Publications

Individual differences in cognitive efficiency, particularly in relation to working memory (WM), have been associated both with personality dimensions that reflect enduring regularities in brain configuration, and with short-term neural plasticity, that reflects task-related changes in brain connectivity. To elucidate the relationship of these two divergent mechanisms, we tested the hypothesis that personality dimensions, which reflect enduring aspects of brain configuration, inform about the neurobiological framework within which short-term, task-related plasticity, as measured by effective connectivity, can be facilitated or constrained. As WM consistently engages the dorsolateral prefrontal (DLPFC), parietal (PAR), and anterior cingulate cortex (ACC), we specified a WM network model with bidirectional, ipsilateral, and contralateral connections between these regions from a functional magnetic resonance imaging dataset obtained from 40 healthy adults while performing the 3-back WM task. Task-related effective connectivity changes within this network were estimated using Dynamic Causal Modelling. Personality was evaluated along the major dimensions of Neuroticism, Extraversion, Openness to Experience, Agreeableness, and Conscientiousness. Only two dimensions were relevant to task-dependent effective connectivity. Neuroticism and Conscientiousness respectively constrained and facilitated neuroplastic responses within the WM network. These results suggest individual differences in cognitive efficiency arise from the interplay between enduring and short-term plasticity in brain configuration.

Empirical evidence suggests the incentive value of an option is affected by other options available during choice and by options presented in the past. These contextual effects are hard to reconcile with classical theories and have inspired accounts where contextual influences play a crucial role. However, each account only addresses one or the other of the empirical findings and a unifying perspective has been elusive. Here, we offer a unifying theory of context effects on incentive value attribution and choice based on normative Bayesian principles. This formulation assumes that incentive value corresponds to a precision-weighted prediction error, where predictions are based upon expectations about reward. We show that this scheme explains a wide range of contextual effects, such as those elicited by other options available during choice (or within-choice context effects). These include both conditions in which choice requires an integration of multiple attributes and conditions where a multi-attribute integration is not necessary. Moreover, the same scheme explains context effects elicited by options presented in the past or between-choice context effects. Our formulation encompasses a wide range of contextual influences (comprising both within- and between-choice effects) by calling on Bayesian principles, without invoking ad-hoc assumptions. This helps clarify the contextual nature of incentive value and choice behaviour and may offer insights into psychopathologies characterized by dysfunctional decision-making, such as addiction and pathological gambling.

Abstract We show how any finite physical system with morphological, i.e. three-dimensional embedding or shape, degrees of freedom and locally limited free energy will, under the constraints of the free energy principle, evolve over time towards a neuromorphic morphology that supports hierarchical computations in which each ‘level’ of the hierarchy enacts a coarse-graining of its inputs, and dually, a fine-graining of its outputs. Such hierarchies occur throughout biology, from the architectures of intracellular signal transduction pathways to the large-scale organization of perception and action cycles in the mammalian brain. The close formal connections between cone-cocone diagrams (CCCD) as models of quantum reference frames on the one hand, and between CCCDs and topological quantum field theories on the other, allow the representation of such computations in the fully-general quantum-computational framework of topological quantum neural networks.

Dynamic causal modelling (DCM) was introduced to study the effective connectivity among brain regions using neuroimaging data. Until recently, DCM relied on deterministic models of distributed neuronal responses to external perturbation (e.g., sensory stimulation or task demands). However, accounting for stochastic fluctuations in neuronal activity and their interaction with task-specific processes may be of particular importance for studying state-dependent interactions. Furthermore, allowing for random neuronal fluctuations may render DCM more robust to model misspecification and finesse problems with network identification. In this article, we examine stochastic dynamic causal models (sDCM) in relation to their deterministic counterparts (dDCM) and highlight questions that can only be addressed with sDCM. We also compare the network identification performance of deterministic and stochastic DCM, using Monte Carlo simulations and an empirical case study of absence epilepsy. For example, our results demonstrate that stochastic DCM can exploit the modelling of neural noise to discriminate between direct and mediated connections. We conclude with a discussion of the added value and limitations of sDCM, in relation to its deterministic homologue.

