Cognitive decline is known to reduce reliability of subjective pain reports. Although facial expressions of pain are generally considered to be less affected by this decline, empirical support for this assumption is sparse. The present study therefore examined how cognitive functioning relates to facial expressions of pain and whether cognition acts as a moderator between nociceptive intensity and facial reactivity. Facial and subjective responses of 51 elderly participants to mechanical stimulation at three intensities levels (50 kPa, 200 kPa, and 400 kPa) were assessed. Moreover, participants completed a neuropsychological examination of executive functioning (planning, cognitive inhibition, and working memory), episodic memory, and psychomotor speed. The results showed that executive functioning has a unique relationship with facial reactivity at low pain intensity levels (200 kPa). Moreover, cognitive inhibition (but not other executive functions) moderated the effect of pressure intensity on facial pain expressions, suggesting that the relationship between pressure intensity and facial reactivity was less pronounced in participants with high levels of cognitive inhibition. A similar interaction effect was found for cognitive inhibition and subjective pain report. Consequently, caution is needed when interpreting facial (as well as subjective) pain responses in individuals with a high level of cognitive inhibition.
In the general population, the self-report of pain is typically viewed as the golden standard in pain assessment. In dementia, however, these self-reports are limited by the strong cognitive decline accompanying the disease, as it impairs language capacities and thereby the patients’ ability to communicate about their pain. Moreover, dementia causes a reduction in abstraction abilities, which reduces the patients’ ability to comprehend and thereby use pain scales to indicate their pain. Experts have therefore recently identified pain behaviors as being crucial in order to obtain reliable and valid pain assessments in dementia. Facial expressions form an important part of pain behaviors in the assessment of pain. Facial responses are not compromised by language impairments and, according to some studies, may be less dependent on the desire of expressing pain since facial expression is a rather automatic process [
However, although facial expressions of pain are thought to be relatively unaffected by cognitive decline compared to self-report, there is evidence that facial expressions are not completely unrelated to cognitive performance. Keltner and coworkers [
The goal of the present study was to add to our understanding of how cognitive functioning (especially executive functioning) affects the self-report and facial expression of pain. Studies so far mainly assessed facial expression of pain using the Facial Action Coding System (FACS) [
As the dementia process induces severe cognitive decline, which hinders both pain report and neuropsychological functions to be reliably assessed, we focused on normal aging adults, as this population is still able to provide reliable pain reports and to undergo a neuropsychological examination. Moreover, we focused on a wide age range, as from the age of 50 years onward a significant decline in cognitive functions, including executive control, can be detected [
Fifty-two older adults between the ages of 50 and 93 years were recruited for this study. Participants were volunteers recruited through advertisements in a local newspaper and through oral advertisement; in addition, some volunteers were acquaintances of the researcher. Education was measured using an ordinal rating scale that ranges from 1 to 7. Here, score 1 represents incomplete primary education, score 2 reflects primary education, score 3 reflects incomplete lower secondary education, score 4 reflects lower general secondary education, score 5 reflects vocational education, score 6 reflects higher general secondary/higher vocational/preuniversity education, and score 7 represents an academic degree [
All neuropsychological tests were administered in a fixed order. This was necessary to include a fixed delay between immediate and delayed memory testing (see Section
Since executive functioning was our main focus, we employed three tasks to assess this heterogeneous domain. These were the Stroop task, the Digit Span Backward task, and the Zoo Map task. The Stroop task was employed as a measure of cognitive inhibition [
In addition to these executive function tasks, episodic memory was measured since this function is known to decline with aging as well [
Finally, psychomotor speed, a function very sensitive to the age-related decline [
For further analyses, standardized scores were calculated for the cognitive outcome measures in order to create cognitive domain scores, so as to reduce the number of statistical tests necessary (which reduces risk of type I error). Hence, an executive domain score (consisting of Stroop interference, Digit Span Backward, and the Zoo Map test), a memory domain score (AVLT immediate recall and delayed recognition, Story Recall immediate and delayed recall), and a psychomotor speed domain score (Stroop Word and Color card) were calculated. Cronbach’s alpha was calculated to test reliability of these domains, in order to determine whether it was appropriate to use these domains for the analyses. As previous studies indicated that specifically cognitive inhibition may play a unique role in facial expressiveness, the executive function measures were also examined separately.
Perception of noxious mechanical pressure was administered using a Wagner FPX
All stimulation sessions were conducted by trained psychology students who also conducted the neuropsychological tests. We used a standardized protocol, and the students performed extensive practice sessions prior to starting the study to assure that they complied to this protocol.
