Signal detection theory in the study of nociceptive and pain perception processes

Abstract

Signal detection theory (SDT) measures (discriminability and response bias) have been proposed to be valid for determining pain perception changes. The construct validity of SDT measures applied to pain perception studies has been questioned on three grounds: interpretation, methodology and theory. Multiple interpretations are possible for the combinations of discriminability and response bias change when the magnitude-rating scale is used for pain perception studies. This is resolved by utilising the confidence-rating scale. The problem of comparability of results between the two scales is bridged by Irwin & Whitehead's (1991) common analytical framework. The results of this thesis supported the framework's prediction that both scales are comparable. Therefore, the confidencerating scale was used for all studies within this thesis for interpretational clarity. Response bias data were not analysed in this thesis due to data artefacts created by correction methods for zero proportions in response categories. Methodologically, the construct validity of discriminability is influenced by the research design and procedures. Therefore, the following procedures were adopted to address weaknesses in previous studies. The one-interval confidence-rating task was used with a six-category confidence-rating scale and post-trial feedback. Based on a methodological study conducted within this thesis, the trial number was pragmatically reduced from 40 trials to 17 trials per stimulus intensity. This trial number reduction would not alter the mean and variance of the data sufficiently to influence the outcome of inferential statistical testing performed. Due to the novel use of the Quantitative Sensory Testing machine for the signal detection study procedures, accuracy and precision study on the machine was performed. This thesis found that the accuracy, repeatability and reproducibility of the machine in generating noxious thermal stimuli is excellent for the purposes of this thesis. Machine error is eliminated as a major source of variance for the thesis results. Theoretically, critics have challenged the construct validity of discriminability as an indicator of pain perception alteration. This thesis examined this issue in two separate contexts: 1) discriminability change as a correlate of local anaesthesia and, 2) discriminability as a correlate of psychological factors (depression and anxiety) in chronic low back pain (CLBP) sufferers. The results failed to establish the construct validity of discriminability for both contexts. However, the higher discriminability in CLBP sufferers compared to healthy individuals is in contrast to past research and warrant further investigation. This thesis addressed the construct validity issues through theoretical, methodological and interpretational modifications. A more robust analysis of the construct validity issue was facilitated. Caution is recommended on the use of discriminability as a pain perception measure until the construct validity issue has been satisfactorily resolved.