Acute stress can change our cognition and emotions, but what specific consequences this has for human prosocial behaviour is unclear. Previous studies have mainly investigated prosociality with financial transfers in economic games and produced conflicting results. Yet a core feature of many types of prosocial behaviour is that they are effortful. We therefore examined how acute stress changes our willingness to exert effort that benefits others. Healthy male participants – half of whom were put under acute stress – made decisions whether to exert physical effort to gain money for themselves or another person. With this design, we could independently assess the effects of acute stress on prosocial, compared to self-benefitting, effortful behaviour. Compared to controls (n=45), participants in the stress group (n=46) chose to exert effort more often for self- than for other- benefitting rewards at a low level of effort. Additionally, the adverse effects of stress on prosocial effort were particularly pronounced in more selfish participants. Neuroimaging combined with computational modelling revealed a putative neural mechanism underlying these effects: more stressed participants showed increased activation to subjective value in the dorsal anterior cingulate cortex and anterior insula when they themselves could benefit from their exerted effort, relative to when someone else could. By using an effort-based task that better approximates real-life prosocial behaviour and incorporating trait differences in prosocial tendencies, our study provides important insights into how acute stress affects prosociality and its associated neural mechanisms.
This study reports valuable findings on the influence of acute stress on prosocial behavior and its neural correlates. Though the evidence is generally solid, concerns about confounding and sampling bias detract from the potential impact.
A perceived lack of resources to deal with the demands of the environment, such as during a tough job interview or a busy day at work, leads to the experience of acute stress. This can have profound effects on cognition and physiology with potential consequences for the way we behave towards others (Faber & Häusser, 2022; von Dawans et al., 2021). However, previous work investigating prosocial decision making under acute stress has yielded mixed findings. Some studies suggest that acute stress can increase prosocial tendencies (Tomova et al., 2017; von Dawans et al., 2012), whereas others have demonstrated decreases (Sollberger et al., 2016; Vinkers et al., 2013) or no effects at all (for a recent meta-analysis, see Nitschke, Forbes, et al., 2022; Veszteg et al., 2021). Several mechanisms have been proposed as to how acute stress leads to increases in prosocial behaviour including an increased affiliative drive to seek social support from others: the ‘tend and befriend’ response (Taylor et al., 2000). Alternatively, decreased prosociality under stress could result from an increased self-focus (Rimmele & Lobmaier, 2012; Tomova et al., 2014) and/or changes in reward sensitivity (Berghorst et al., 2013).
Differences in prosocial decision making under stress have typically been found in economic games requiring financial transfers to others, such as the dictator game or trust game (Camerer, 2003). These involve decisions about how to share a financial endowment with others but they do not involve much effort. Yet, prosocial behaviour in everyday life—from filling in a form to donate to charity to helping a colleague at work—requires the investment of differing amounts of effort (Lockwood et al., 2017, 2021, 2022). It is unclear how stress impacts this type of prosocial behaviour. Additionally, we do not know whether changes in sharing financial endowments reflect genuine differences in prosocial motivation under stress or differences in how stressed individuals value their own monetary gains (Berghorst et al., 2013; Carvalheiro et al., 2021; Tomova et al., 2020). Therefore, we directly compared the effects of acute stress on effortful prosocial behaviour compared to effortful self-benefitting behaviour. In the task, participants had to exert actual physical effort to obtain rewards either for themselves or another person. We then used computational modelling to investigate the neural mechanisms responsible for the potential stress-induced changes in effortful prosocial behaviour. Finally, we tested whether individual differences in existing prosocial tendencies modulated these effects.
Acute stress may reduce effortful prosocial behaviour via its more general effects on motivation and effort exertion. In rodents, acute stress results in a reduced willingness to exert effort for rewards without changing preferences for these rewards themselves (Shafiei et al., 2012); these effects are dependent on the hypothalamic–pituitary–adrenal (HPA) axis (Bryce & Floresco, 2016). In humans, acute stress also leads to an increased avoidance of tasks involving cognitive (Bogdanov et al., 2021) and physical (Voulgaropoulou et al., 2022) effort. However, previous work has focused solely on effort-based decisions for self-relevant outcomes. It is not known how acute stress affects effort-based decisions when someone else benefits from our effort, nor is it clear how acute stress affects the neurocomputational mechanisms of effort-based decision-making either for ourselves or for others.
