A test-retest assessment of the effects of mental load on ratings of affect, arousal and perceived exertion during submaximal cycling

ABSTRACT

This study aimed to test the effects of mental (i.e. executive) load during a dual physical-mental task on ratings of perceived exertion (RPE), affective valence, and arousal. The protocol included two dual tasks with matched physical demands but different executive demands (2-back and oddball), carried out on different days. The procedure was run twice to assess the sensitivity and stability of RPE, valence and arousal across the two trials. Linear mixed-effects analyses showed less positive valence (−0.44 points on average in a 1–9 scale; R_β² = 0.074 [CI90%, 0.052–0.098]), and heightened arousal (+0.13 points on average in a 1–9 scale; R_β² = 0.006 [CI90%, 0.001–0.015]), for the high executive load condition, but showed no effect of mental load on RPE. Separated analyses for the two task trials yielded best-fitting models that were identical across trials for RPE and valence, but not for arousal. Model fitting was improved by assuming a 1-level autoregressive covariance structure for all analyses. In conclusion, executive load during a dual physical-mental task modulates the emotional response to effort, but not RPE. The autoregressive covariance suggests that people tend to anchor estimates on prior ones, which imposes certain limits on scales' usability.

KEYWORDS:

• Arousal
• perceived exertion
• mental workload
• affective valence
• Effort assessment
• linear mixed-effects modelling

Disclosure statement

No potential conflict of interest was reported by the authors.

Supplemental Material

Supplemental data for this article can be accessed here.

Notes

¹ As later described in detail, HRR was used to set physical load constraints constant across sessions and participants. In spite of its potential limitations, this method adequately controls for within-participant load variations (Borresen & Lambert, 2009), which was complementarily checked with power output measures. Possible variations in actual physical load across participants were not problematic in the present design, first, because all theoretically and practically relevant effects are within-participant; and, second, because variations across participants are explicitly taken into account in linear mixed-effects models (as the participant is included in the model as a random factor). The same case can be made about other non-controlled sources of between-participant variability.

² For the model including the effect of task type upon the best model, AIC = 3576.759, L.Ratio = 1.444, p = 0.230. Following Wagenmakers (2007), the approximation to a Bayes Factor between these two models yielded BF₀₁ = 20.430, which can be interpreted as strong support for the non-existence of a task type effect on RPE.