Facial expressions were video-taped during the mechanical pain test and during a baseline period using a camera that was located in front of the participant at a distance of approximately 1.5 meters. Participants were instructed to maintain focus to a predefined location in front of them, in order to guarantee a frontal view and to avoid talking while pressure was applied. Facial expressions were analyzed offline in time windows of 7 seconds (covering the stimulation period or, in case of baseline trials, the time period before starting the pressure stimulation) using facial descriptors extracted out of existing observational pain assessment scales. This extraction has led to the development of the Pain Assessment in Impaired Cognition (PAIC) metatool [
Observation of facial items within the painful trials (400 kPa) in 51 participants. Selection of pain-indicative items was based on frequency of occurrence (>5%) as well as on a more frequent occurrence during pain compared to baseline (effect size
Facial items | Percentage of occurrence |
Effect size |
---|---|---|
Pained expression | 11.8 |
|
|
|
|
|
|
|
Closing eyes | 5.0 |
|
|
|
|
|
|
|
|
|
|
Clenched teeth |
|
|
Empty gaze | 50.0 |
|
Seeming disinterested |
|
|
Pale face |
|
|
Teary eyed |
|
|
|
|
|
Looking sad |
|
|
Looking frightened |
|
|
After each stimulation, participants rated their pain using a 0–10 numerical rating scale (NRS). Average NRS pain scores were calculated for each of the stimulus intensities, resulting in one average NRS pain score for 50 kPa pressure, one for 200 kPa pressure, and one for 400 kPa pressure.
(i) To test the hypothesis that a decline in executive control is associated with increased facial expressions of pain and increased NRS scores, regression analyses were employed, entering the executive function scores as predictor variables and facial expressions or NRS ratings, respectively, as criterion variables. Given that we applied 2 pressure intensities that lay in the noxious range (200 and 400 kPa), analyses were conducted separately for facial and subjective responses to 200 kPa and 400 kPa, respectively, resulting in 2 (NRS scores, facial expression) × 2 (200, 400 kPa) = 4 regression analyses.
(ii) To test whether potential associations are indeed specific for executive functioning, as was suggested by previous studies, we conducted blockwise regression analyses, this time entering memory and speed function in the first block of predictors and executive functioning in the second block. This allows us to test whether executive functioning can add predictive power beyond that already explained by memory and speed performances. Again, analyses were conducted separately for NRS ratings and facial expressions and separately for the 2 noxious intensities.
(iii) In order to examine whether cognition moderates the relationship between pressure intensity and facial and subjective pain responses, repeated measures analysis was employed with pressure intensity (50 kPa, 200 kPa, and 400 kPa) as within-subjects variable and the executive functioning scores as covariates. This analysis was conducted twice, once with the NRS scores as dependent variable and once with the facial expression scores as dependent variable.
Analyses were conducted with SPSS 22 and alpha level was set to 0.05. In case of directed hypotheses, we used one-sided testing.
Participant characteristics, together with the results from the pain assessment and the neuropsychological examination, can be found in Table
Characteristics, pain NRS, and facial expression scores of the participants.
Variable |
|
|
---|---|---|
Age (yrs) | 51 | 66.7 (12.0) |
Sex (M/F) | 51 | 26/25 |
MMSE | 51 | 28.7 (1.4) |
Education | 51 | 5 (2) |
NRS 50 kPa | 51 | 1.2 (1.6) |
NRS 200 kPa | 51 | 2.9 (2.3) |
NRS 400 kPa | 51 | 5.0 (2.6) |
Facial expressions 50 kPa | 51 | 0.4 (0.6) |
Facial expressions 200 kPa | 50 | 0.7 (0.9) |
Facial expressions 400 kPa | 51 | 1.2 (1.4) |
Stroop Word card (s) | 51 | 55.9 (15.9) |
Stroop Color card (s) | 51 | 67.3 (20.1) |
Stroop Color/Word card (s) | 51 | 127.5 (95.2) |
Zoo Map score | 51 | 11.0 (4.0) |
Digit Span Backward | 51 | 5.8 (2.3) |
AVLT immediate recall | 51 | 38.6 (11.1) |
AVLT delayed recognition | 51 | 27.3 (2.9) |
RBMT immediate story recall | 51 | 7.5 (3.4) |
RBMT delayed story recall | 51 | 6.3 (3.2) |
Descriptive represent means (±SD), with the exception of sex, where frequencies (male (M)/female (F)) are presented, and education where the median score (IQR) is presented. The facial expression scores represent the average number of pain-specific expressions. AVLT: Auditory Verbal Learning Test; MMSE: Mini Mental State Examination; NRS: numerical rating scale; RBMT: Rivermead Behavioural Memory Test; s: seconds; yrs: years.