We examined these mechanisms by focussing on how the computationally-derived subjective values of rewards are altered by changing the effort costs required to obtain them (Bartra et al., 2013; Chong et al., 2017; Lockwood et al., 2017, 2021, 2022). Subjective value decreases as the effort costs required to obtain the reward increase. By using computational modelling, we could extract participant-specific effort discounting parameters (K) which represented the extent to which each participant devalued (or ‘discounted’) rewards by effort (Lockwood et al., 2017). That is, the discount parameters represent how much the subjective value of a reward decreases when a certain amount of effort has to be invested to obtain it. We used functional magnetic resonance imaging (fMRI) to establish brain regions where blood oxygen level–dependent (BOLD) activation responded to subjective value (Lockwood et al., 2022) and how this activation was affected by acute stress. We had a particular focus on the dorsal anterior cingulate cortex (dACC) and anterior insula (AI), as these areas are strongly implicated in effort-based decision making (for a meta-analysis, see Lopez-Gamundi et al., 2021; Pessiglione et al., 2018). For instance, Lockwood et al. (2022) found responses in these areas to subjective value on both self- and other-benefitting trials when participants were deciding whether to exert effort. In addition, both areas are sensitive to the effects of stress (Berretz et al., 2021; Cerqueira et al., 2007; King et al., 2009; Kogler et al., 2015; Radley et al., 2005). Thus, we hypothesised that if acute stress results in reduced effort-related prosocial behaviour (e.g. Sollberger et al., 2016; Vinkers et al., 2013), it may do so via a preferential response of these areas to subjective value related to self-compared to other-benefitting decisions in stressed participants. Conversely, if acute stress leads to increased effort-related prosocial behaviour (e.g. Tomova et al., 2017; von Dawans et al., 2012), dACC and AI should respond more to subjective value for other- compared to self-benefitting rewards in stressed participants.
The willingness to engage in prosocial behaviour shows substantial individual differences (Murphy et al., 2011; Thielmann et al., 2020) and may account for previous inconsistencies in terms of how acute stress affects social behavior (Nitschke, Forbes, et al., 2022). For example, Azulay et al. (2022) showed that elevations in cortisol following stress induction resulted in greater generosity in a one-shot dictator game in participants scoring high on trait empathy, with such cortisol increases resulting in reduced generosity in those scoring low on empathy. Similarly, Speer et al. (Speer et al., 2022) showed that acute stress can accentuate existing differences in dishonesty (cf. Schulreich et al., 2022; Ying et al., 2022). Thus, we aimed to test whether individual differences in participants’ trait prosocial tendencies, as measured by Social Value Orientation (SVO; Murphy et al., 2011), which quantifies participants’ tendency to distribute resources between themselves and another person, modulate the impact of acute stress on effortful prosocial behaviour. Specifically, we tested the hypothesis that individualistic participants (i.e., those who are more selfish) would show the most marked changes in effortful prosocial behaviour following acute stress.
Ninety-six male participants were randomly assigned to undergo an established stress induction protocol or a control task (Dedovic et al., 2005; Tomova et al., 2017). Saliva samples and perceived stress ratings were collected as indicators of physiological and subjective stress responses throughout the experiment. Participants made incentivised choices in an fMRI scanner between ‘work’ (exerting effort) or ‘rest’ (not exerting effort) (Figure 1; 10–12). If they chose ‘rest’, participants waited until the end of the trial and received one credit; if they chose ‘work’, then they could receive a larger reward (2, 4, 6, 8, or 10 credits; converted to money at the end of the experiment). To receive this larger reward, participants had to squeeze a hand dynamometer for at least one second at the required effort level (30, 40, 50, 60, or 70% of their maximum voluntary contraction; MVC) during a three second window. To establish whether stress affected effort-based decision making differently for self-benefitting rewards compared to other-benefitting rewards, on half the trials, participants themselves were the recipients of the credits (self trials), whereas on the other half of trials participants were told the next participant in the study would receive the credits (other trials). We hypothesised that acute stress would decrease participants’ willingness to exert other-benefitting prosocial effort compared to self-benefitting effort. Moreover, using neuroimaging combined with computational modelling, we examined whether the effects of stress were related to changes in subjective value representations in dACC and AI on self-compared to other-benefitting trials.
