Image Processing, Computer-Assisted/methods , Magnetic Resonance Imaging/methods , Algorithms , Humans , Models, Neurological , Neural Pathways/physiology , Neurons/physiologyABSTRACT
Transfer entropy, an information-theoretic measure of time-directed information transfer between joint processes, has steadily gained popularity in the analysis of complex stochastic dynamics in diverse fields, including the neurosciences, ecology, climatology, and econometrics. We show that for a broad class of predictive models, the log-likelihood ratio test statistic for the null hypothesis of zero transfer entropy is a consistent estimator for the transfer entropy itself. For finite Markov chains, furthermore, no explicit model is required. In the general case, an asymptotic χ2 distribution is established for the transfer entropy estimator. The result generalizes the equivalence in the Gaussian case of transfer entropy and Granger causality, a statistical notion of causal influence based on prediction via vector autoregression, and establishes a fundamental connection between directed information transfer and causality in the Wiener-Granger sense.
ABSTRACT
An outstanding challenge in neuroscience is to develop theoretically grounded and practically applicable quantitative measures that are sensitive to conscious level. Such measures should be high for vivid alert conscious wakefulness, and low for unconscious states such as dreamless sleep, coma and general anaesthesia. Here, we describe recent progress in the development of measures of dynamical complexity, in particular causal density and integrated information. These and similar measures capture in different ways the extent to which a system's dynamics are simultaneously differentiated and integrated. Because conscious scenes are distinguished by the same dynamical features, these measures are therefore good candidates for reflecting conscious level. After reviewing the theoretical background, we present new simulation results demonstrating similarities and differences between the measures, and we discuss remaining challenges in the practical application of the measures to empirically obtained data.
Subject(s)
Consciousness , Information Theory , Models, Neurological , Multivariate AnalysisABSTRACT
Granger causality (G-causality) is increasingly employed as a method for identifying directed functional connectivity in neural time series data. However, little attention has been paid to the influence of common preprocessing methods such as filtering on G-causality inference. Filtering is often used to remove artifacts from data and/or to isolate frequency bands of interest. Here, we show [following Geweke (1982)] that G-causality for a stationary vector autoregressive (VAR) process is fully invariant under the application of an arbitrary invertible filter; therefore filtering cannot and does not isolate frequency-specific G-causal inferences. We describe and illustrate a simple alternative: integration of frequency domain (spectral) G-causality over the appropriate frequencies ("band limited G-causality"). We then show, using an analytically solvable minimal model, that in practice G-causality inferences often do change after filtering, as a consequence of large increases in empirical model order induced by filtering. Finally, we demonstrate a valid application of filtering in removing a nonstationary ("line noise") component from data. In summary, when applied carefully, filtering can be a useful preprocessing step for removing artifacts and for furnishing or improving stationarity; however filtering is inappropriate for isolating causal influences within specific frequency bands.
Subject(s)
Algorithms , Electrophysiology/methods , Nerve Net/physiology , Signal Processing, Computer-Assisted , Action Potentials/physiology , Animals , Artifacts , Computer Simulation , Humans , Neural Networks, ComputerABSTRACT
Granger causality analysis is a popular method for inference on directed interactions in complex systems of many variables. A shortcoming of the standard framework for Granger causality is that it only allows for examination of interactions between single (univariate) variables within a system, perhaps conditioned on other variables. However, interactions do not necessarily take place between single variables but may occur among groups or "ensembles" of variables. In this study we establish a principled framework for Granger causality in the context of causal interactions among two or more multivariate sets of variables. Building on Geweke's seminal 1982 work, we offer additional justifications for one particular form of multivariate Granger causality based on the generalized variances of residual errors. Taken together, our results support a comprehensive and theoretically consistent extension of Granger causality to the multivariate case. Treated individually, they highlight several specific advantages of the generalized variance measure, which we illustrate using applications in neuroscience as an example. We further show how the measure can be used to define "partial" Granger causality in the multivariate context and we also motivate reformulations of "causal density" and "Granger autonomy." Our results are directly applicable to experimental data and promise to reveal new types of functional relations in complex systems, neural and otherwise.
Subject(s)
Multivariate Analysis , Entropy , Normal DistributionABSTRACT
Granger causality is a statistical notion of causal influence based on prediction via vector autoregression. Developed originally in the field of econometrics, it has since found application in a broader arena, particularly in neuroscience. More recently transfer entropy, an information-theoretic measure of time-directed information transfer between jointly dependent processes, has gained traction in a similarly wide field. While it has been recognized that the two concepts must be related, the exact relationship has until now not been formally described. Here we show that for Gaussian variables, Granger causality and transfer entropy are entirely equivalent, thus bridging autoregressive and information-theoretic approaches to data-driven causal inference.