Symptom perception, placebo effects, and the Bayesian brain

This topical review addresses a fundamental challenge to the traditional biomedical model: how to explain symptoms that occur without clear physical pathology and symptom relief from placebo treatments. The authors propose that the Bayesian brain model—which conceptualizes perception as a predictive process integrating sensory inputs, prior experience, and contextual cues—provides a compelling framework for understanding these phenomena. Rather than treating the brain as a passive receiver of signals, this approach views perception as the brain's continuously updated hypothesis about bodily and environmental states, governed by Bayes' rule and prediction error minimization.

The Bayesian framework suggests that symptom perception is not a direct readout of physiological dysfunction but rather an inference the brain makes about whether the body has deviated from a "healthy body condition" baseline. For acute conditions with clear pathology, sensory signals have high precision and align well with symptoms. However, in chronic pain and medically unexplained symptoms—where sensory signals are ambiguous or shifting—the brain's high-precision predictions about pain can override weak sensory evidence, explaining the poor correlation between objective pathology and subjective experience. Similarly, symptom relief is understood not as direct restoration of function but as a revision of the brain's predictions that the body is returning to health, enhanced by external evidence from medical rituals, clinician support, and contextual cues.

The authors demonstrate that placebo effects operate through the same inferential processes as active treatments, with both strengthening predictions of health through different pathways. Evidence from "open-hidden" studies shows that patients receiving treatments openly experience greater relief than those receiving identical treatments covertly, supporting the role of prediction. Importantly, the framework transcends the artificial distinction between "real" and "imaginary" symptoms, revealing that all symptoms arise from the same inferential process that is never strictly reducible to physiology alone.

This predictive processing approach has significant implications for clinical practice. It suggests that truly patient-focused medicine must attend to the predictive processes underlying symptom perception and leverage the therapeutic context—including clinician communication, perceived treatment value, and clinical ritual—as legitimate mechanisms for promoting health. Understanding that placebo and nocebo effects are normal features of nervous system function rather than anomalies validates the therapeutic value of clinical presence and contextual factors, while explaining why such approaches are most effective for symptoms loosely coupled to organic pathology.