The effects of acute stress on the calibration of persistence

Abstract

People frequently fail to wait for delayed rewards after choosing them. These preference reversals are sometimes thought to reflect self-control failure. Other times, however, continuing to wait for a delayed reward may be counterproductive (e.g., when reward timing uncertainty is high). Research has demonstrated that people can calibrate how long to wait for rewards in a given environment. Thus, the role of self-control might be to integrate information about the environment to flexibly adapt behavior, not merely to promote waiting. Here we tested effects of acute stress, which has been shown to tax control processes, on persistence, and the calibration of persistence, in young adult human participants. Half the participants (n = 60) performed a task in which persistence was optimal, and the other half (n = 60) performed a task in which it was optimal to quit waiting for reward soon after each trial began. Each participant completed the task either after cold pressor stress or no stress. Stress did not influence persistence or optimal calibration of persistence. Nevertheless, an exploratory analysis revealed an “inverted-U” relationship between cortisol increase and performance in the stress groups, suggesting that choosing the adaptive waiting policy may be facilitated with some stress and impaired with severe stress.

Willingness to wait covaries with endogenous variation in cortisol

“…We observed that the smaller were changes in hair cortisol level between the months of the study and higher were the life change units (from the questionnaire) the less subjects were choosing the later option. Among curvilinear relationships we did not confirm previous results about subjects’ earnings (Lempert et al, 2018a), but found some evidence for the reaction times: when chronic stress level increased slightly, it took longer for participants to decide in the short delay task, but when the cortisol measure from hair samples increased to higher levels, they decided quicker whether to pick a sooner or later option. Nevertheless, tests for curvilinear relationships included only a small portion of participants, those for whom there was an increase in biological stress level compared to the baseline.…”

“…According to the literature, we expected significant medium level correlation between hair and saliva measures around r = 0.4 (Vanaelst et al, 2012; van Holland et al, 2012; Zhang et al, 2018; Weckesser et al, 2019) and low level to no correlation between biological measures and questionnaires (Vanaelst et al, 2012; Stalder et al, 2017; Prado-Gascó et al, 2019). Some scholars suggest log-transformation of cortisol values to better approximate a Gaussian distribution of cortisol values across subjects (Lenow et al, 2017; Lempert et al, 2018a). Indeed, we decided it is reasonable to modify all of our cortisol values based on the distributions before and after log-transformation (SI, Figure S5).…”

“…Previous studies have shown that the magnitude of the physiological response to stress can predict behavior in PFC-dependent tasks, importantly, with both linear (Goldfarb et al, 2017; Maier et al, 2015) and curvilinear (Luksys and Sandi, 2011; Lempert et al, 2018a) relationships being possible. To that end, we explored several other per task / session measures, including subjects’ profit, total waiting time (wait), proportion of later choices (PL, or lottery choices for risk task) and reaction time (RT) in relation to stress levels and their change.…”

“…We did not stress our subjects and only looked at changes in cortisol levels compared to baseline session or month. Still, following Lempert et al (2018a) we checked whether we can find any curvilinear relationships for those with increase in biological stress level compared to the baseline. There were only 13 subjects with such positive changes from lh1 to lh2 (Δlh2), 16 subjects from lh1 to lh3 (Δlh3) and 17 subjects from lh2 to lh3 (Δlh32).…”

“…We did not stress our subjects and only looked at changes in cortisol levels compared to baseline session or month. Still, following Lempert et al (2018a) we checked whether we can find any curvilinear relationships for 12/23 (which was not certified by peer review) is the author/funder. All rights reserved.…”