Background & Aims

The ability to adapt to environmental changes and learn in the face of uncertainty is critical for generating precise and flexible responses to a wide range of stimuli. In the context of thermosensation and nociception, adaptability allows us to effectively detect temperature shifts and avert potential tissue damage, even under conditions of incomplete or ambiguous information. In this context, the thermal grill illusion (TGI) presents a striking case in which the simultaneous presentation of innocuous warm and cold stimuli can evoke illusory burning sensations. The illusion’s occurrence, which cannot be solely attributed to the physical characteristics of the stimuli, suggests that top-down expectations may shape the veridical perception of temperature and pain. An intriguing possibility is that the TGI may stem from thermosensory predictive coding, where increased uncertainty about stimulus temperatures give rise to the perception of pain.

Methods

In this study, we apply computational methods to reveal how both innocuous thermosensation and the thermal grill illusion (TGI) are shaped by precision-weighted expectations in a large cohort of 267 healthy participants. We further utilized high-resolution quantitative Magnetic Resonance imaging to identify how inter-individual variations in brain microstructure are associated with computational fingerprints of thermosensory learning. To this aim, participants completed a probabilistic thermosensory learning (PTL) task of approx. 300 trials, in which we strategically embedded simultaneous cold and warm stimuli to induce the thermal grill illusion within the learning sequence. This experimental approach offers a comprehensive analysis of the role of expectations in thermosensory learning and provides a unique opportunity to test the hypothesis that the precision of thermal expectations plays a crucial role in the perception of illusory pain.

Results

Our results demonstrate that innocuous thermosensory stimuli were perceived in a veridical manner, and the thermal grill illusion (TGI) manipulation effectively induced illusory heat and burning sensations. Furthermore we highlight that participants’ thermosensory predictions significantly influence the perceived intensity of the innocuous stimuli. Using computational modeling we show that learned trial-wise expectations and lower level uncertainties shaped innocuous thermosensory percepts, whereas the burning sensation elicited by the TGI was driven by higher level uncertainties regarding consistencies in the cue-stimulus association. Lastly we investigated the neurobiological underpinnings, such as myeloarchitecture and iron concentration, of inter-individual variability in the computational parameters using quantitative magnetic resonance imaging. Finding that these relate to key areas (e.g., insula, amygdala and brainstem) associated with pain, fear and somatosensory processing.

Conclusions

In summary, this study not only refines our understanding of human thermosensation from a Bayesian perspective but also elucidates how thermal uncertainty can underpin the mis-interpretation of harmless stimuli as painful, transforming objectively innocuous stimuli into a subjective experience of pain. Our findings align with the growing body of evidence supporting Bayesian models in pain perception and learning, extending such frameworks to the domains of thermosensation and thermo-nociceptive illusions. By showing that innocuous thermosensation and pain perception can be modulated by learned expectations, our study provides a computational framework that may be used to identify computational profiles in chronic pain conditions, where pain perception is decoupled from peripheral nociceptive inputs. Overall, our study offers a computational framework to investigate how individual differences in learning could underlie experiences of chronic pain.

References

Craig, A. D., and M. C. Bushnell. 1994. “The Thermal Grill Illusion: Unmasking the Burn of Cold Pain.” Science 265 (5169): 252–55. https://doi.org/10.1126/science.8023144.

Fardo, Francesca, Brianna Beck, Micah Allen, and Nanna Brix Finnerup. 2020. “Beyond Labeled Lines: A Population Coding Account of the Thermal Grill Illusion.” Neuroscience and Biobehavioral Reviews 108 (January): 472–79. https://doi.org/10.1016/j.neubiorev.2019.11.017.

Geuter, Stephan, Sabrina Boll, Falk Eippert, and Christian Büchel. 2017. “Functional Dissociation of Stimulus Intensity Encoding and Predictive Coding of Pain in the Insula.” Edited by Heidi Johansen-Berg. eLife 6 (May): e24770. https://doi.org/10.7554/eLife.24770.

Mathys, Christoph, Jean Daunizeau, Karl Friston, and Klaas Stephan. 2011. “A Bayesian Foundation for Individual Learning Under Uncertainty.” Frontiers in Human Neuroscience 5. https://www.frontiersin.org/articles/10.3389/fnhum.2011.00039.

Presenting Author

Jesper Fischer Ehmsen

Poster Authors

Jesper Fischer Ehmsen

B.S

center of functionally integrative neuroscience

Lead Author

Topics

  • Pain Imaging