An investigation into the effects of social evaluation on cardiovascular and endocrine responses to the CO₂ stress test in humans

Abstract

The present study examined whether social evaluation could heighten individuals' physiological responses to the CO₂ stress test, and the hypothalamic-pituitary-adrenal (HPA) response in particular. Twenty-five healthy volunteers undertook the CO₂ test under three conditions: (i) standard CO₂ protocol, (ii) standard CO₂ protocol conducted in front of a full-length mirror (mirror) and (iii) standard CO₂ protocol conducted in front of a video camera deemed to be transmitting live images of the procedure to investigators evaluating participant performance (video). Despite counterbalancing for task order, there were significant differences in anger and depression among the conditions. Repeated measures analysis of variances (ANOVAs), controlling for these mood indices, revealed that salivary cortisol, heart rate and systolic blood pressure responses to the CO₂ test were not affected by social evaluation (i.e. mirror or video). Although the data provide no evidence that endocrine and cardiovascular responses to the CO₂ test are affected by social evaluation, the potency of the social evaluation manipulation in this study is in question. Thus, further research is warranted which includes evidence of, or instructions suggesting negative social evaluation.

Keywords::

• Acute stress
• blood pressure
• cortisol
• CO₂ stress test
• heart rate
• social evaluation

Introduction

The acute administration of carbon dioxide (CO₂) has been used historically in psychiatry to stimulate anxiety. However, in recent years, modifications to the original procedure have enabled investigators to use it in clinical and non-clinical contexts as an acute physiological stressor (Argyropoulos et al. 2002; Kaye et al. 2004, 2005; Wetherell et al. 2006; Loft et al. 2007). This work has revealed that the acute administration of a mixture of 35% CO₂ and 65% oxygen produces: significant changes in self-reported mood, cardiovascular changes indicative of sympathetic and parasympathetic activation, and endocrine changes indicative of hypothalamic-pituitary-adrenal (HPA) axis activation.

Proponents of the test have suggested that it confers several advantages over many other laboratory stress tasks including: a time and labour efficient protocol (the challenge itself lasts only 4 s and does not require multiple experimenters); it activates both sympathetic and parasympathetic responses (few tasks achieve the latter) and, unlike most psychological stress paradigms, responses to the test appear not to habituate with repeated exposures (Wetherell et al. 2006). However, one aspect of the task which has been associated with some controversy is its effectiveness as an activator of the HPA axis. While several studies have reported evidence of increases in cortisol levels, particularly when measured in plasma (Argyropoulos et al. 2002; Kaye et al. 2005), other investigators have failed to see significant changes in the HPA axis following the inhalation of 35% CO₂ (Van Duinen et al. 2004; Loft et al. 2007). Thus, the present study was designed to examine whether physiological responses to this task, and the HPA response in particular, may be affected by external non-physiological stimuli. Specifically, we explored whether the introduction of social evaluation to the standard CO₂ test protocol would affect individuals' cardiovascular and endocrine responses to the task.

The potency of social evaluation as a stressor, and one which activates the HPA axis in particular, is supported by existing theory and empirical evidence. Specifically, stress and coping theory posits that psychological stress occurs when an individual perceives that the demands of their environment exceed their available coping resources (Lazarus and Folkman 1984). Social self-preservation theory (Dickerson et al. 2004) further specifies that when these external demands threaten the social self they can be particularly aversive, resulting in widespread changes in psychological and physical outcomes (including mood, self-esteem, cortisol secretion). Empirical evidence supports this position with results from a recent meta-analysis of over 200 experimental studies indicating that psychological stressors involving a social-evaluative threat were among the most potent activators of the stress response, as denoted by changes in cortisol secretion (Dickerson and Kemeny 2004). Indeed, this observation has been replicated and advanced further in several recent publications (Gruenewald et al. 2004; Rohleder et al. 2007; Dickerson et al. 2008). For example, the recent work of Dickerson et al. (2008) showed that increased cortisol levels in response to a standard speech task were only evident in participants who performed the task in front of a panel they were told would evaluate their performance, compared to participants who completed the task in front an individual who did not evaluate them, or indeed individuals who completed the speech task on their own. Rohleder et al. (2007) further demonstrated that the magnitude of the cortisol response to naturalistic social evaluative threat (ballroom dancing) was greater than that observed during laboratory-based stressors with a social evaluative component.

