Article Text

Download PDFPDF

Cognition and performance: anxiety, mood and perceived exertion among Ironman triathletes
  1. David Parry1,
  2. Camilla Chinnasamy1,
  3. Eleni Papadopoulou1,
  4. Timothy Noakes2,
  5. Dominic Micklewright1
  1. 1Department of Biological Sciences, Centre for Sports and Exercise Science, University of Essex, Colchester, UK
  2. 2Department of Human Biology, UCT/MRC Research Unit for ESSM, University of Cape Town, Cape Town, South Africa
  1. Correspondence to Dr Dominic Micklewright, Department of Biological Sciences, Centre for Sports and Exercise Science, University of Essex, Wivenhoe Park, Colchester, Essex CO4 3SQ, UK; dpmick{at}


Objective The authors examined the changing patterns of mood before and after an Ironman triathlon, and the relationships between expected performance outcomes, perception of effort and pacing.

Design Twelve participants in the 2008 Ironman Austria triathlon competition were studied before, during and after the event. Each participant completed measures of mood, anxiety and perceived exertion, while pacing was calculated from official race timings at various points on the course.

Results Positive correlations were found between distance covered and rating of perceived exertion (RPE) during each of the individual disciplines, and also between RPE and the percentage of overall race time completed (r=0.826, p<0.001). A negative correlation was found between average speed and distance covered during the run segment (r=−0.911, p<0.005) with pace gradually declining. Differences occurred in the profile of mood states mood subscales of tension and fatigue between the baseline, prerace and postrace trials. Somatic anxiety was higher before the race compared with baseline measures.

Discussion RPE followed a linear progression of RPE during each discipline followed by a re-setting of the perception of effort at the start of the next discipline. The increase in RPE for the entire event followed a linear increase. The linear decline in run pace is consistent with a recent model in which expected RPE is used to modulate pacing. Anxiety and mood responses of participants in this study indicate that the emotional response of athletes before and after ultra-endurance exercise is closely aligned with their conscious thoughts.

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

It has been suggested that the regulation of intensity during self-paced endurance exercise occurs in a feed-forward manner derived from information about the physiological state of the body,1,,4 sensory information about the external environment5,,7 and cognitive interactions between previous experience and knowledge of distance to the endpoint.8,,14 This teleoanticipation operates by means of a feedback control loop, monitoring afferent feedback from the body as well as processing sensory information from the external environment.6 7 This endogenous and exogenous information facilitates pace monitoring and, by efferent motor output to working muscles, modifies pace in order to preserve homeostasis in physiological systems and avert premature fatigue during exercise.

Ultra-endurance exercise places unique demands on teleoanticipation because of the difficulty of predicting endogenous and exogenous changes that could occur over a long time period. The influence of uncertainty on pacing was discussed in a recent review by Tucker and Noakes,10 and is demonstrated by studies in which uncertainty about elapsed distance leads to more conservative pacing,15 with faster ultra-endurance runners making fewer adjustments in speed16 and Ironman triathletes adopting a non-linear cycling pace.17 Such results suggest that changes in power output/pace cannot be used as a reliable predictor of time-to-completion in closed-loop endurance exercise because of adjustments needed to compensate for unpredicted disruptions to physiological homeostasis.

The perception of effort may play a key role in the modulation of exercise intensity,5 10 11 with an expected increases in the rating of perceived exertion (RPE) being contrasted against subjective feelings at any given moment. Disparity between expected and experienced RPE leads to a modification of exercise intensity to restore parity.10 RPE, as measured by the Borg scale,18 increases linearly relative to the percentage of exercise time completed such that the initial rate of increase can accurately predict the endpoint of an exercise bout,19 20 supporting theories of anticipatory pacing strategies. Although a linear relationship between RPE and heart rate (HR) during laboratory-based incremental exercise tests has been observed,21 22 this relationship does not always occur during self-paced exercise. A dissociative relationship was observed during an ultramarathon study in which RPE increased linearly despite declining HR.23

