James M. Kilner

Predictive coding: an account of the mirror neuron system

Is it possible to understand the intentions of other people by simply observing their actions? Many believe that this ability is made possible by the brain’s mirror neuron system through its direct link between action and observation. However, precisely how intentions can be inferred through action observation has provoked much debate. Here we suggest that the function of the mirror system can be understood within a predictive coding framework that appeals to the statistical approach known as empirical Bayes. Within this scheme the most likely cause of an observed action can be inferred by minimizing the prediction error at all levels of the cortical hierarchy that are engaged during action observation. This account identifies a precise role for the mirror system in our ability to infer intentions from actions and provides the outline of the underlying computational mechanisms.

The mismatch negativity: A review of underlying mechanisms

The mismatch negativity (MMN) is a brain response to violations of a rule, established by a sequence of sensory stimuli (typically in the auditory domain) [Näätänen R. Attention and brain function. Hillsdale, NJ: Lawrence Erlbaum; 1992]. The MMN reflects the brain’s ability to perform automatic comparisons between consecutive stimuli and provides an electrophysiological index of sensory learning and perceptual accuracy. Although the MMN has been studied extensively, the neurophysiological mechanisms underlying the MMN are not well understood. Several hypotheses have been put forward to explain the generation of the MMN; amongst these accounts, the “adaptation hypothesis” and the “model adjustment hypothesis” have received the most attention. This paper presents a review of studies that focus on neuronal mechanisms underlying the MMN generation, discusses the two major explanatory hypotheses, and proposes predictive coding as a general framework that attempts to unify both.

Dynamic causal modeling of evoked responses in EEG and MEG.

Neuronally plausible, generative or forward models are essential for understanding how event-related fields (ERFs) and potentials (ERPs) are generated. In this paper, we present a new approach to modeling event-related responses measured with EEG or MEG. This approach uses a biologically informed model to make inferences about the underlying neuronal networks generating responses. The approach can be regarded as a neurobiologically constrained source reconstruction scheme, in which the parameters of the reconstruction have an explicit neuronal interpretation. Specifically, these parameters encode, among other things, the coupling among sources and how that coupling depends upon stimulus attributes or experimental context. The basic idea is to supplement conventional electromagnetic forward models, of how sources are expressed in measurement space, with a model of how source activity is generated by neuronal dynamics. A single inversion of this extended forward model enables inference about both the spatial deployment of sources and the underlying neuronal architecture generating them. Critically, this inference covers long-range connections among well-defined neuronal subpopulations. In a previous paper, we simulated ERPs using a hierarchical neural-mass model that embodied bottom-up, top-down and lateral connections among remote regions. In this paper, we describe a Bayesian procedure to estimate the parameters of this model using empirical data. We demonstrate this procedure by characterizing the role of changes in cortico-cortical coupling, in the genesis of ERPs. In the first experiment, ERPs recorded during the perception of faces and houses were modeled as distinct cortical sources in the ventral visual pathway. Category-selectivity, as indexed by the face-selective N170, could be explained by category-specific differences in forward connections from sensory to higher areas in the ventral stream. We were able to quantify and make inferences about these effects using conditional estimates of connectivity. This allowed us to identify where, in the processing stream, category-selectivity emerged. In the second experiment, we used an auditory oddball paradigm to show that the mismatch negativity can be explained by changes in connectivity. Specifically, using Bayesian model selection, we assessed changes in backward connections, above and beyond changes in forward connections. In accord with theoretical predictions, there was strong evidence for learning-related changes in both forward and backward coupling. These examples show that category- or context-specific coupling among cortical regions can be assessed explicitly, within a mechanistic, biologically motivated inference framework.

