Bayesian Inference in Pain-Related Processing

Background

Bayesian inference is a unifying theory of perception. Under this theory, perception occurs through combining our senses with beliefs and experiences. Research about how humans make inferences is less studied in the context of pain, including whether people actually integrate sensory and belief information to judge pain, and whether personal attributes and experiences (e.g., experience with pain, mental health) shape expectations about pain and thus the experience of it.

Research Questions / Hypotheses

Primary research questions:

1. Do people use inferential strategies to judge pain that is being experienced in others?
2. Do personal factors, such as previous experience with pain and depression, influence pain perception?

Aims:

1. Explore whether people use Bayesian inferential strategies to interpret pain-related stimuli.
2. Explore correlations between demographic variables (prior pain, depression) and the degree to which people rely on sensory information when perceiving pain in others.

Participants

662 completed the task. 57 participants were excluded due to either erroneous or outlier or missing data.

Methods

Experiment: Participants are asked to complete an online experiment. The experiment involves participants viewing a series of symbolic pain cues (ratings on a pain scale), in order to infer the underlying intensity (in degrees Celsius) of a noxious heat stimuli. The ratings are provided by model participants who are attempting to rate their pain as accurately as possible based on a single underlying temperature in each trial. One of the two models (Model A or Model B), is more precise than the other at appraising their pain (i.e., their pain ratings tend to be more consistent with the temperatures experienced). Participants are required to guess the underlying temperature (from 32-50 degrees) based on 5x pain ratings (0-100) provided by each model participant, taking into account that one model tends to be more accurate at rating their pain. Surveys: The study includes questionnaires related to somatic symptoms (e.g., Somatic Symptoms Scale), including the experience of pain (e.g., back pain in the past 7 days) to determine the participants' prior experience with pain. Given that previous work has also suggested potential links between pain and depression, the questionnaire included a survey on depression, to determine whether depression mediates the relationship between pain and Bayesian inference strategies.

Results

There was no direct relationship between prior pain experiences and Bayesian preferences observed in a pain-free student population. Depression status significantly predicted Bayesian preferences for priors: individuals with moderate-to-severe depression were less likely to rely on sensory information (i.e., are more likely to rely on prior beliefs) about pain, compared to people with mild depressive symptoms. Higher levels of depression predicted greater weekly pain, which is consistent with previous research of depression being associated with pain symptoms.

Implications

Increasing pain complaints with higher depression levels might influence how the pain perception informed. However, given the present study involved participants from a largely healthy and pain-free population, it is unclear whether pain actually influences Bayesian inferential strategies. Further research would benefit from involving individuals with chronic pain to explore this relationship. Further research is also needed to determine if this effect is specific to pain or generalisable to other domains involving the perception and judgement of stimuli. Overall, the findings highlight the importance of considering depression in understanding the relationship between pain and Bayesian decision-making. Given the present study was limited to a healthy adult population, future studies should focus on depression with varying levels of severity.