The above body of research has, therefore, demonstrated that physiological responses, in particular HPA responses, can be heightened when stressful tasks contain a social-evaluative component, and that the potency of these responses is greater for naturalistic stressors than experimental stressors. However, this literature has been largely concerned with psychological stressors. Thus, it is not clear whether social evaluation would exhibit comparable effects when presented in the context of a physiological stressor, such as the CO₂ stress test. One investigation which is of relevance in this regard is the recent work of Schwabe et al. (2008). They explored the effects of social evaluation on physiological responses to the cold pressor test. This test, which involves the immersion of the participant's hand in ice water, has previously been shown to exhibit robust sympathetic responses, but usually fails to produce an HPA response. In order to examine the effects of social evaluation, participants performed the test according to the standard cold pressor protocol, or the standard protocol with the addition of a social evaluation component. The latter was achieved by undertaking the test under the observation of an experimenter and informing subjects that they were being video-taped so that their facial expressions could be submitted to later analysis. The authors observed that, while the standard protocol failed to elicit changes in salivary cortisol, the protocol that included social evaluation did indeed result in significant increases in cortisol levels.

Thus, Schwabe et al. (2008) demonstrated that physiological stressors may indeed be affected by social evaluation in the same way that psychological stressors appear to be. Consequently, the present study was designed to examine whether social evaluation could heighten individuals' physiological responses, in particular HPA responses, to the CO₂ test. In order to explore this, individuals participated in the CO₂ stress test under three conditions. The first involved the standard CO₂ paradigm, which included no social-evaluative features. The second and third conditions involved modifications to the CO₂ paradigm in order to achieve a sense of social evaluation. This was achieved by conducting the test with (a) participants facing a mirror and (b) in the presence of a video camera which, participants were informed, was transmitting live images to investigators who were evaluating their performance. We sought to elicit social evaluation through these two approaches in order to examine a secondary aim, namely: could social evaluative threat be achieved through simple observation of the social self (i.e. one's mirror image) or did it require observation by another (as captured by the video camera condition).

Methods

Recruitment and participants

The present study was conducted with a higher education student sample at the end of the academic year and during summer vacations. Thus, only post-graduate students were approached to participate in order to minimise loss to follow-up during the observation period. Post-graduate courses were identified and the administrator of each course was contacted to request permission for their students to be approached regarding the study. Students were then approached, either via e-mail (distributed by the course administrator or the study team), or by a member of the research team attending a designated lecture.

Standard inclusion/exclusion criteria for the CO₂ stress test were adopted. Thus, healthy volunteers who reported that they did not have a history of hypertension, panic disorder or any other psychiatric disorder; who were not currently asthmatic or pregnant and who were not currently being treated for a psychiatric condition were approached. A sample of 28 volunteers was recruited initially. However, three participants withdrew following the initial stress testing procedure (all stating that they could not attend one or more of the subsequent scheduled test days), so data are reported from the 25 participants with complete data. The mean age of this sample was 28 years (19 female and 6 male), 13 participants described their marital status as single, 8 single but in a significant relationship, and 4 married or co-habiting. Five participants stated they were smokers. All participants were free of any medication (apart from a contraceptive pill) and were asked to not drink alcohol, caffeinated drinks or smoke after midnight before each testing day.

The study was approved by the local Research Ethics Committee and all participants gave written informed consent. Participants were paid an honorarium of £30 for their participation in the research.

Measures

Mood

The short form of the Profile of Mood States (POMS; Curran et al. 1995) was completed at each stage. The scale captures self-reported tension, vigour, depression, anger, confusion and fatigue. Cronbach's alpha reliability coefficients for the present study ranged from 0.60 to 0.96 across the six subscales across the three sessions.

Daily hassles

The hassles subscale from the hassles and uplifts scale (Kanner et al. 1981) was used to assess the severity of hassles reported at each stage. Cronbach's alpha reliability coefficients for the present study ranged from 0.87 to 0.97 across the three conditions.

Salivary cortisol

All saliva samples were initially refrigerated and then frozen at − 80°C until analysis. Cortisol was assayed using Neogen cortisol ELISA kits (Neogen Corporation, Lansing, MI, USA) using standard methods. Assays were conducted by an independent organisation (Cultech Ltd, West Glamorgan, UK).