Results from the regression analyses are presented in Table
Association between executive functioning and facial or subjective responses to noxious stimulation (200 and 400 kPa).
Criterion variable |
|
Pressure intensity |
|
|
|
|
| ||
---|---|---|---|---|---|---|---|---|---|
Stroop interference | Digit Span Backward | Zoo Map test | |||||||
Facial expression | 50 | 200 | .575 | .054 | −.203 | .596 | .355 | 8.438 |
|
51 | 400 | −.124 | −.233 | .123 | .223 | .050 | .820 | .244 | |
|
|||||||||
NRS score | 51 | 200 | .299 | −.281 | .055 | .446 | .199 | 3.894 |
|
51 | 400 | .321 | −.231 | .170 | .435 | .190 | 3.664 |
|
NRS: numerical rating scale.
Results of the blockwise regression analyses are displayed in Table
Specificity of the association between executive functioning and facial or subjective responses to noxious stimulation (200 and 400 kPa).
Criterion variable | Pressure intensity |
|
Predictor variables |
|
|
Significance of |
|
---|---|---|---|---|---|---|---|
Facial expression | 200 | 50 | Block 1 | Memory & speed | .349 | ||
50 | Block 2 | Executive functioning | .438 | .089 |
|
||
400 | 51 | Block 1 | Memory & speed | .005 | |||
51 | Block 2 | Executive functioning | .067 | .063 | .198 | ||
|
|||||||
NRS rating | 200 | 51 | Block 1 | Memory & speed | .307 | ||
51 | Block 2 | Executive functioning | .335 | .028 | .293 | ||
400 | 51 | Block 1 | Memory & speed | .229 | |||
51 | Block 2 | Executive functioning | .281 | .052 | .184 |
Results of blockwise regression analyses are presented. NRS: numerical rating scale.
Repeated measures analysis, with pressure intensity as within-subjects variable, the NRS scores as dependent variable, and the executive function measures as covariates, revealed a significant interaction between the Stroop interference score and pressure intensity (
Characteristics of the three interference control groups.
Variable | Low interference control | Average interference control | High interference control | Statistical test |
---|---|---|---|---|
|
17 | 17 | 17 | — |
Age | 74.0 (12.3) | 67.5 (11.5) | 58.6 (6.5) |
|
Sex (M/F) | 9/8 | 11/6 | 6/11 |
|
MMSE | 28.4 (1.5) | 28.6 (1.8) | 29.1 (0.9) |
|
Education | 4 (1.5) | 5 (2) | 6 (2.0) |
|
Stroop interference score | 2.26 (1.85–4.66) | 1.71 (1.54–1.84) | 1.41 (1.18–1.54) | — |
Means (±SD) are presented for age and the MMSE, means (range) for the Stroop interference score, frequencies for sex, and median score (IQR) for education. F: females; M: males; MMSE: Mini Mental State Examination;
Numerical rating scale (NRS) scores at different pressure intensities of participants with low (
The Digit Span Backward and the Zoo test did not interact with the effect of pressure intensity on subjective ratings (all
A same repeated measures analysis, now with facial expressions as dependent variable, demonstrated a significant effect of pressure intensity (
Facial expressions at different pressure intensities of participants with low (
The Digit Span Backward and the Zoo test did not interact with the effect of pressure intensity on facial expressions (all
The present study examined the interrelatedness between cognitive functioning, noxious intensity, and facial expressions of pain in elderly people. Our primary goal was to investigate whether executive function in particular would show a relationship with facial expressions following painful stimulation and if these functions moderated the effect of noxious intensity on facial expressiveness. Moreover, given that subjective pain reports are regarded as being prone to the age-related cognitive decline, in contrast to the more automatically generated facial expressions [
Overall, the results showed that variations in subjective and facial responses to noxious stimulation could indeed be explained by executive functioning. The associations were strong for subjective responses, whereas only variations in facial expressiveness to mild noxious stimuli were significantly associated with executive functioning. However, when investigating how specific the associations were for executive functioning compared to the neuropsychological domains “memory” and “speed,” we only found a specific association between executive functioning and facial expressiveness (to mild pain). Here, executive functioning still added explained variance even when controlling for memory and speed performances. In contrast, variance in self-report ratings to noxious stimulation was sufficiently explained by memory and speed performances.