Greater cortisol and perceived stress following stress induction
Participants rated their perceived stress at eight timepoints distributed across the experiment. We conducted a repeated measures ANOVA with group as a between subject-factor and the eight sample timepoints as a within-subject factor. This revealed a significant interaction between group and sample timepoint, F(7,644) = 19.05, P < 0.001, η2 = 0.101, showing that the increase in stress over time was greater in the stress group than in the control group (Figure 2). Follow up tests revealed a significant difference between the groups at timepoints 2 through to 7 (all Ps < 0.04) but not at timepoints 1 or 8 (all P values were Bonferroni corrected for multiple comparisons). For salivary cortisol (Figure 2), we also found a significant interaction between group x sample timepoint, F(5,445) = 6.945, P < 0.001, η2 = 0.028. Follow up tests revealed a significant difference between the groups at timepoints 3, 4 and 5 (all Ps < 0.001) but not at timepoints 1, 2 or 6 (all P values were Bonferroni corrected). To create a measure of participants’ total stress response throughout the experiment, we calculated the area under the curve (AUC) with respect to ground for participants’ stress ratings and salivary cortisol (Pruessner et al., 2003).
Acute stress reduces effortful prosocial behaviour at a low level of effort
To determine differences in choices to engage in effortful behaviour, we conducted a mixed effects logistic regression using the glmer function from the R package lme4 (Bates et al., 2015). Participants’ choices were entered as the dependent variable (work=1; rest=0), with the factors Group (stress, control), Recipient (self, other), Effort (5 levels), and Reward (5 Levels) as fixed effects. As random effects, we included random intercepts for participant and a random slope for Recipient (Barr et al., 2013). Models with a more complex random effects structure (i.e., with Reward or Effort as random slopes) were singular or did not converge (despite using BOBYQA optimisation). A model containing all the three-way interactions between the fixed effects was compared to a more complex model containing the four-way interaction using the anova function in R. The more complex model did not provide a better fit to the data (AIC: 7109.5 [four-way interaction model] vs. 7094.1 [three-way interactions model]; BIC: 7885.6 vs. 7749.6; P=0.413). For the more parsimonious three-way interaction model, there was a significant interaction between Group, Recipient, and Effort (Type III Wald test, χ2=21.48, P<0.001). We ran follow up tests using the emmeans package in R (Lenth, 2022). Here, we tested the interaction between Group and Recipient at each level of effort, while keeping the reward level constant (average reward level). We found a significant Group x Recipient interaction at effort level 2 (P<0.05: Bonferroni corrected for multiple comparisons). Similarly, in an exploratory follow up analysis, when we contrasted the two lower effort levels (effort levels 1 and 2) with the two higher effort levels (effort levels 4 and 5), we found an interaction between Group and Recipient at the combined lower effort levels (P=0.035; Bonferroni corrected) but not at the combined higher effort levels (P=1.00).
This pattern of results was driven by the fact that whereas participants in the control group did not show a significant difference in the likelihood of deciding to exert effort on self trials compared to other trials at effort level 1 (P=0.595) or 2 (P=0.882), participants in the stress group favoured putting in effort on self trials compared to other trials even at effort level 1 (P<0.001) and 2 (P<0.001). Thus, both our planned and exploratory analysis suggest that unlike participants in the control group, those in the stress group were less willing to put in effort to gain rewards for another person relative to themselves at a lower level of effort. The estimated marginal means for the three-way interaction between group, recipient and effort are shown in Figure 3.
There was a significant interaction between Group and Effort (Type III Wald test, χ2=17.27, P=0.002), but follow up tests did not reveal any significant differences at any effort level between the groups (all Ps > 0.26 Bonferroni corrected for multiple comparisons). There was a significant interaction between Recipient and Reward (Type III Wald test, χ2=19.12, P<0.001) and also between Recipient, Effort, and Reward (Type III Wald test, χ2=50.65, P<0.001) showing that the effect of Effort on Reward was different on self- compared to other- benefitting trials. The three-way interactions between Group x Recipient x Reward (P=0.285) and Group x Effort x Reward (P=0.064) were not significant (see Table S3 and S4 for the full model output of the choice data).
