Mutual information (MI) is a useful information-theoretic measure to quantify the statistical dependence between two random variables: X and Y. Often, we are interested in understanding how the dependence between X and Y in one set of samples compares to another. Although the dependence between X and Y in each set of samples can be measured separately using MI, these estimates cannot be compared directly if they are based on samples from a non-stationary distribution. Here, we propose an alternative measure for characterising how the dependence between X and Y as defined by one set of samples is expressed in another, cross MI. We present a comprehensive set of simulation studies sampling data with X-Y dependencies to explore this measure. Finally, we discuss how this relates to measures of model fit in linear regression, and some future applications in neuroimaging data analysis.

Contemporary models of perceptual awareness lack tractable neurobiological constraints. Inspired by recent cellular recordings in a mouse model of tactile threshold detection, we constructed a biophysical model of perceptual awareness that incorporated essential features of thalamocortical anatomy and cellular physiology. Our model reproduced, and mechanistically explains, the key in vivo neural and behavioural signatures of perceptual awareness in the mouse model, as well as the response to a set of causal perturbations. We generalised the same model (with identical parameters) to a more complex task - visual rivalry - and found that the same thalamic-mediated mechanism of perceptual awareness determined perceptual dominance. This led to the generation of a set of novel, and directly testable, electrophysiological predictions. Analyses of the model based on dynamical systems theory show that perceptual awareness in simulations of both threshold detection and visual rivalry arises from the emergent systems-level dynamics of thalamocortical loops.

Biological neural networks, unlike artificial neural networks (ANNs), can process information from data that is inherently noisy, unstructured, sparse and dynamic. How is this possible? And how can we replicate this? A crucial clue comes from neuroscience: the brain is a complex physical system and the network topology of its neural circuitry is a determinant of its emergent collective properties. Indeed, as the brain's neural network is already entrained in its physical hardware, it clearly does not require an ANN software add-on to learn from natural data. Self-assembled nanowire networks with memristive junctions represent arguably the closest hardware architecture to real biological neural networks and are thus uniquely placed to demonstrate genuinely neuromorphic information processing. Here, we present preliminary results on polymer-coated silver nanowire networks. Their neuromorphic architecture (densely structured network topology, memristive switch junctions and efficient interconnect) gives rise to a rich repertoire of collective nonlinear dynamics manifested through adaptive current transport pathways. The potential for associative learning is demonstrated in a test protocol in which a nanowire network is stimulated by multiple electrodes mapped to different spatial patterns. The capacity to process information in the temporal domain is demonstrated via simulations of a reservoir computing implementation in which nanowire networks are shown to perform tasks such as time series prediction and handwritten digit recognition. Overall, their unique properties and neuromorphic information processing capabilities make nanowire networks promising candidates for emerging applications in cognitive devices in particular, at the edge.

A key component of the flexibility and complexity of the brain is its ability to dynamically adapt its functional network structure between integrated and segregated brain states depending on the demands of different cognitive tasks. Integrated states are prevalent when performing tasks of high complexity, such as maintaining items in working memory, consistent with models of a global workspace architecture. Recent work has suggested that the balance between integration and segregation is under the control of ascending neuromodulatory systems, such as the noradrenergic system, via changes in neural gain (in terms of the amplification and non-linearity in stimulus-response transfer function of brain regions). In a previous large-scale nonlinear oscillator model of neuronal network dynamics, we showed that manipulating neural gain parameters led to a 'critical' transition in phase synchrony that was associated with a shift from segregated to integrated topology, thus confirming our original prediction. In this study, we advance these results by demonstrating that the gain-mediated phase transition is characterized by a shift in the underlying dynamics of neural information processing. Specifically, the dynamics of the subcritical (segregated) regime are dominated by information storage, whereas the supercritical (integrated) regime is associated with increased information transfer (measured via transfer entropy). Operating near to the critical regime with respect to modulating neural gain parameters would thus appear to provide computational advantages, offering flexibility in the information processing that can be performed with only subtle changes in gain control. Our results thus link studies of whole-brain network topology and the ascending arousal system with information processing dynamics, and suggest that the constraints imposed by the ascending arousal system constrain low-dimensional modes of information processing within the brain.