Canonical neural networks perform active inference

This work considers a class of canonical neural networks comprising rate coding models, wherein neural activity and plasticity minimise a common cost function—and plasticity is modulated with a certain delay. We show that such neural networks implicitly perform active inference and learning to minimise the risk associated with future outcomes. Mathematical analyses demonstrate that this biological optimisation can be cast as maximisation of model evidence, or equivalently minimisation of variational free energy, under the well-known form of a partially observed Markov decision process model. This equivalence indicates that the delayed modulation of Hebbian plasticity—accompanied with adaptation of firing thresholds—is a sufficient neuronal substrate to attain Bayes optimal inference and control. We corroborated this proposition using numerical analyses of maze tasks. This theory offers a universal characterisation of canonical neural networks in terms of Bayesian belief updating and provides insight into the neuronal mechanisms underlying planning and adaptive behavioural control.

Neuronal mechanisms of ginseng on antiobesity effects: implication of its synergistic benefits with physical exercise

Obesity is a chronic disease of increasing prevalence in most countries, which leads to substantial increase in morbidity, and mortality in association with diabetes, hyperlipidaemia, hypertension, and other cardiovascular diseases. Many factors have been attributed to an epidemic of obesity including sedentary lifestyle, high-fat diets (HFD), and consumption of large amount of modern fast foods. Panax ginseng C. A. Meyer (PG) has several pharmacological and physiological effects. In particular, PG and saponin fractions from PG show a variety of efficacies such as antifatigue, hyperlipidemia, hypertension and noninsulin-dependent diabetes mellitus and obesity. We have revealed that ginseng and ginsenosides can decrease food intake energy expenditure by stimulating appetite regulatory hormones and can reduce energy intake. Exercise/physical activity is well known as modality for treating the disease of overweight and obesity. It is suggested that natural products and their combinations with exercise may produce a synergistic activity that increases their bioavailability and action on multiple molecular targets, offering advantages over chemical treatments. This review is aimed at evaluating the antiobesity efficacy of ginseng and ginsenosides and delineating the mechanisms by which they function. Finally, we review information regarding interactions between ginseng and physical exercise in protecting against weight gain and obesity.

Enhanced learning and sensory salience in a cerebellar mouse autism model

Among the impairments manifested by autism spectrum disorder (ASD) are sometimes islands of enhanced function. Although neuronal mechanisms for enhanced functions in ASD are unknown, the cerebellum is a major site of developmental alteration, and early-life perturbation to it leads to ASD with higher likelihood than any other brain region. Here we report that a cerebellum-specific transgenic mouse model of ASD shows faster learning on a sensory evidence-accumulation task. In addition, transgenic mice showed enhanced sensitivity to touch and auditory cues, and prolonged electrophysiological responses in Purkinje-cell complex spikes and associative neocortical regions. These findings were replicated by pairing cues with optogenetic stimulation of Purkinje cells. Computational latent-state analysis of behavior revealed that both groups of mice with cerebellar perturbations exhibited enhanced focus on current rather than past information, consistent with a role for the cerebellum in retaining information in memory. We conclude that cerebellar perturbation can activate neocortex via complex spike activity and reduce reliance on prior experience, consistent with a weak-central-coherence account in which ASD traits arise from enhanced detail-oriented processing. This recasts ASD not so much as a disorder but as a variation that, in particular niches, can be adaptive.

Situational Understanding in the Human and the Machine

The Air Force research programs envision developing AI technologies that will ensure battlespace dominance, by radical increases in the speed of battlespace understanding and decision-making. In the last half century, advances in AI have been concentrated in the area of machine learning. Recent experimental findings and insights in systems neuroscience, the biophysics of cognition, and other disciplines provide converging results that set the stage for technologies of machine understanding and machine-augmented Situational Understanding. This paper will review some of the key ideas and results in the literature, and outline new suggestions. We define situational understanding and the distinctions between understanding and awareness, consider examples of how understanding—or lack of it—manifest in performance, and review hypotheses concerning the underlying neuronal mechanisms. Suggestions for further R&D are motivated by these hypotheses and are centered on the notions of Active Inference and Virtual Associative Networks.