To analyse the force data (i.e. how much effort participants actually exerted once they decided to do so), we ran a mixed effects model using the lmer function in R (Bates et al., 2015). The outcome variable was defined as the area under the curve during the duration of the force period relative to participants’ MVC (i.e. the cumulated effort exerted over time). The model contained the same fixed effects as above and again random intercepts for participant and a random slope for Recipient. A Type III Wald χ2 test revealed neither a main effect of Group nor any significant interaction involving Group (see Table S5 for the full model output) but a significant three-way interaction between Recipient, Effort and Reward (χ2=46.56, P<0.001) and a two-way interaction between Effort and Reward (χ2=67.92, P<0.001).
Finally, if participants chose to put in effort, there were no significant differences between the groups regarding how often they successfully squeezed to the required effort level and received the reward (control group: mean success rate = 98.73%, SD = 2.19%; stress group: 97.66%, SD = 3.52%).
To summarise, at a lower level of effort, participants in the stress group showed a preference for exerting effort for self- compared to other-benefitting rewards. The control group, in contrast, shown no difference in their willingness to exert effort for a reward on self- compared to other-benefitting trials at this level of effort. There were no group differences in the amount of force exerted by the participants on each trial nor in how successful participants were in squeezing to the required effort level. This suggests that acute stress did not impact participants’ ability to perform the effort task but that it did affect their decisions to put in effort for another person’s benefit.
Social value orientation modulates the impact of perceived stress on effortful prosocial behaviour
To test the hypothesis whether more individualistic participants become even more selfish under stress, we first calculated the proportion of effortful prosocial choices participants made relative to the total number of effortful choices they made (% prosocial choices). In other words, this measured how often participants chose to put in effort (i.e. ‘work’) for the other person and thereby gain a higher reward for the other person relative to the total number of times they chose to put in effort to gain a higher reward (i.e., both for themselves and the other person). This enabled us to determine each participant’s prosocial effort relative to any general stress-induced decline in effortful behaviour.
First, we tested whether SVO angle and Group (i.e., stress vs. control) interacted to influence the proportion of prosocial choices using linear regression in R but did not find a significant interaction (B=0.164, SE=0.181, P=0.368). Despite average differences between the groups in AUC for perceived stress and cortisol, there was considerable heterogeneity in the way participants responded to the stress induction. Thus, using linear regression in R, we investigated whether AUC for perceived stress (z scored across participants), which captured the total stress response of each participant in both the stress group and the control group throughout the experiment, interacted with SVO angle to influence the proportion of prosocial choices.
Here, we found a significant interaction between SVO angle (range=0-45°) and perceived stress (B=0.246, SE=0.103, P=0.020). Simple slopes analysis revealed that those with a more individualistic SVO angle (-1SD; 10.92°) showed a significant decline in the proportion of prosocial choices at increasing levels of perceived stress (B=-2.97, SE=1.08, P=0.008), but this was not the case for participants with a more prosocial SVO angle (+1SD; 27.64°; B=1.14, SE=1.18, P=0.337). This shows that more individualistic individuals became made relatively more self-benefitting (‘selfish’) decisions at increasing levels of perceived stress (Figure 4). Cortisol (AUC) did not modulate the effect of SVO angle on the proportion of prosocial choices (B=0.069, SE=0.093, P=0.459).
Perceived stress modulates AI and dACC responses to SVself relative to SVother during effort-based decisions
To identify the neural mechanisms underlying the effects of stress on prosocial behaviour, we analysed brain areas which during the offer period responded to the subjective value (SV) of the chosen option relative to the non-chosen option (i.e., work vs. rest) on each trial. We focused our analysis on the AI and dACC, using anatomical masks independently determined by Lockwood et al. (2022). Both areas were of strong a priori interest as they have been reliably implicated in effort-based decision making (Chong et al., 2017; Croxson et al., 2009; Engström et al., 2015; Lockwood et al., 2022; Prévost et al., 2010) and show a consistent sensitivity to stress (Åhs et al., 2006; Berretz et al., 2021; Cerqueira et al., 2007; Dedovic et al., 2009; Gathmann et al., 2014; King et al., 2009; Kogler et al., 2015; Morgado et al., 2015; Pruessner et al., 2008; Radley et al., 2005; Starcke & Brand, 2012; Wang et al., 2005).