Procedure

Participants were asked to attend three testing sessions during which they were informed they would complete the CO₂ stress test on each occasion. The sessions were conducted at 2-week intervals over a 6-week period (i.e. once every 2 weeks). In order to control for time of day effects on salivary cortisol levels all testing occurred between 11.30 and 14.30 h, and for each participant the timing of the first test session was maintained for the subsequent two sessions. At recruitment, participants were informed that the purpose of the study was to explore how external factors may influence their responses to a stress task. However, they were not informed of precisely which external factors were being examined, or that features of the CO₂ stress test procedure would vary on each occasion. The three conditions under which the test was performed are described below. The order in which participants undertook each condition was counterbalanced to control for task order effects. Measures of mood and daily hassles were completed prior to the completion of the CO₂ stress test at each testing session.

Standard CO₂ stress test protocol (standard condition)

This has been described in detail elsewhere (Wetherell et al. 2006; Loft et al. 2007). In brief, the protocol involved two stages (familiarisation and the stress task). First, participants were familiarised with all aspects of the procedure. This included the use of salivettes (Sarstedt, Germany) for the collection of saliva to determine cortisol levels. Participants were advised that they would be asked to collect saliva samples on four occasions during the procedure: 2 min prior to the stress test and again at 10, 20 and 30 min after the stress test. They were asked to place the cotton swab contained in the salivette in their mouths and to roll it around until complete saturation had been achieved. They were also familiarised with the equipment used to monitor blood pressure and heart rate. This involved the use of an automated cuff (Dinamap, GE Healthcare UK, St Giles, Chalfont, UK) which was programmed to take readings at 10, 5–0 min before CO₂/O₂ inhalation and again at 1–5 min after inhalation. Finally, they were trained in the inhalation protocol itself. They were advised that, for the stress task, they would be required to have their nose occluded (using a nose clip) and to inhale a single vital capacity breath of a mixture of 35% CO₂ and 65% O₂ (BOC gases, Guildford, UK). The mixture was contained in a 10 L Douglas bag attached to a silicone mouthpiece via a 3-way stopcock (Hans-Rudolf Inc, Shawnee, Kansas, USA) and a spirometer (Ko Ko, Louiseville, Kentucky, USA).

To ensure participants understood the experimental procedure and also to determine their vital capacities, they were asked to undertake a practice breath. For this, the participant's nose was occluded and mouthpiece inserted, but the Douglas bag was detached so that the participant inhaled normal air. Participants were asked to inhale a full breath, exhale fully and then take a second fast full breath. This second breath was held for 4 s after which the participant exhaled and the mouthpiece and nose clip were removed. During this practice period, inspired vital capacity was determined with reference to the spirometer. In order that participants had a direct visual reference of their vital capacity, they were shown a computerised image of their vital capacity displayed as a graph.

Once participants were familiar with the experimental procedure, the CO₂ stress test was undertaken. For this, blood pressure and heart rate recordings and saliva samples were collected as described above and the inhalation procedure was also repeated. However, for the stress test the Douglas bag was attached to the mouthpiece so that the second inhaled breath contained the CO₂/O₂ mixture. Participants were asked to monitor the vital capacity graph to ensure that they inhaled the vital capacity achieved during their practice breath. For all participants, the stress test breaths equalled or exceeded the vital capacity levels achieved during practice.

Standard CO₂ stress test protocol plus mirror (mirror condition)

In this condition, the standard protocol described above was repeated. However, all testing occurred with participants facing a full-length mirror so that they were able to see their own image throughout the testing procedure. No explanation was given by the experimenter for the presence of the mirror.

Standard CO₂ stress test protocol plus video camera (video condition)

In this condition, the standard protocol described above was repeated. However, participants were advised that their responses to the CO₂ test were to be recorded and that the images were being transmitted live to investigators in a nearby room who would be evaluating how the participant responded to the CO₂ test. They were further informed that, depending on how their responses compared with those of previous participants, the investigators may ask to conduct a brief interview with them after the stress task was completed.

Following completion of all three testing sessions, participants were asked to complete a brief structured interview designed to examine their perceptions of the experimental conditions. Accordingly, they were asked to reflect on the three testing sessions and describe whether and, where appropriate, why they perceived differences between the sessions. In this way, these interviews were a crude evaluation of the fidelity of the experimental manipulations. Following the structured interview, all participants were debriefed fully as to the nature of the study and received their honorarium.