RPE has a psychological component with the evaluation of the remaining task affecting it24 and the propensity either to continue the task or succumb to fatigue.8,,11 Indeed, the perception of effort has been likened to an emotion.5 This view is supported by recent neuro-anatomical models of consciousness in which mental representations of the self at any given moment in time are generated from the integration of all conditions pertinent to homeostasis within the body.25 In cognitive theories, the relationship between an event and an emotional response is mediated by conscious information processing. According to attribution theory26 and appraisal theory,27 28 emotional responses and, in this context, feelings of fatigue, depend partly upon the interpretation of physical sensations. The sensation of fatigue is dependent, therefore, upon the individual's unique interpretation of his or her experience,5 which, in turn, will modify pacing.10

Complex patterns of mood change after an ultramarathon have been recorded, with positive and negative mood changes observed as a result of completing the race.29 30 According to these studies, the emotional response after ultra-endurance exercise is not merely a symptom of the associated physical demands, but is also influenced by the interpretation of performance given the expectations.30 What is not well understood is what kind of cognitions give rise to particular postexercise emotional responses.

The aims of this study were to examine changing patterns of mood and anxiety leading up to an Ironman triathlon, the relationship between RPE and pacing behaviour and the relationship between performance expectations, performance appraisals and postrace mood.



Twelve triathletes (nine men, three women) participated in this study. Their mean (SD) age, height and body mass was 42.5 (3.6) years, 175.9 (7.2) cm and 73.8 (12.6) kg. Participants had been competing in triathlons for 67 (50) months and during the 3 months before the study had completed 8.2 (2.4) training sessions each week on 5.9 (0.7) days per week. The average total training time each week was 13.3 (3.8) h, comprising 2.3 (1.1) h swimming, 7.5 (2.5) h cycling and 3.0 (0.9) h running.

Each participant provided written informed consent to take part in the study. All of the procedures used in this study were in accordance with the Declaration of Helsinki and were approved by the University of Essex ethics committee.


A within-subjects design was used in which repeated measurements of anxiety, mood and perceived exertion were taken for each participant before, during and after competing in the 2008 Ironman Austria Triathlon. The 2008 Ironman Austria Triathlon began and ended in the town of Klagenfurt and consisted of a 3.8 km freshwater lake and river swim, a 180 km moderately undulating two circuit cycling course and a 42.2 km flat two circuit running course.

One week before competing in the triathlon each participant completed a short questionnaire providing details of their training during the preceding 3 months as well as time predictions for the swimming, cycling and running components of their race. Post race participants completed a 20-item Likert questionnaire to measure their performance self-appraisals.

Anxiety measurements

The competitive state anxiety inventory 2 (CSAI-2)31 was completed by each triathlete 1 week before (baseline trial) and again on the morning of the race (prerace trial). The CSAI-2 is a 27-item Likert scale questionnaire comprising three subscales of somatic anxiety, cognitive anxiety and self-confidence.

Mood measurements

Mood state was measured using the ‘right now’ version of the profile of mood states (POMS) questionnaire.32 33 The POMS short form comprises 30 single-word mood descriptors, each with a five-point Likert response scale, from which subscale scores for tension, depression, anger, vigour, fatigue and confusion are calculated. The POMS short version was used to minimise questionnaire fatigue in this study.

Each participant completed a POMS approximately 1 week before (baseline trial), on the morning of (prerace trial) and immediately after the race (postrace trial). POMS raw scores for each subscale were converted into T-scores in accordance with McNair et al (p 36).33 POMS total mood disturbance score was calculated by subtracting the vigour subscale score from the sum of the tension, anger, depression, fatigue and confusion subscale scores.32 33

Perceived exertion

One week before the race participants were provided with the Borg 6-20 RPE scale,18 which they were asked to memorise. During the swim participants memorised an RPE rating at the first and second turn-point buoys (1.5 and 2.4 km), and immediately before getting out of the water (3.8 km). Once participants had completed transition and settled on their bike they recorded their swim RPE scores onto a waterproof recording sheet that had been secured to their bike. Participants also recorded their RPE during the cycle ride at 30, 60, 90, 120, 150 and 180 km using the same sheet. After completing the second transition, RPE scores during the run were recorded for each participant by an experimental observer at 6.2, 16.3, 21.5, 26.7, 32.6 and 36.8 km.