To calculate the subjective value for each participant on each trial, we used an established model which has consistently shown to best characterise participants’ choices on this task (Lockwood et al., 2017, 2021, 2022). The model contains two separate parabolic effort discounting parameters (Kself, Kother) and one temperature parameter (β):
Subjective value = Reward - (Discount x Effort2)
Discount parameter on self trials = Kself
Discount parameter on other trials = Kother.
Maximum likelihood estimation and a model comparison approach confirmed that this was the best fitting model - in terms of the lowest Bayesian Information Criterion (see Figure S1). We further validated this winning model by performing parameter recovery (See Table S1). Each participant’s Kself and Kother values were used to calculate the SV of the chosen option relative to the non-chosen option (i.e., work vs. rest) on each trial using the winning model.
These values were then used during the first level analysis as a parametric modulator to determine which regions scaled with SV for each participant on self trials (SVself) and other trials (SVother). Using each participant’s contrast images for SVself and SVother, we extracted the parameter estimates from the regions of interest (ROI) in the AI and dACC, and analysed them using linear mixed effect models with the lmer function in R. The anticipated Group (stress vs. control) by Recipient (self vs. other) interaction was not significant in either ROI (dACC: B=-0.242, SE=0.486, P=0.619; AI: B=-0.719, SE=0.498, P=0.152). As above, and in light of the heterogeneity of responses to the stress induction, we included perceived stress (AUC) as a between-subject predictor and Recipient as a within-subject predictor. This revealed a significant interaction between perceived stress and Recipient in both the dACC (B=-0.566, SE=0.237, P=0.019) and AI (B=-0.547, SE=0.247, P=0.029), shown in Figure 5. Simple slopes analyses revealed that in the dACC there was a negative association between perceived stress and SVother (B=-0.413, SE=0.176, P=0.020) but not between perceived stress and SVself (B=0.153, SE=0.176, P=0.386). Thus, the results suggest that greater perceived stress perturbs responses to SVother in the dACC. Simple slopes analysis in the AI did not reveal any significant associations (all Ps > 0.096). Cortisol (AUC) did not modulate the effect of Recipient in either ROI (dACC: B=-0.342, SE=0.250, P=0.174; AI: B=- 0.225, SE=0.257, P=0.383).
To complement the ROI analyses, we conducted whole brain analyses with a statistical threshold of p<0.05 family-wise error (FWE) corrected at the cluster level with a cluster defining threshold of p<0.001 across the whole brain. For each participant, we created contrast images for SVself – SVother (1 -1) and SVother – SVself (-1 1). These contrast images were then used in one sample t-tests with Perceived Stress as a covariate. This revealed several complementary regions which responded preferentially to SVself compared to SVother at increasing levels of perceived stress, including the dorsolateral prefrontal cortex (dlPFC; middle frontal gyrus: x = -46, y = 20, z = 40; Table 1).
As with the behavioural analysis, we investigated whether SVO angle modulated the impact of perceived stress on responses to SVself and SVother in the dACC and AI. We conducted a linear mixed effect model with SVO angle, Perceived Stress, and Recipient as predictors and the parameters estimates as the dependent variable. In the dACC, there was a significant interaction between Perceived Stress and SVO angle (B=-0.062, SE=0.024, P=0.011). Simple slopes analyses revealed that those with a more prosocial SVO (+1SD; 27.35°) showed a significant decline in activation to SV at increasing levels of perceived stress (B=-0.991, SE=0.267, P<0.001), but this was not the case for participants with a more individualistic SVO (- 1SD; 11.00°; B=0.016, SE=0.252, P=0.949). In the dACC, there was also a significant interaction between SVO angle and Recipient (B=-0.057, SE=0.027, P=0.039). Simple slopes analyses revealed a positive association between SVO angle and SVother (B=0.041, SE=0.021, P=0.046) but not SVself (B=-0.016, SE=0.021, P=0.440). There was no three-way interaction between SVO angle, Perceived Stress, and Recipient in the dACC (B=0.035, SE=0.032, P=0.273). In the AI, there was no significant interactions involving SVO angle (all Ps>0.089).