Statistical analysis

As described above, data were collected on heart rate, systolic blood pressure and salivary cortisol before and after performance of the CO₂ stress test. The raw scores obtained for each outcome were transformed to capture cardiovascular and endocrine reactivity and recovery from the test. For systolic blood pressure and heart rate, the reactivity measure reflected the sum of the pre-CO₂ measures (i.e. 10, 5–0 min pre-inhalation) minus the values obtained at +1 min post-CO₂ inhalation. The recovery measure for these outcomes was captured by the difference between the +1 and +5 min values obtained post-CO₂ inhalation. For cortisol, the reactivity measure reflected the difference between the − 2 min pre-CO₂ inhalation and +20 min post-CO₂ inhalation values; and recovery was captured by the difference between the +20 and +30 min post-CO₂ inhalation values. Although test order was counterbalanced, analyses were conducted to examine the effects of test order on each of the outcomes. These revealed that the effect of test order did not attain statistical significance for any outcome (data not shown) and that effect sizes for task order ranged from 0.013 to 0.216 (mean η² = 0.08).

Results

Participants' views on experimental conditions

Responses to the structured interview revealed that, in answer to the question ‘Did you notice any differences about the place/the environment in which the testing sessions took place’, 72% commented on the presence of the video camera, mirror or both. When asked, ‘Do you think you responded differently on any of the 3 days?’ 92% stated that they did. However, when asked why they thought they may have responded differently, none of the participants implicated the presence of the mirror or video camera. Rather, 48% suggested that their expectations of the test varied on each occasion (most commonly that on the first occasion they thought they may have been more stressed/anxious because they were unsure what to expect); 20% thought that there may have been some variability in their completion of the CO₂ stress test across the three sessions and the rest were unable to specify a particular reason. When probed further regarding ‘why do you think the mirror was there?’ 68% stated it was there to make them self-conscious; and in answer to the question, ‘what effect did the mirror have on you?’ 52% reported that their feelings of discomfort were heightened during the mirror condition. The same questions were asked of the video condition: 64% stated that they felt the video camera was there so that their responses could be observed by others; and in answer to the question ‘what effect did the video camera have on you?’ 48% indicated that it heightened their feelings of discomfort during the video testing session.

Self-reported mood and daily hassles across the three experimental conditions

The mean number of hassles and mood scores on the POMS scale at each testing session are shown in Table I. Repeated measures ANOVAs revealed that individuals did not differ significantly in their mood and hassles scores between the three conditions, apart from on the measures of depression (F = 6.840, df 2,48, p = 0.003) and anger (F = 4.558, df 2,48, p = 0.015). Paired-samples t-tests revealed that, for depression, differences existed between the standard and video conditions (t = 3.122, p = 0.005) and the mirror and video conditions (t = 2.672, p = 0.013), such that levels of depression were lower during the video condition in both comparisons. For anger, paired samples t-tests revealed differences between the standard and mirror conditions (t = 2.224, p = 0.036) and the standard and video conditions (t = 2.492, p = 0.02), such that levels of anger were higher in the standard condition for both comparisons.

Table I.  Mean (SEM) scores for daily hassles and mood across the three experimental conditions.

	Mean (SEM), N = 25
	Standard condition	Mirror condition	Video condition
Daily hassles	33.32 (4.98)	25.13 (2.42)	32.39 (4.01)
Tension	8.26 (1.31)	6.36 (0.77)	7.13 (0.78)
Depression	8.09 (1.65)	6.04 (1.10)	4.52 (0.85)
Anger	7.57 (1.27)	5.48 (0.87)	4.83 (0.63)
Vigour	9.83 (0.96)	10.45 (1.16)	12.13 (0.98)
Fatigue	6.96 (1.19)	7.09 (1.24)	7.26 (0.87)
Confusion	5.83 (0.83)	5.61 (0.79)	5.74 (0.71)

SEM, standard error of the mean.

Cardiovascular and endocrine responses to the CO₂ stress test across the three experimental conditions

Mean cardiovascular and endocrine responses during the CO₂ stress test protocol are displayed for each experimental condition in Figures 1–3. Repeated measures ANOVAs were conducted to examine whether the computed measures of cardiovascular and endocrine reactivity and recovery differed significantly between the three conditions. In view of the significant differences in depression and anger between the standard and video and mirror conditions, respectively, these analyses were conducted with the measures of depression during the video condition and anger during the standard condition included as covariates. The results revealed that there were no significant differences in either reactivity or recovery indices for heart rate, blood pressure or cortisol across the three conditions. Furthermore, the amount of variance accounted for by the conditions was modest, ranging from only between 1 and 7% (Table II).

Figure 1.  Mean (and standard error, SE) heart rate responses during the CO₂ stress test period for the standard, mirror and video conditions (n = 25).

