Balancing Active Inference and Active Learning with Deep Variational Predictive Coding for EEG

2020 ◽  
Author(s):  
Andre Ofner ◽  
Sebastian Stober
Keyword(s):  
Active Learning ◽  
Predictive Coding ◽  
Active Inference

scholarly journals Implementing Predictive Processing and Active Inference: Preliminary Steps and Results

2019 ◽  
Author(s):  
Beren Millidge
Keyword(s):  
Predictive Coding ◽  
Predictive Processing ◽  
Active Inference ◽  
Inference Models ◽  
Perceptual Model ◽  
Challenging Tasks ◽  
Generalised Coordinates

Initial and preliminary implementations of predictive processing and active inference models are presented. These include the baseline hierarchical predictive coding models of (Friston 2003, 2005), and dynamical predictive coding models using generalised coordinates (Friston 2008, 2010, Buckley 2017). Additionally, we re-implement and experiment with the active inference thermostat presented in (Buckley 2017) and also implement an active inference agent with a hierarchical predictive coding perceptual model on the more challenging cart-pole task from OpanAI gym. We discuss the initial performance, capabilities, and limitations of these models in their preliminary stages and consider how they might be further scaled up to tackle more challenging tasks.

scholarly journals ‘Through others we become ourselves’: The dialectics of predictive coding and active inference

2019 ◽  
Author(s):  
Dimitris Bolis ◽  
Leonhard Schilbach
Keyword(s):  
Free Energy ◽  
Social Interaction ◽  
Predictive Coding ◽  
Real Life ◽  
Other Minds ◽  
Active Inference ◽  
Energy Principle ◽  
Free Energy Principle ◽  
Multiscale Dynamics ◽  
Cultural Processes

Thinking Through Other Minds (TTOM) creatively situates the free energy principle within real-life cultural processes, thereby enriching both sociocultural theories and Bayesian accounts of cognition. Here, shifting the attention from thinking to becoming, we suggest complementing such an account by focusing on the empirical, computational and conceptual investigation of the multiscale dynamics of social interaction.

scholarly journals The cutaneous borders of interoception: Active and social inference of pain and pleasure on the skin

2018 ◽  
Author(s):  
Mariana von Mohr ◽  
Aikaterini Fotopoulou
Keyword(s):  
Social Factors ◽  
Predictive Coding ◽  
Active Inference ◽  
Top Down ◽  
Bottom Up ◽  
Social Inference ◽  
Brain Levels ◽  
The Brain

Pain and pleasant touch have been recently classified as interoceptive modalities. This reclassification lies at the heart of long-standing debates questioning whether these modalities should be defined as sensations on their basis of neurophysiological specificity at the periphery or as homeostatic emotions on the basis of top-down convergence and modulation at the spinal and brain levels. Here, we outline the literature on the peripheral and central neurophysiology of pain and pleasant touch. We next recast this literature within a recent Bayesian predictive coding framework, namely active inference. This recasting puts forward a unifying model of bottom-up and top-down determinants of pain and pleasant touch and the role of social factors in modulating the salience of peripheral signals reaching the brain.

scholarly journals Combining Active Inference and Hierarchical Predictive Coding: A Tutorial Introduction and Case Study

2019 ◽  
Author(s):  
Beren Millidge
Keyword(s):  
Free Energy ◽  
Reinforcement Learning ◽  
Predictive Coding ◽  
Action Selection ◽  
Proof Of Concept ◽  
Active Inference ◽  
Energy Principle ◽  
Free Energy Principle ◽  
Perceptual Model

This paper combines the active inference formulation of action (Friston, 2009) with hierarchical predictive coding models (Friston, 2003) to provide a proof-of-concept implementation of an active inference agent able to solve a common reinforcement learning baseline -- the cart-pole environment in OpenAI gym. It demonstrates empirically that predictive coding and active inference approaches can be successfully scaled up to tasks more challenging than the mountain car (Friston 2009, 2012). We show that hierarchical predictive coding models can be learned from scratch during the task, and can successfully drive action selection via active inference. To our knowledge, it is the first implemented active inference agent to combine active inference with a hierarchical predictive coding perceptual model. We also provide a tutorial walk-through of the free-energy principle, hierarchical predictive coding, and active inference, including an in-depth derivation of our agent.