From decision to action: Detailed modelling of frog tadpoles reveals neuronal mechanisms of decision-making and reproduces unpredictable swimming movements in response to sensory signals

How does the brain process sensory stimuli, and decide whether to initiate locomotor behaviour? To investigate this question we develop two whole body computer models of a tadpole. The “Central Nervous System” (CNS) model uses evidence from whole-cell recording to define 2000 neurons in 12 classes to study how sensory signals from the skin initiate and stop swimming. In response to skin stimulation, it generates realistic sensory pathway spiking and shows how hindbrain sensory memory populations on each side can compete to initiate reticulospinal neuron firing and start swimming. The 3-D “Virtual Tadpole” (VT) biomechanical model with realistic muscle innervation, body flexion, body-water interaction, and movement is then used to evaluate if motor nerve outputs from the CNS model can produce swimming-like movements in a volume of “water”. We find that the whole tadpole VT model generates reliable and realistic swimming. Combining these two models opens new perspectives for experiments.

Neuronal congruency effects in macaque frontal cortex

The interplay between task-relevant and task-irrelevant stimulus features induces conflicts which impair human behavioral performance in many perceptual and cognitive tasks, a.k.a. a behavioral congruency effect. The neuronal mechanisms underlying behavioral congruency effects, however, are poorly understood. We recorded single unit activity in monkey frontal cortex using a novel task-switching paradigm and discovered a neuronal congruency effect that is carried by task-relevant and -irrelevant neurons. The former neurons provide more signal, the latter less noise in congruent compared to incongruent conditions. Their relative activity levels determine the neuronal congruency effect and behavioral performance. Although these neuronal congruency signals are sensitive to selective attention, they cannot be entirely explained by selective attention as gauged by response time. We propose that such neuronal congruency effects can explain behavioral congruency effects in general, as well as previous fMRI and EEG results in various conflict paradigms.

Neuronal activity reorganization in motor cortex for successful locomotion after a lesion in the ventrolateral thalamus

Thalamic stroke leads to ataxia if the cerebellum-receiving ventrolateral thalamus (VL) is affected. The compensation mechanisms for this deficit are not well understood, particularly the roles that single neurons and specific neuronal subpopulations outside the thalamus play in recovery. The goal of this study was to clarify neuronal mechanisms of the motor cortex involved in mitigation of ataxia during locomotion when part of the VL is inactivated or lesioned. In freely ambulating cats, we recorded the activity of neurons in layer V of the motor cortex as the cats walked on a flat surface and horizontally placed ladder. We first reversibly inactivated approximately 10% of the VL unilaterally using glutamatergic transmission antagonist CNQX and analyzed how the activity of motor cortex reorganized to support successful locomotion. We next lesioned 50-75% of the VL bilaterally using kainic acid and analyzed how the activity of motor cortex reorganized when locomotion recovered. When a small part of the VL was inactivated, the discharge rates of motor cortex neurons decreased, but otherwise the activity was near normal, and the cats walked fairly well. Individual neurons retained their ability to respond to the demand for accuracy during ladder locomotion; however, most changed their response. When the VL was lesioned, the cat walked normally on the flat surface but was ataxic on the ladder for several days post-lesion. When ladder locomotion normalized, neuronal discharge rates on the ladder were normal, and the shoulder-related group was preferentially active during the stride's swing phase.