Performance measurements

During the race each participant wore an electronic chip that enabled split times to be recorded as they passed over electronic mats at various points during the race. These official split times were used as the performance measures in this study. A split time was recorded at the end of the swim, at 32, 63, 90, 122, 153 and 180 km during the cycle course, and at 6.2, 12.1, 16.3, 21.5, 26.7, 32.6 and 42.2 km during the run. The average swim, run and cycle speed was calculated using split times and distances covered.

Performance appraisals and attributions

At the end of the race participants completed a 20-item Likert questionnaire, which is presented in figure 1. The questionnaire was designed to measure their appraisals of how well they performed (items 1, 3, 7, 10, 14, 17 and 18), what they attributed their performance to (items 2, 4, 8, 11, 12, 13, 15, 16, 19 and 20) and their level of enjoyment and motivation during the race (items 5, 6 and 9). The questions given were written for this study to gather specific information and we acknowledge that, other than face validity, the reliability and validity of the questionnaire is not established.

Figure 1

Post-race questionnaire designed to measure participants' cognitive appraisals of their performance and of the race conditions.

Statistical analysis

Pearson's product moment correlation tests were used to measure relationships between RPE and various swim, bike, run and overall performance outcomes. Changes in the POMS subscale score were analysed using a within-subjects multivariate analysis of variance with post hoc paired-sample t tests. Changes in CSAI-2 scores were examined using paired-samples t tests. An α level of 0.05 was used to indicate statistical significance and effect sizes are reported as either eta-squared (η2) or partial eta-squared ().


Performance, pacing and RPE outcomes

The average (SD) completion time was 12 h, 39 min, 44 s (1 h, 49 min, 2 s). Average swim, bike, run completion times are provided in figure 2 together with the aggregate transition time and overall completion time. Average (SD) speeds were 3.1 (0.3) km/h for the swim segment, 28.8 (4.0) km/h for the bike segment and 9.2 (1.6) km/h for the run segment.

Figure 2

Average Ironman completion times with each subdiscipline and transitions. Error bars represent ±1 SD. The average aggregate transition time is presented because separate times for the swim-to-bike (T1) and bike-to-run (T2) transitions were not recorded. The aggregate transition time was calculated as the total completion time minus the sum of swim, bike and run completion times.

There were positive correlations between distance covered and RPE during the swim segment (r=0.994, p<0.05), the bike segment (r=0.899, p<0.005) and the run segment (r=0.993, p<0.0001). Distance against RPE correlations are presented separately for the swim, bike and run segments in figure 3. A positive correlation was also found between RPE and the percentage of overall race time completed (r=0.826, p<0.001), which is illustrated in figure 4.

Figure 3

Correlations between distance completed and rating of perceived exertion (RPE) for swimming (A), cycling (B) and running (C) disciplines. Correlation between the distance completed and average speed is also presented for cycling (B) and running (C) but not for swimming because split times were not recorded meaning the average speed could not be calculated.

Figure 4

Rating of perceived exertion (RPE) responses correlated against the percentage of overall race time completed. Y-axis error bars represent ±1 SD of Borg RPE. X-axis error bars represent ±1 SD of the percentage of time overall race time completed when RPE was recorded. As it was not possible to record the split time during the swim the first two percentages (†) were calculated based upon the assumption that the swim speed was constant. Transition times between disciplines are represented by a rectangle box.

Pacing during the bike and run segments are illustrated in figure 3B and 3C. No correlation was found between average speed and distance covered during the cycle segment (r=0.012, p>0.05) but a negative correlation was found between average speed and distance covered during the run segment (r=−0.911, p<0.005). Pacing was not calculated for the swim segment because split times for each participant at the turn buoys were not available.