Evidence of mirror neurons in human inferior frontal gyrus

There is much current debate about the existence of mirror neurons in humans. To identify mirror neurons in the inferior frontal gyrus (IFG) of humans, we used a repetition suppression paradigm while measuring neural activity with functional magnetic resonance imaging. Subjects either executed or observed a series of actions. Here we show that in the IFG, responses were suppressed both when an executed action was followed by the same rather than a different observed action and when an observed action was followed by the same rather than a different executed action. This pattern of responses is consistent with that predicted by mirror neurons and is evidence of mirror neurons in the human IFG.

A free energy principle for the brain.

By formulating Helmholtzs ideas about perception, in terms of modern-day theories, one arrives at a model of perceptual inference and learning that can explain a remarkable range of neurobiological facts: using constructs from statistical physics, the problems of inferring the causes of sensory input and learning the causal structure of their generation can be resolved using exactly the same principles. Furthermore, inference and learning can proceed in a biologically plausible fashion. The ensuing scheme rests on Empirical Bayes and hierarchical models of how sensory input is caused. The use of hierarchical models enables the brain to construct prior expectations in a dynamic and context-sensitive fashion. This scheme provides a principled way to understand many aspects of cortical organisation and responses. In this paper, we show these perceptual processes are just one aspect of emergent behaviours of systems that conform to a free energy principle. The free energy considered here measures the difference between the probability distribution of environmental quantities that act on the system and an arbitrary distribution encoded by its configuration. The system can minimise free energy by changing its configuration to affect the way it samples the environment or change the distribution it encodes. These changes correspond to action and perception respectively and lead to an adaptive exchange with the environment that is characteristic of biological systems. This treatment assumes that the systems state and structure encode an implicit and probabilistic model of the environment. We will look at the models entailed by the brain and how minimisation of its free energy can explain its dynamics and structure.

The role of synchrony and oscillations in the motor output

Abstract There is currently much interest in the synchronisation of neural discharge and the potential role it may play in information coding within the nervous system. We describe some recent results from investigations of synchronisation within the motor system. Local field potentials (LFPs) and identified pyramidal tract neurones (PTNs) were recorded from the primary motor cortex of monkeys trained to perform a precision grip task. The LFPs showed bursts of oscillatory activity at 20–30 Hz, which were coherent with the rectified electromyographs (EMG) of contralateral hand and forearm muscles. This oscillatory synchronisation showed a highly specific task dependence, being present only during the part of the task when the animal maintained a steady grip and not during the movement phases before or after it. PTNs were phase-locked to LFP oscillations, implying that at least part of the coherence between cortical activity and EMG was mediated by corticospinal fibres. The phase locking of the PTNs to LFP oscillations produced task-dependent oscillatory synchronisation between PTN pairs, as assessed by the single-unit cross-correlation histogram. Recordings were also made from normal human subjects performing a precision grip similar to that used in the monkey recordings. Pairs of EMGs recorded from intrinsic hand and forearm muscles showed 20–30 Hz coherence, which modulated during task performance, being present only during periods of steady contraction. We suggest that these changes in EMG-EMG synchronisation reflect changing levels of synchronous drive from the corticospinal system. The generation of oscillations in the cortex is discussed in the light of results from a model of local cortical circuits. Other modelling work has shown that synchrony in the corticospinal inputs could act to recruit motoneurones more efficiently, producing more output force from a muscle than asynchronous inputs firing at the same mean rate. A speculative hypothesis is presented on the role of synchronous oscillations in the motor system, which is consistent with experimental observations to date.

EEG and MEG Data Analysis in SPM8

SPM is a free and open source software written in MATLAB (The MathWorks, Inc.). In addition to standard M/EEG preprocessing, we presently offer three main analysis tools: (i) statistical analysis of scalp-maps, time-frequency images, and volumetric 3D source reconstruction images based on the general linear model, with correction for multiple comparisons using random field theory; (ii) Bayesian M/EEG source reconstruction, including support for group studies, simultaneous EEG and MEG, and fMRI priors; (iii) dynamic causal modelling (DCM), an approach combining neural modelling with data analysis for which there are several variants dealing with evoked responses, steady state responses (power spectra and cross-spectra), induced responses, and phase coupling. SPM8 is integrated with the FieldTrip toolbox , making it possible for users to combine a variety of standard analysis methods with new schemes implemented in SPM and build custom analysis tools using powerful graphical user interface (GUI) and batching tools.

