The human brain has a remarkable ability to learn and update its beliefs about the world. Here, we investigate how thermosensory learning shapes our subjective experience of innocuous temperature and the misperception of pain in response to harmless thermal stimuli. Through computational modeling, we demonstrate that the brain uses a probabilistic predictive coding scheme to update beliefs about temperature changes based on their uncertainty. We find that these expectations directly modulate the perception of pain in the thermal grill illusion as well as the constituent innocuous cold and warm stimulus. Quantitative microstructural brain imaging revealed that the myeloarchitecture and iron concentration of the somatosensory cortex, the posterior insula and the amygdala reflect inter-individual variability in computational parameters related to learning and the degree to which uncertainty modulates illusory pain perception. Our findings offer a new framework to explain how the brain infers pain from innocuous thermal inputs. Our model has important implications for understanding the etiology of thermosensory symptoms in chronic pain conditions.