Figure 2.  Mean (and SE) systolic blood pressure responses during the CO₂ stress test period for the standard, mirror and video conditions (n = 25).

Figure 3.  Mean (and SE) cortisol responses during the CO₂ stress test period for the standard, mirror and video conditions (n = 25).

Table II.  Results from repeated measures ANOVA examining endocrine and cardiovascular reactivity and recovery during the CO₂ stress test across the three experimental conditions.

Condition	Mean (SEM)	F (df)	η^2	p
Salivary cortisol reactivity (nmol/l)
Standard	0.54 (0.17)	0.544 (2,48)	0.026	0.584
Mirror	0.52 (0.23)
Video	0.08 (0.43)
Salivary cortisol recovery (nmol/l)
Standard	− 0.20 (0.21)	0.629 (2,48)	0.030	0.539
Mirror	0.06 (0.16)
Video	− 0.04 (0.12)
Heart rate reactivity (BPM)
Standard	1.51 (1.97)	0.012 (2,48)	0.001	0.988
Mirror	3.99 (1.41)
Video	4.82 (1.80)
Heart rate recovery (BPM)
Standard	0.08 (1.84)	1.462 (2,48)	0.062	0.243
Mirror	− 2.56 (1.52)
Video	− 0.68 (2.38)
Systolic blood pressure reactivity (mm/Hg)
Standard	− 8.86 (2.76)	0.609 (2,48)	0.027	0.548
Mirror	− 5.39 (1.62)
Video	− 12.77 (2.40)
Systolic blood pressure recovery (mm/Hg)
Standard	11.24 (3.24)	1.784 (2,48)	0.075	0.180
Mirror	7.88 (1.88)
Video	16.08 (2.71)

SEM, standard error of the mean; nmol/l, nanomoles per litre; BPM, beats per minute; mm/Hg, millimetres of mercury.

Further post-hoc analyses were conducted to examine whether the results obtained were influenced by whether or not the participant smoked or their gender. With regard to smoking status, no statistically significant main or interaction effects were observed (all p>0.05: data not shown). No main effects were observed for gender either. However, significant interaction effects were observed for the measures of cortisol reactivity (F = 3.698, p = 0.034, η² = 0.163), heart rate reactivity (F = 6.193, p = 0.004, η² = 0.228) and heart rate recovery (F = 3.456, p = 0.051, η² = 0.257). Figure 4 reveals that, for all three indices, men showed greater variation in their responses between the conditions than women.

Figure 4.  Gender differences in mean (and SE) cortisol reactivity, heart rate reactivity and heart rate recovery scores between the three conditions (male, n = 6; female, n = 19). Significant gender effect, p = 0.034, p = 0.004, p = 0.051, respectively (ANOVA).

Discussion

The present study was designed to examine whether physiological responses to the CO₂ stress test, in particular the HPA response, could be enhanced by the addition of social evaluation to the standard protocol. We further explored whether social evaluative threat can be achieved through simple observation of the social self (i.e. one's mirror image) or whether it required observation by another (as captured by the video camera condition). Responses to the structured interviews revealed that the vast majority of participants perceived the three conditions as being different in some way, and approximately half of the sample reported that the video and mirror conditions were indeed associated with greater unease and discomfort. However, the endocrine and cardiovascular data did not correspond with these reports. For all outcomes, i.e. heart rate, systolic blood pressure and salivary cortisol level, no significant differences were evident in the indices of reactivity or recovery among the three experimental conditions.

At least two interpretations of these results are possible. First, one could conclude that responses to the CO₂ stress test appear not to be affected by social evaluation. Contrary to what has been reported by Dickerson and Kemeny (2004) and others, neither the video nor mirror manipulations resulted in heightened cardiovascular or endocrine responses to the task. This apparent contradiction may, at first, appear to be because the CO₂ stress test is a physiological stressor and social evaluation has most often been found to heighten physiological responses when included in psychological, rather than physiological, stress paradigms. Thus, there may be something unique about physiological stressors, and the CO₂ test in particular, that makes them impervious to the effects of social evaluation. However, the findings of Schwabe et al. (2008) caution against such a conclusion, in that they observed that social evaluation was able to heighten cortisol responses to a physiological stressor, i.e. the cold pressor test.