scholarly journals Neural Underpinning of Japanese Particle Processing in Non-native Speakers

2021 ◽  
Author(s):  
Chise Kasai ◽  
Motofumi Sumiya ◽  
Takahiko Koike ◽  
Takaaki Yoshimoto ◽  
Hideki Maki ◽  
...  
Keyword(s):  
Working Memory ◽  
Native Speakers ◽  
Inferior Frontal Gyrus ◽  
Predictive Coding ◽  
Primary Source ◽  
Verbal Working Memory ◽  
Forward Model ◽  
Neural Representation ◽  
Middle Temporal Gyrus ◽  
Active Inference

Abstract Grammar acquisition by non-native learners (L2) is typically less successful and may produce fundamentally different grammatical systems than that by native speakers (L1). The neural representation of grammatical processing between L1 and L2 speakers remains controversial. We hypothesized that working memory is the primary source of L1/L2 differences, and operationalized working memory is an active inference within the predictive coding account, which models grammatical processes as higher-level neuronal representations of cortical hierarchies, generating predictions (forward model) of lower-level representations. A functional MRI study was conducted with L1 Japanese speakers and highly proficient Japanese learners requiring oral production of grammatically correct Japanese particles. Selecting proper particles requires forward model-dependent active inference as their functions are highly context-dependent. As a control, participants read out a visually designated mora indicated by underlining. Particle selection by L1/L2 groups commonly activated the bilateral inferior frontal gyrus/insula, pre-supplementary motor area, left caudate, middle temporal gyrus, and right cerebellum, which constituted the core linguistic production system. In contrast, the left inferior frontal sulcus, known as the neural substrate of verbal working memory, showed more prominent activation in L2 than in L1. Thus, the active inference process causes L1/L2 differences even in highly proficient L2 learners.

scholarly journals Goal-Directed Planning for Habituated Agents by Active Inference Using a Variational Recurrent Neural Network

Entropy ◽  
2020 ◽  
Vol 22 (5) ◽  
pp. 564
Author(s):  
Takazumi Matsumoto ◽  
Jun Tani
Keyword(s):  
Lower Bound ◽  
Latent Variables ◽  
Predictive Coding ◽  
Forward Model ◽  
Training Data ◽  
Active Inference ◽  
Model Framework ◽  
Model Learning ◽  
Proposed Model ◽  
Sensory Motor

It is crucial to ask how agents can achieve goals by generating action plans using only partial models of the world acquired through habituated sensory-motor experiences. Although many existing robotics studies use a forward model framework, there are generalization issues with high degrees of freedom. The current study shows that the predictive coding (PC) and active inference (AIF) frameworks, which employ a generative model, can develop better generalization by learning a prior distribution in a low dimensional latent state space representing probabilistic structures extracted from well habituated sensory-motor trajectories. In our proposed model, learning is carried out by inferring optimal latent variables as well as synaptic weights for maximizing the evidence lower bound, while goal-directed planning is accomplished by inferring latent variables for maximizing the estimated lower bound. Our proposed model was evaluated with both simple and complex robotic tasks in simulation, which demonstrated sufficient generalization in learning with limited training data by setting an intermediate value for a regularization coefficient. Furthermore, comparative simulation results show that the proposed model outperforms a conventional forward model in goal-directed planning, due to the learned prior confining the search of motor plans within the range of habituated trajectories.

scholarly journals Interoceptive accuracy moderates the response to a glucose load: a test of the predictive coding framework

2019 ◽  
Vol 286 (1898) ◽  
pp. 20190244 ◽  
Author(s):  
Hayley A. Young ◽  
Chantelle M. Gaylor ◽  
Danielle de Kerckhove ◽  
Heather Watkins ◽  
David Benton
Keyword(s):  
Heart Rate ◽  
Blood Glucose ◽  
Emotional Health ◽  
Predictive Coding ◽  
The Body ◽  
Active Inference ◽  
Glucose Load ◽  
Postprandial Period ◽  
Interoceptive Accuracy