Brain systems for assessing facial attractiveness

Attractiveness is a facial attribute that shapes human affiliative behaviours. In a previous study we reported a linear response to facial attractiveness in orbitofrontal cortex (OFC), a region involved in reward processing. There are strong theoretical grounds for the hypothesis that coding stimulus reward value also involves the amygdala. The aim of the present investigation is to address whether the amygdala is also sensitive to reward value in faces, indexed as facial attractiveness. We hypothesized that contrary to the linear effects reported previously in OFC, the amygdala would show a non-linear effect of attractiveness by responding to both high and low attractive faces relative to middle attractive faces. Such a non-linear response would explain previous failures to report an amygdala response to attractiveness. Human subjects underwent fMRI while they were presented with faces that varied in facial attractiveness where the task was either to rate faces for facial attractiveness or for age. Consistent with our hypothesis, right amygdala showed a predicted non-linear response profile with greater responses to highly attractive and unattractive faces compared to middle-ranked faces, independent of task. Distinct patterns of activity were seen across different regions of OFC, with some sectors showing linear effects of attractiveness, others exhibiting a non-linear response profile and still others demonstrating activation only during age judgments. Significant effects were also seen in medial prefrontal and paracingulate cortices, posterior OFC, insula, and superior temporal sulcus during explicit attractiveness judgments. The non-linear response profile of the amygdala is consistent with a role in sensing the value of social stimuli, a function that may also involve specific sectors of the OFC.

Action and behavior: a free-energy formulation

We have previously tried to explain perceptual inference and learning under a free-energy principle that pursues Helmholtz’s agenda to understand the brain in terms of energy minimization. It is fairly easy to show that making inferences about the causes of sensory data can be cast as the minimization of a free-energy bound on the likelihood of sensory inputs, given an internal model of how they were caused. In this article, we consider what would happen if the data themselves were sampled to minimize this bound. It transpires that the ensuing active sampling or inference is mandated by ergodic arguments based on the very existence of adaptive agents. Furthermore, it accounts for many aspects of motor behavior; from retinal stabilization to goal-seeking. In particular, it suggests that motor control can be understood as fulfilling prior expectations about proprioceptive sensations. This formulation can explain why adaptive behavior emerges in biological agents and suggests a simple alternative to optimal control theory. We illustrate these points using simulations of oculomotor control and then apply to same principles to cued and goal-directed movements. In short, the free-energy formulation may provide an alternative perspective on the motor control that places it in an intimate relationship with perception.

Action understanding and active inference

We have suggested that the mirror-neuron system might be usefully understood as implementing Bayes-optimal perception of actions emitted by oneself or others. To substantiate this claim, we present neuronal simulations that show the same representations can prescribe motor behavior and encode motor intentions during action–observation. These simulations are based on the free-energy formulation of active inference, which is formally related to predictive coding. In this scheme, (generalised) states of the world are represented as trajectories. When these states include motor trajectories they implicitly entail intentions (future motor states). Optimizing the representation of these intentions enables predictive coding in a prospective sense. Crucially, the same generative models used to make predictions can be deployed to predict the actions of self or others by simply changing the bias or precision (i.e. attention) afforded to proprioceptive signals. We illustrate these points using simulations of handwriting to illustrate neuronally plausible generation and recognition of itinerant (wandering) motor trajectories. We then use the same simulations to produce synthetic electrophysiological responses to violations of intentional expectations. Our results affirm that a Bayes-optimal approach provides a principled framework, which accommodates current thinking about the mirror-neuron system. Furthermore, it endorses the general formulation of action as active inference.