Recognition of facial expressions of pain using spatial frequency information

  • Shan Wang

Student thesis: Doctoral ThesisPhD


Individuals’ pain experiences can be communicated both verbally and non-verbally. Facial expressions are a primary non-verbal channel of pain communication, and so need to be clearly and unambiguously recognised and differentiated from other non-noxious emotional expressions. It is known that we are able to identify others’ pain from their facial expressions in an accurate and efficient manner, even under challenging visual conditions. However, little is known about how facial expressions are processed by observers, and what information is actually used, to make the identification of pain possible. To account for this, the current thesis considered facial expressions as a type of visual stimulus and investigated possible mechanisms that underpin the recognition of pain expressions from the perspective of perceptual information analysis.Spatial frequency (SF) information is a type of fundamental perceptual information that encodes different characteristics of a visual display. For a facial expression, low-SF information conveys the large-scale facial configuration and structural changes, whereas high-SF information depicts the fine details of facial features. In order to understand how we recognise pain expressions in terms of SF analysis, a series of experiments were conducted within this thesis to primarily investigate the role of low-SF and high-SF information in the recognition of pain expressions (Experiment 1–4), and the temporal feature of low-SF and high-SF information processing in pain recognition (Experiment 5–7).Data of this thesis revealed that although pain expressions could be recognised with either low-SF or high-SF information available, low-SF information plays a prominent role that leads to more accurate judgements (Experiment 1) and is preferentially perceived by observers (Experiment 2–4). Moreover, the processing of low-SF information shows a temporal advantage over high-SF information (Experiment 5). Pain expressions presented with low-SF information only was decoded more rapidly than those presented with high-SF (Experiment 6), and the asynchrony between low-SF and high-SF processing originated from a very early stage of information extraction (Experiment 7). Therefore, the decoding of low-SF pain expressions is not only faster in duration but also precedes the decoding of high-SF pain. Altogether, these findings suggest that when we differentiate facial expressions of pain from non-noxious emotions, the coarse low-SF information plays a key role by providing a preliminary understanding of the overall quality of pain expressions rapidly, and the fine-detailed high-SF information is integrated at a later stage and plays a trivial role. More interestingly, this pattern was found not only for the recognition of pain expressions, but also the core emotions investigated, which suggests that expressions of pain and core emotions share similar visual perceptual properties.This thesis provides a visuoperceptual account of how we recognise facial expressions of pain and suggests that in addition to analysing a series of facial action units the recognition of pain expressions is also a visual perceptual process that relies heavily on the perceptual information analysis. Limitations that were associated with the research contained within this thesis were acknowledged, suggesting directions for future research.
Date of Award1 Mar 2017
Original languageEnglish
Awarding Institution
  • University of Bath
SupervisorEdmund Keogh (Supervisor) & Christopher Eccleston (Supervisor)

Cite this