Feasibility and Challenges of Performing Magnetoencephalography Experiments in Children With Arthrogryposis Multiplex Congenita

Arthrogryposis multiplex congenita (AMC) has recently drawn substantial attention from researchers and clinicians. New effective surgical and physiotherapeutic methods have been developed to improve the quality of life of patients with AMC. While it is clear that all these interventions should strongly rely on the plastic reorganization of the central nervous system, almost no studies have investigated this topic. The present study demonstrates the feasibility of using magnetoencephalography (MEG) to investigate brain activity in young AMC patients. We also outlined the general challenges and limitations of electrophysiological investigations on patients with arthrogryposis. We conducted MEG recordings using a 306-channel Elekta Neuromag VectorView system during a cued motor task performance in four patients with arthrogryposis, five normally developed children, and five control adults. Following the voice command of the experimenter, each subject was asked to bring their hand toward their mouth to imitate the self-feeding process. Two patients had latissimus dorsi transferred to the biceps brachii position, one patient had a pectoralis major transferred to the biceps brachii position, and one patient had no elbow flexion restoration surgery before the MEG investigation. Three patients who had undergone autotransplantation prior to the MEG investigation demonstrated activation in the sensorimotor area contralateral to the elbow flexion movement similar to the healthy controls. One patient who was recorded before the surgery demonstrated subjectively weak distributed bilateral activation during both left and right elbow flexion. Visual inspection of MEG data suggested that neural activity associated with motor performance was less pronounced and more widely distributed across the cortical areas of patients than of healthy control subjects. In general, our results could serve as a proof of principle in terms of the application of MEG in studies on cortical activity in patients with AMC. Reported trends might be consistent with the idea that prolonged motor deficits are associated with more difficult neuronal recruitment and the spatial heterogeneity of neuronal sources, most likely reflecting compensatory neuronal mechanisms. On the practical side, MEG could be a valuable technique for investigating the neurodynamics of patients with AMC as a function of postoperative abilitation.

Neuronal Circuits for Social Decision-Making and Their Clinical Implications

Social living facilitates individual access to rewards, cognitive resources, and objects that would not be otherwise accessible. There are, however, some drawbacks to social living, particularly when competing for scarce resources. Furthermore, variability in our ability to make social decisions can be associated with neuropsychiatric disorders. The neuronal mechanisms underlying social decision-making are beginning to be understood. The momentum to study this phenomenon has been partially carried over by the study of economic decision-making. Yet, because of the similarities between these different types of decision-making, it is unclear what is a social decision. Here, we propose a definition of social decision-making as choices taken in a context where one or more conspecifics are involved in the decision or the consequences of it. Social decisions can be conceptualized as complex economic decisions since they are based on the subjective preferences between different goods. During social decisions, individuals choose based on their internal value estimate of the different alternatives. These are complex decisions given that conspecifics beliefs or actions could modify the subject’s internal valuations at every choice. Here, we first review recent developments in our collective understanding of the neuronal mechanisms and circuits of social decision-making in primates. We then review literature characterizing populations with neuropsychiatric disorders showing deficits in social decision-making and the underlying neuronal circuitries associated with these deficits.

Energy Homeostasis Principle: A Naturalistic Approach to Explain the Emergence of Behavior

Explaining the emergence of behavior and understanding on the basis of neuronal mechanisms is still elusive. One renowned proposal is the Free Energy Principle (FEP), which uses an information-theoretic framework derived from thermodynamic considerations to describe how behavior and understanding would emerge. FEP starts from a whole organism approach, based on mental states and phenomena, mapping them into the neuronal substrate. An alternative approach, the Energy Homeostasis Principle (EHP), initiates a similar explanatory effort, but starting from single neuron phenomena and building up to the whole organism’s behavior and understanding. In this work, we develop the EHP as an alternative but complementary vision to FEP and try to explain how behavior and understanding would emerge from the local requirements of the neurons. Based on EHP and a strict naturalist approach that sees living beings as physical and deterministic systems, we explain scenarios where learning would emerge without the need for volition or goals. Given these starting points, we state several considerations of how we see the nervous system, particularly the role of function, purpose, and the conception of goal-oriented behaviors. We problematize these conceptions, giving an alternative teleology-free framework in which behavior and, ultimately, understanding would still emerge. We reinterpret neural processing explaining basic learning situations up to simple anticipatory behavior. Finally, we end the work with an evolutionary perspective of how this non-goal-oriented behavior appears. We acknowledge that in the current form of our proposal, we are still far from explaining the emergence of understanding. Still, we set the ground for an alternative neuron-based framework to ultimately explain understanding.