Posted on January 15, 2022

• Abstract
  • This paper is about the study of the cognitive process behind people's pain expectations, and specially how these differ between individuals. This is done by proposing a model of pain perception that leverages a Bayesian process. After studying how people reacted to "short and deception-free predictive cue tasks," the authors show the robustness of their model at predicting pain ratings. Also, they mention two big results when evaluating groups: the perception of pain can be conditional to the expectation of it; and, the more uncertainty there is in the expectation, the less impact the perception will have.
• Introduction
  • The model study was based on three pillars:
    • relation between expectation uncertainty and pain perception,
    • define what are the cognitive processes behind the expression "pain expectations", and
    • test the model to see if it can distinguish people based solely on those cognitive processes.
• The models
  • Model 1 - Simple:
    • Baseline
  • Model 2 - Multimodal
    • pain expectations follow a multi-model distribution
  • Model 3 - Mean-only:
    • pain expectations as the average of the possible pain levels
  • Model 4 - Mean-and-variance
    • similar to Model 3, but introducing the variance over the possible pain levels
  • Model 5 - Full
    • similar to Model 4, but included this time the effect of cue-independent pain expectations
      • note that cue-independent expectations are characteristics of the individual, driven by their own belief of pain expectation regardless the cue given
  • Model 6
    • similar to Model 5, but focusing on the individuals instead of the group. This adds random effects
  • Questions
    • when "predictive coding framework" is mentioned, what is the meaning of the word "coding"? Is it coding as in programming, or coding as in synthesizing information (e.g. genetic code)?
    • Is the linear transformation of the responses a standard procedure? (i.e. R'_{ij} = a_i + b_i x R_ij)
      • Isn't the cue-independent information removed after applying the transformation to the response?