Active inference is a corollary of the Free Energy Principle that prescribes how self-organizing biological agents interact with their environment. The study of active inference processes relies on the definition of a generative probabilistic model and a description of how a free energy functional is minimized by neuronal message passing under that model. This paper presents a tutorial introduction to specifying active inference processes by Forney-style factor graphs (FFG). The FFG framework provides both an insightful representation of the probabilistic model and a biologically plausible inference scheme that, in principle, can be automatically executed in a computer simulation. As an illustrative example, we present an FFG for a deep temporal active inference process. The graph clearly shows how policy selection by expected free energy minimization results from free energy minimization per se, in an appropriate generative policy model.

People with autism spectrum disorder (ASD) experience the world differently, but the neurocognitive mechanisms that underlie these differences are poorly understood. In neurotypical (NT) adults, everyday perceptions are accompanied a sense of reliability, or confidence, which promotes flexible learning and adaptive behaviour. One novel hypothesis is that perception is not compromised in ASD per se, but rather the ability to identify the fidelity of one’s own perception; i.e. metacognition. To test this, we measured perceptual metacognitive efficiency in adults with and without ASD using a visual discrimination task in combination with hierarchical decision theoretic modelling. Our results reveal that adults with ASD exhibit reduced correspondence between accuracy and confidence in perception, despite equivalent task performance. We then demonstrate that the relationship between confidence and pupil dilation (a proxy for noradrenergic function) shows reduced dynamic range in autism; diminished response pre-perceptually, and enhanced response post-perceptually. Our results offer empirical support for previously hypothesised, but never tested, metacognitive and noradrenergic alterations in ASD, offering mechanistic insights into the non-social features of the condition.

Abstract To what extent is predictive processing enactivist? This overview of predictive processing and active inference offers an account of embodied exchange with the world that associates neuronal processing with inferring the causes of our sensations. Crucially, one of the most important causes of the sensorium is our own behaviour, which—perhaps counterintuitively—has to be inferred. The agenda—in active inference—is to link formal (mathematical) descriptions of dynamical systems to a description of active perception in terms of beliefs and goals. In short, it tries to understand why and how we gather evidence for our own existence. There are several approaches to this sort of formulation. This chapter compares and contrasts a Helmholtzian or Kantian (bottom-up) approach—via perception—with the sort of (top-down) approaches offered by the statistical physics of self-organisation in sentient systems.

The aim of this paper is twofold: first, to introduce a neural field model motivated by a well-known neural mass model; second, to show how one can estimate model parameters pertaining to spatial (anatomical) properties of neuronal sources based on EEG or LFP spectra using Bayesian inference. Specifically, we consider neural field models of cortical activity as generative models in the context of dynamic causal modeling (DCM). This paper considers the simplest case of a single cortical source modeled by the spatiotemporal dynamics of hidden neuronal states on a bounded cortical surface or manifold. We build this model using multiple layers, corresponding to cortical lamina in the real cortical manifold. These layers correspond to the populations considered in classical (Jansen and Rit) neural mass models. This allows us to formulate a neural field model that can be reduced to a neural mass model using appropriate constraints on its spatial parameters. In turn, this enables one to compare and contrast the predicted responses from equivalent neural field and mass models respectively. We pursue this using empirical LFP data from a single electrode to show that the parameters controlling the spatial dynamics of cortical activity can be recovered, using DCM, even in the absence of explicit spatial information in observed data.

Despite their groundbreaking performance, autonomous agents can misbehave when training and environmental conditions become inconsistent, with minor mismatches leading to undesirable behaviors or even catastrophic failures. Robustness towards these training-environment ambiguities is a core requirement for intelligent agents and its fulfillment is a long-standing challenge towards their real-world deployments. Here, we introduce a Distributionally Robust Free Energy model (DR-FREE) that instills this core property by design. Combining a robust extension of the free energy principle with a resolution engine, DR-FREE wires robustness into the agent decision-making mechanisms. Across benchmark experiments, DR-FREE enables the agents to complete the task even when, in contrast, state-of-the-art models fail. This milestone may inspire both deployments in multi-agent settings and, at a perhaps deeper level, the quest for an explanation of how natural agents - with little or no training - survive in capricious environments.