POMS and CSAI-2 outcomes

Multivariate analysis of variance outcomes indicated a difference in POMS subscales between the UK, prerace and postrace trials, F(12,58)=5.4, p<0.0001, =0.99. Univariate within-subjects analyses of variance indicated trial differences for the POMS subscales of tension, F(2,33)=14.7, p<0.0001, =0.47 and fatigue, F(2,33)=25.1, p<0.0001, =0.6, but not for depression, anger, confusion or vigour. Post hoc paired-samples t tests found higher tension before the race compared with UK baseline, t(11)=−3.7, p<0.005, η2=0.55, and lower tension after the race compared with prerace levels, t(11)=4.6, p<0.001, η2=0.66. Fatigue was higher after the race compared with UK baseline levels, t(11)=−6.3, p<0.0001, η2=0.78 and prerace levels, t(11)=−7.3, p<0.0001, η2=0.83. Changes in POMS profiles are presented in figure 5.

Figure 5

Changes in mood state profile between baseline, prerace and postrace. Post hoc paired samples t test p values are presented. NS indicates a non-significant univariate within-subjects analysis of variance outcome. MANOVA, multivariate analysis of variance; POMS, profile of mood states.

Somatic anxiety was found to be higher before the race compared with baseline measures taken in the UK, t(11)=−1.9, p<0.05, η2=0.25. No differences in cognitive anxiety or self-confidence were found between baseline and prerace scores. Changes in CSAI-2R scores are presented in figure 6.

Figure 6

Changes in competitive state anxiety inventory 2 (CSAI-2R) scores between baseline and prerace trials. Post hoc paired samples t test p values are presented. NS indicates a non-significant univariate within-subjects analysis of variance outcome.

Performance appraisals and attributions

The modal questionnaire responses for each item are highlighted in grey in figure 1. Performance appraisals, performance attributions and levels of motivation and enjoyment were all found to be favourable, as indicated by the calculated questionnaire subscale scores. Subscale scores and the overall questionnaire score are presented in figure 7.

Figure 7

Mean questionnaire outcomes indicating positive performance appraisals, attributions with feelings of enjoyment and motivation. Scores above the broken line indicate positive performance appraisals and attributions, whereas scores below the line indicate negative appraisals and attributions. Error bars represent ±1 SD.


Perception of effort

We found a linear increase in RPE throughout the ultradistance triathlon, similar to a previously reported increase in RPE in an ultradistance running marathon23 but, to our knowledge, this is the first time this has been described in an ultradistance multidiscipline event. We also describe how the RPE follows a more complex pattern whereby RPE follows a ‘dog tooth’ pattern, with each discipline displaying a linear increase in RPE, with a re-setting at the start of the next discipline, although only to a relatively small extent, such that the overall increase for the event is maintained. The reason for this re-setting of RPE between disciplines is not clear from our study. One possible explanation could relate to the time spent in transition between disciplines, during which athletes change clothing and equipment before proceeding to the next discipline. While no single physiological variable has been identified that consistently relates to perceived exertion,34 the time spent in transition is effectively a recovery period and may allow for a reduction in the integrated afferent signals that may contribute to the perception of effort. The time spent in transition is relatively short and it is likely that physiological mediators such as HR will return to pretransition levels soon after starting the next discipline, so this is unlikely to be the full explanation for the minor re-setting. Although not reported, HR data for one of the participants showed that post-transition values returned to and exceeded the HR immediately before the corresponding transition within a few minutes, in contrast to the RPE, which took hours to return to pretransition values. HR during an ultradistance triathlon has previously been shown to return very quickly to pretransition values when moving from one discipline to the next.35

Another possible explanation is that the changes in mode of exercise during a triathlon necessitate a change in the muscle groups that are predominantly used in locomotion in each discipline36 so that progressive local muscle fatigue during each discipline may account for some of the minor re-setting in RPE, as previously relatively unused, and therefore relatively unfatigued, muscle groups are utilised when switching to the next discipline. Further studies are required to investigate which physiological mediators contribute to RPE.