In addition to these regression analyses, we also investigated whether executive functioning might moderate the association between noxious intensity and facial as well as subjective responses using covariate analyses of variance. We found that the interference score as measured with the Stroop test significantly moderated the relationship between pressure intensity and both the subjective pain report and the facial expressions. These interactions indicated that those older adults with better interference control abilities show a less pronounced increase in both the pain report and the facial expressions following painful stimulation, when compared to elderly people with a lower (i.e., worse) level of interference control. A further comparison of these subgroups showed that the only significant group difference was one in age, which supports the notion that it is an age-related decline in inhibition capability that plays an important role in the facial expressions of pain. Moreover, as the interaction effect was isolated to the Stroop task (other executive tasks did not demonstrate an interaction with the facial or subjective expression of pain), we believe that it does not reflect a general effect of age. If it were a general age effect, other executive tests should have also yielded significant interaction effects.
The interpretation of these findings has crucial clinical implications. The relationship between subjective pain ratings and executive functioning appears to be nonspecific: although significant associations with executive control were found, they disappeared after the significant confounding effect of memory and psychomotor speed was included. Hence, regardless of the specific cognitive domain, a general relationship is present where cognitive decline is associated with higher NRS scores. This contrasts with findings for facial expressions, where a unique association with executive functioning was found.
Furthermore, the current study suggests that the level of cognitive inhibition is crucial for the extent to which facial expressions of pain are displayed. A previous study in healthy participants already suggested that cognitive inhibition, as measured with the Stroop interference control score, but not other executive functions such as shifting, working memory, and planning, is associated with experimental pain sensitivity [
Finally, some other findings deserve attention in the discussion regarding interpretation of facial expressions of pain. First of all, the low interference control group demonstrated increased facial expressions but similar NRS scores compared to the participants with average levels of interference control. This might indicate that this group is specifically unable to inhibit their facial expression to painful stimulation but does not actually experience more pain. Second, given that facial expressions were reduced and did not increase substantially across intensities in the high cognitive inhibition group, interpreting pain based on the facial expressions in adults with high levels of cognitive inhibition could result in an
In contrast to “interference control” the other types of executive functioning (planning and working memory ability) were less strongly related to facial reactivity to pain. It is possible that the commonly assumed heterogeneous nature of executive functioning is also evident in distinct associations with regard to pain outcomes. This suggests that whereas interference control may show an inverse relationship with the level of pain that is reported and facially expressed, this association may be different for other frontal functions. In the existing literature, also indirect evidence for a heterogeneous link between the frontal lobes and pain can be found. Patients with frontotemporal dementia, for example, who normally show severe frontal brain damage, have been shown to display reduced pain awareness as indicated by patients’ proxies [
Some limitations of the present study need to be addressed. First, participants were instructed to refrain from talking. This might have caused subjects to keep a still face in general and to consciously inhibit facial pain expressions. The fact that the level of facial expressions was generally low might also be related to this point. It is crucial to realize though that the threshold for facial expression of pain is much higher than the subjective pain threshold. That means that individuals just start to facially express their pain once the pain is of moderate or sometimes even strong intensity [
A second drawback of the current study is that the intensities of the applied stimuli were rather low so that some subjects might not have experienced any noteworthy pain. This impression was supported by several participants hesitating or looking doubtful about their pain ratings, especially on the second intensity, as if they were expected to feel actual pain and give higher ratings than on the first occasion but did not really perceive the stimulus as painful. Nonetheless, the fact that we did find a comparable effect of interference control on the increase in pain responses, whether measured by report (subjective) or by facial expressions (objective), supports reliability of our findings.
Third, all stimulation intensities were applied in the same ascending order to prevent that a first stimulation at a high intensity might have an analgesic or even induce a hyperalgesic effect for subsequent lower-level stimulations. Stated otherwise, starting with, for example, 400 kPa stimulation intensity might mask or exacerbate responses to subsequent lower stimulation levels. Although there was a clear rationale for using this fixed order, it might have influenced how participants reacted to the pain. A solution would be to let participants first get accustomed to different pressure intensities, in order to increase reliability of the pain assessment protocol. Regarding the rating of the facial expressions, this was also always accomplished in a fixed order. Hence, the rater may have been influenced by expectations regarding facial expression as the pressure intensity increased. Nonetheless, as the rater was blinded to all study outcomes (e.g., cognitive test results), it is unlikely that expectations of the rater can explain the observed interaction with the interference control score.
Finally, the current study examined only elderly people that were not diagnosed with a neurodegenerative disorder. In order to generalize the results to other populations such as children or patients with dementia, a replication within these populations is necessary.
The present study indicates that cognitive inhibition moderates the effect of stimulus intensity on pain ratings and facial expressions. Nonetheless, the results also indicate that, in contrast to subjective pain ratings, facial expressions are less likely to be influenced by a general cognitive decline, supporting the clinical utility of these expressions for pain assessment purposes in populations with limited communicative abilities. Future studies are needed, addressing these associations in diverse populations.
The authors declare that there are no competing interests regarding the publication of this paper.