This leads us necessarily to a second, and competing, explanation, i.e. rather than the results suggesting that responses to the CO₂ stress test are unaffected by social evaluation, that instead our manipulation of social evaluation was inadequate and this led to equivalence in the physiological responses across the three experimental conditions. Research by Dickerson et al. (2004, 2008) offers some insight into why our social evaluation manipulation may have failed. Dickerson and colleagues have shown that social evaluation is most effective, and cortisol responses greater, when the evaluation is conducted by a live audience, rather than remotely (as might occur with video recordings), and when explicit negative social evaluation takes place. Thus, it would seem that the anxiogenic properties of social evaluation lie in part in the ‘actual presence’ of others, but also if there is explicit evidence of ‘negative’ social evaluation. Our video condition was intended to provide the most robust manipulation of social evaluation in that participants were advised explicitly that their responses were being assessed by a panel based in another location. However, according to the proposal of Dickerson et al. (2004, 2008), this manipulation may have been ineffective because the evaluation occurred remotely and participants were given no feedback, negative or otherwise. Accordingly, the equivalence in the physiological responses observed in the standard and video condition may indeed have been because our manipulation of social evaluation was inadequate. Accordingly, perhaps the most parsimonious interpretation of these data would be to suggest that, as the potency of the social evaluation in the present study is unclear, further research is necessary to establish whether or not physiological responses to the CO₂ stress test can be enhanced by the addition of social evaluation to the standard protocol.

Notwithstanding the above conclusion, a caveat is apparent from the post-hoc analyses examining the effects of gender. The findings revealed that, for the measures of cortisol reactivity and heart rate reactivity and recovery, there was a significant interaction between the responses of men and women on these indices. Specifically, Figure 4 reveals evidence of greater variability between the three experimental conditions in men, while women's responses remained largely stable. Indeed, the data from the male participants indicate that, in line with our expectations, cortisol reactivity and heart rate reactivity and recovery were greatest in the video condition; this indicates that our male participants may have been more sensitive to the social evaluation demands of the different conditions. This observation could be interpreted as being in accordance with the conclusions of Kudielka and Kirschbaum (2005). In their recent review of gender differences in the HPA axis, they concluded that male participants show greater reactivity to laboratory stressors than female participants. While the overall sample size in the present study, and the modest number of men, demand that our findings should be interpreted with caution, the results nevertheless highlight the importance of not assuming comparable responses to stressors between the genders.

The second aim of this study was to examine whether social evaluative threat could be achieved through simple observation of the social self. Here too, the findings of Dickerson et al. (2004, 2008) have implications for our understanding of the results pertaining to the mirror condition. As with the video condition, the data revealed there were no differences in responses to the CO₂ stress test between the mirror and standard conditions. As participants were not informed of the purpose of the mirror or even directed towards it, this condition failed to meet three of the task requirements set forth by Dickerson et al. (2004, 2008), i.e. that person or persons be present to undertake the evaluation; that it should be made explicit that social evaluation is taking place and that the evaluation be negative. Thus, again, it is perhaps not surprising that the manipulation failed to alter physiological responses to the CO₂ stress test. In this way, the present findings not only support but also extend the observations of Dickerson et al. (2004, 2008), i.e. they suggest that both observation of the social self and the mere presence of others are, on their own, inadequate techniques for eliciting the distress associated with social evaluation. It is clear that further research into the effects of social evaluation on responses to the CO₂ stress test would benefit from the inclusion of explicit negative appraisals of participants.

A further issue worthy of comment concerns the finding that the results of this study confirm that the CO₂ stress test has at best modest effects on the HPA axis. Consistent with previous work (Van Duinen et al. 2004; Loft et al. 2007), the present results showed that the test failed to produce significant changes in levels of salivary cortisol. Thus, it would seem that there remains room for further methodological refinement with this stress paradigm. In particular, for investigators to consider the introduction of methods such as social evaluation, which may enhance the HPA response.

In sum, the present study was designed to explore whether physiological responses to the CO₂ stress test, in particular the HPA response, could be enhanced by the addition of social evaluation to the standard protocol. Responses to the test did not vary significantly between the standard condition and the two experimental conditions designed to achieve social evaluation. These findings may, however, have been affected by both the modest sample size and weaknesses in the methodology employed to achieve social evaluation. As our findings also indicate that the CO₂ stress test continues to have only modest effects on the HPA axis, further research is warranted into whether the introduction of explicit negative social evaluation or other procedural modifications can enhance the HPA response to this stressor.

Declaration of interest: The authors report no conflicts of interest. The authors alone are responsible for the content and writing of the paper.