The role of inference in musculoskeletal pain: an (En)Active Inference Masterclass for clinical reasoning, communication and person-centred care

Esteves, Jorge E, Rossettini, Giacomo, Cerritelli, Francesco, Cook, Chad, Cormack, Ben, Newell, Dave, Palese, Alvisa, Friston, Karl and Thacker, Mick (2026) The role of inference in musculoskeletal pain: an (En)Active Inference Masterclass for clinical reasoning, communication and person-centred care. Musculoskeletal Science and Practice, 86. ISSN 2468-7812

Text Newell, D MSP full article.pdf - Accepted Version Restricted to Repository staff only until 17 October 2027. Available under License Creative Commons Attribution Non-commercial No Derivatives. Download (928kB)

Abstract

Abstract Background: Persistent musculoskeletal (MSK) pain cannot be fully explained by models that focus only on damaged or irritated tissues. At the same time, the biopsychosocial model, meant to integrate biological, psychological, and social factors, has often been used as a simple checklist rather than as a unified explanation of how these factors interact to produce pain. Active Inference frames perception, action, attention, emotion, and agency as processes of prediction and regulation. Read alongside enactive, ecological, and embodied predictive-processing–neuro-endocrine-immune (EPP-NEI) accounts, it offers a clinically grounded way to understand MSK pain as an embodied, relational, and inferential process. Purpose: This Masterclass translates Active Inference into MSK clinical reasoning, communication, and person-centred care, showing how kinesiophobia, hypervigilance, contextual effects, and persistent pain — after tissue healing — can be read through a single inferential lens. Approach: We synthesise theoretical, computational, and empirical work from Active Inference, enactivism, pain science, and rehabilitation. Key constructs, including generative models, prediction error, precision, allostasis, and Markov blankets, are introduced without mathematics and used pragmatically, while noting where inferential and enactive traditions converge and where tensions remain. Clinical implications: We propose five clinical levers: therapeutic alliance, manual therapy and touch, language, exercise, and context. Each is a distinct entry point into a distributed inferential process, acting by shifting priors, recalibrating precision, or generating prediction errors. These levers are mapped onto a heuristic ten-level neuraxis, L0–L9, and illustrated through three clinical vignettes. Conclusion: Active Inference does not replace the biopsychosocial model; it offers the mechanistic architecture the model has often lacked. Its value lies not in new techniques, but in organising evidence-based care and generating testable clinical predictions.

Item Type:	Article
Schools:	Centre for PAIn Research
Depositing User:	Bridget Roberts
Date Deposited:	02 Nov 2026 11:04
Last Modified:	02 Nov 2026 11:04
URI:	https://hsu.repository.guildhe.ac.uk/id/eprint/683

Actions (login required)

Edit Item