Whereas the within-discipline RPE displayed a linear upward progression for each of the swim, cycle and run, the pattern of pacing during the cycle showed significant variation, largely due to the terrain of the course. The cycle discipline covered a two-lap course and no significant relationship between distance covered and pacing was found, indicating a consistent pace. During the run, the pace gradually declines in a linear fashion. In neither the bike nor the run was there any evidence of a sudden failure in the ability to maintain pace. Rather, during the run in particular, there appears to be a continuous process of pace reduction. It is not possible from our data to suggest whether this reduction in pace is solely the result of reduced power output from working muscles, or is also due to reduced biomechanical efficiency associated with the loss of ability to maintain a good running technique.37 It may be possible to measure stride rate and length in future studies to examine this variable. Our findings are consistent with a recent model proposed by Tucker and Noakes38 in which the rate of increase in RPE is compared with the individual's maximum tolerable RPE and the anticipated remaining duration of the exercise bout, resulting in changes to power output and therefore pace, in order to keep the projected final RPE at a tolerable level. Indeed, the remarkably linear increase in RPE and reduction in pace during the run would suggest that a process of regular assessment has taken place, assuming that pacing was subject to some level of conscious control, and that the athletes were not simply powerless to prevent the gradual decline in pace due to fatigue processes causing a decline in the functional capacity of working muscles. Previous studies have reported that reductions in power output and pace during endurance exercise are associated with a reduction in ‘central drive’39 40 indicating that reductions in pace are, at least to some extent, due to conscious or subconscious processes.

The seemingly gradual decrease in pace during the run is consistent with the idea that RPE is the main mediator that is used by the athletes to control their pace. This may be because of the nature of the ultradistance triathlon and the standard of athlete that we studied, in which the athletes are focused upon completing the race rather than pursuing a particular finishing time. Micklewright et al8 have previously demonstrated that athletes are prepared to ignore their conscious perception of effort and risk premature fatigue in attempting to pursue a performance level based upon their expectations and previous experience. The mean completion time of the athletes in the current study was faster than the mean anticipated finishing time based upon the predictions provided by each athlete, and this may explain, to some extent, why the athletes seem prepared so readily to accept a decline in pace during the run discipline, as they were exceeding their expectations, although actual run completion times were also faster than the athletes' own expectations. It could, of course, be possible that athletes were being conservative in their predictions, but the positive results from the postrace appraisal and attribution questionnaire suggest that athletes did indeed perform well relative to their expectations. Further studies with faster athletes pursuing expected completion times more closely matched to their actual performance times may shed further light on the relationship between performance expectations, perceived exertion and pacing.

Mood and anxiety changes

Prerace, participants reported an increase in feelings of tension and somatic anxiety compared with baseline, demonstrating an anticipatory affective state. Consistent with the cognitive perspective26,,28 these prerace emotional changes are likely to be due to conscious thoughts about the task ahead and the likelihood of success or failure given the significant levels of preparation undertaken. The modal responses to the questionnaire presented in figure 1 indicate positive appraisals and attributions among participants and high levels of enjoyment. These positive cognitions among participants also coincide with postrace feelings of tension that were lower than their baseline measurements. As predicted, this is an indication that the emotional response of athletes after ultra-endurance exercise is more aligned with their conscious thoughts30 than their physical state.


The mechanisms involved in setting and modulating pace during an ultradistance triathlon were examined during this study. The linear progression of RPE for the event and within individual disciplines lends support to the theories in which conscious awareness of the intensity of exercise is used to modulate the pace, while the ‘dog tooth’ pattern of RPE progression for the entire event suggests a re-assessment of coping resources when progressing from one discipline to the next.

What is already known on this topic

  • RPE is known to influence pacing strategies.

  • The rate of RPE increase can be used to predict exercise duration.

What this study adds

  • How multidiscipline events lead to a pattern of RPE increase.

  • RPE displays an overall linear increase for a multidiscipline event, which is characterised by minor re-setting between disciplines implying a cognitive pacing strategy for the entire event and for each component discipline.


View Abstract


  • Competing interests None.

  • Patient consent Obtained.

  • Ethics approval All of the procedures used in this study were conducted in accordance with the Declaration of Helsinki and were approved by the University of Essex ethics committee.

  • Provenance and peer review Not commissioned; externally peer reviewed.