Recently, interoception and homeostasis have been described in terms of predictive coding and active inference. Afferent signals update prior predictions about the state of the body, and stimulate the autonomic mediation of homeostasis. Performance on tests of interoceptive accuracy (IAc) may indicate an individual's ability to assign precision to interoceptive signals, thus determining the relative influence of ascending signals and the descending prior predictions. Accordingly, individuals with high IAc should be better able to regulate during the postprandial period. One hundred females were allocated to consume glucose, an artificially sweetened drink, water or no drink. Before, and 30 min after a drink, IAc, heart rate (HR) and blood glucose (BG) were measured, and participants rated their hunger, thirst and mood. A higher IAc was related to lower BG levels, a decline in anxiety and a higher HR, after consuming glucose. A higher IAc also resulted in a larger decline in hunger if they consumed either glucose or sucralose. These data support the role of active inference in interoception and homeostasis, and suggest that the ability to attend to interoceptive signals may be critical to the maintenance of physical and emotional health.

scholarly journals Active Inference and Learning in the Cerebellum

2016 ◽  
Vol 28 (9) ◽  
pp. 1812-1839 ◽  
Author(s):  
Karl Friston ◽  
Ivan Herreros
Keyword(s):  
Eyeblink Conditioning ◽  
Predictive Coding ◽  
Trace Conditioning ◽  
Face Validity ◽  
Active Inference ◽  
Focal Lesions ◽  
Delay Conditioning ◽  
Neuronal Populations ◽  
Intrinsic Connectivity ◽  
The Face

This letter offers a computational account of Pavlovian conditioning in the cerebellum based on active inference and predictive coding. Using eyeblink conditioning as a canonical paradigm, we formulate a minimal generative model that can account for spontaneous blinking, startle responses, and (delay or trace) conditioning. We then establish the face validity of the model using simulated responses to unconditioned and conditioned stimuli to reproduce the sorts of behavior that are observed empirically. The scheme’s anatomical validity is then addressed by associating variables in the predictive coding scheme with nuclei and neuronal populations to match the (extrinsic and intrinsic) connectivity of the cerebellar (eyeblink conditioning) system. Finally, we try to establish predictive validity by reproducing selective failures of delay conditioning, trace conditioning, and extinction using (simulated and reversible) focal lesions. Although rather metaphorical, the ensuing scheme can account for a remarkable range of anatomical and neurophysiological aspects of cerebellar circuitry—and the specificity of lesion-deficit mappings that have been established experimentally. From a computational perspective, this work shows how conditioning or learning can be formulated in terms of minimizing variational free energy (or maximizing Bayesian model evidence) using exactly the same principles that underlie predictive coding in perception.

scholarly journals A tale of two densities: active inference is enactive inference

2019 ◽  
Vol 28 (4) ◽  
pp. 225-239 ◽  
Author(s):  
Maxwell JD Ramstead ◽  
Michael D Kirchhoff ◽  
Karl J Friston
Keyword(s):  
Predictive Coding ◽  
Generative Models ◽  
Predictive Processing ◽  
Active Inference ◽  
Energy Principle ◽  
Free Energy Principle ◽  
The One ◽  
And Control ◽  
The Brain ◽  
Selection Of

The aim of this article is to clarify how best to interpret some of the central constructs that underwrite the free-energy principle (FEP) – and its corollary, active inference – in theoretical neuroscience and biology: namely, the role that generative models and variational densities play in this theory. We argue that these constructs have been systematically misrepresented in the literature, because of the conflation between the FEP and active inference, on the one hand, and distinct (albeit closely related) Bayesian formulations, centred on the brain – variously known as predictive processing, predictive coding or the prediction error minimisation framework. More specifically, we examine two contrasting interpretations of these models: a structural representationalist interpretation and an enactive interpretation. We argue that the structural representationalist interpretation of generative and recognition models does not do justice to the role that these constructs play in active inference under the FEP. We propose an enactive interpretation of active inference – what might be called enactive inference. In active inference under the FEP, the generative and recognition models are best cast as realising inference and control – the self-organising, belief-guided selection of action policies – and do not have the properties ascribed by structural representationalists.

Sign in / Sign up

Export Citation Format

Share Document