Pain Prediction Can Bias Pain Sensation

Post by Elisa Guma 

What's the science?

Pain is necessary for survival, both for quickly responding to aversive stimuli and for predicting potential harm from objects or situations. It is generally known that the experience of pain is related to nociception: the activation of special nerve endings in the body by potentially harmful stimuli such as mechanical (e.g. a sharp nail) or thermal (e.g. high heat). However, our experience of pain often deviates from nociception. Previously, it was shown that pain is influenced by our predictions of how painful the stimulus will be by using simple-Pavlovian cues. This week in the Journal of Neuroscience, Lim and colleagues show that pain can be strongly influenced by predictions from a conceptual schema. 

How did they do it?

The authors recruited 42 healthy adults and administered a series of tests in order to investigate how pain ratings are affected by a mismatch in pain prediction (top-down cognition; from the brain) and sensation due to nociception (bottom-up; from the body/nerves). First, a baseline pain rating was recorded for two heat stimuli used as the pain stimuli: 45°C (low) or 47°C (high). Next, all participants underwent functional magnetic resonance imaging while the pain tests were administered. This allowed the authors to make associations between neural responses to a) pain prediction and b) prediction errors. In the matched condition, participants learned a schema: when the value shown in the visual cue increased the pain produced by the heat stimulus also increased. Thus, visual cues stated “The incoming heat stimulus is at x% intensity”. The x value was a number between 0-100, and was increased linearly, with a 10-point increase corresponding to a 0.4 degree increase in stimulus temperature. 

In a second task set (mismatch level 1), the authors gradually introduced prediction errors where initially the cues continued to vary between 1-100, but the heat was held at either 45 ° C for cue values  0-40 or at 47 ° C for cue values from 61-100. Subsequently, the mismatch level was increased over a third and fourth task set (mismatch level 2) where the cues kept changing between 1-100, but the stimulus was always at 47°C . Participants were not given information on the mismatch and made their own decision on how much pain they felt after experiencing each cue-heat stimulus pairing. Finally, to better understand the factors that might mediate perceptual biases, a pain catastrophizing scale was administered, as well as a mindfulness questionnaire.

What did they find?

First, the authors established that participants were able to successfully detect a linear pattern between pain (heat) stimuli and cued (visual) stimulus intensities. Interestingly, the authors observed a bias in pain perception driven by the cue, such that the cue value had a greater influence on pain perception than the sensory stimulus. For the majority (5/8) of mismatched conditions for mismatch level 2, and (to a lesser degree) for mismatch level 1, participants partially updated their pain ratings with the change in heat stimulus, but the predictions had a stronger influence on perception than the prediction errors from the heat stimulus. 

The authors observed that areas typically involved in pain perception were responsive to increases in cued threat (cue stimulus increases), such as somatosensory regions, posterior insula and periaqueductal gray area (corrected, FLAME1). When prediction errors were high, more cognitive cortical areas became active. Finally, the authors investigated individual differences in pain perception and found that people who had higher catastrophizing behaviours and lower mindfulness or sensory awareness (as measured by a self-report scale) tended to be more affected by the cues or threat predictions. These behaviors involved the ventral and dorsal striatal circuitry, which are implicated in value-based vs. model-free or habitual decisions respectively.   

What’s the impact?

This study demonstrates that predictions that arise from learning concepts have a strong impact on pain perception even if there are prediction errors arising from sensory inputs. In other words, our prediction of pain based on a reported upcoming stimulus intensity impacts our pain even if the level of stimulus intensity eventually administered is not the same as the reported intensity. Further, changes in the levels of cued-threats altered activity in cognitive and sensory networks in an opposite manner, which underscores the role of these top-down and bottom-up networks in biasing pain perception. Finally, the authors were able to attribute behavioural patterns with perception, showing that individuals with higher mindful awareness and less fear of pain are better able to update their pain perception.

Lim et al. Threat prediction from schemas as a source of bias in pain perception. Journal of Neuroscience (2020). Access the original scientific publication here.