Much of our prosocial behaviour entails a willingness to exert effort. We investigated how acute stress–a prevalent everyday occurrence–impacts effortful prosocial behaviour and report three important findings. First, prosocial behaviour under stress was dependent on the amount of effort involved. Compared to participants in the control group, those in the stress group were less willing to exert effort for other-benefitting rewards at a relatively low level of effort. Secondly, by using neuroimaging combined with computational modelling, we found that participants who perceived more stress showed increased activation in dACC and AI to subjective value (SV) on self- relative to other-benefitting trials. Thirdly, the detrimental effects of acute stress on effortful prosocial behaviour were especially marked in more selfish participants, who also showed differences in dACC activation. This reveals how individual differences in prosocial tendencies can shape prosocial behaviour under acute stress.
Our task separated the effect of acute stress on participants’ willingness to exert effort for self-benefitting, compared to other-benefitting, rewards (Lockwood et al., 2017, 2021, 2022). Thus, changes in prosocial behaviour were not only the result of general changes in motivation to exert effort following stress (Bogdanov et al., 2021; Bryce & Floresco, 2016; Shafiei et al., 2012). Moreover, in previous studies using economic games in which participants could make financial transfers to others (for a meta-analysis, see Nitschke, Forbes, et al., 2022), it was not clear whether differences in prosocial behaviour were the result of changes in reward sensitivity (e.g. how participants value their own monetary gains following stress; Berghorst et al., 2013), or the result of actual changes in prosocial motivation. As we manipulated effort costs, rather than financial costs, we showed that acute stress specifically changed participants’ motivation to help others when the effort costs involved in gaining rewards for others were relatively low (i.e. effort level 2; 40% of participant’s MVC). Increased or unchanged prosocial behaviour under stress could therefore be restricted to situations in which effort costs are absent or when participants are helping close others (Margittai et al., 2015, 2018).
A potential explanation for why we found group differences in other-benefitting relative to self-benefitting effortful behaviour at a lower level of effort and not at higher levels, is that acute stress also had a general effect on participants’ willingness to exert effort as shown by the presence of a Group x Effort interaction (see also Figure 3). At higher effort levels, any effects of recipient (i.e. self vs. other) may have been masked by a more general effect of acute stress on participants’ willingness to exert effort. Thus, future studies investigating differences in prosocial motivation under stress will need to separate general reductions in motivation from specific differences in prosocial motivation. One way to achieve this could be to investigate lower effort levels in a more fine-grained manner.
We used a computational modelling approach combined with neuroimaging to provide insight into potential neural mechanisms underlying the behavioural results. At increasing levels of perceived stress, the dACC and AI showed greater activation to SV on self-benefitting (SVself) relative to other-benefitting trials (SVother). In the dACC, this difference was driven by reduced activation to SVother at increasing levels of perceived stress. Both the dACC and AI have been consistently implicated in effort-based decision making (Chong et al., 2017; Croxson et al., 2009; Engström et al., 2015; Lockwood et al., 2022; Prévost et al., 2010; Rudebeck et al., 2006; Walton et al., 2003) and are sensitive to stress-effects (Berretz et al., 2021; Gathmann et al., 2014; Kogler et al., 2015; Morgado et al., 2015; Starcke & Brand, 2012). The dACC is thought to track effort costs in a domain general manner - regardless of whether effort costs are physical or cognitive (Chong et al., 2017), or whether effort is being exerted for one’s own or someone else’s benefit (Contreras-Huerta et al., 2020; Lockwood et al., 2022). Thus, reduced tracking of SV in the dACC on other-benefitting trials under heightened stress could reflect the reduced priority placed on other people’s gains under when stressed.
It is important to acknowledge that we did not find the anticipated Group by Recipient interaction in either the AI or dACC; the effect of Recipient on parameter estimates was only modulated by perceived stress (AUC) – the amount of stress participants reported across the experiment. Similarly, we did not see the anticipated Group by SVO angle interaction for the proportion of prosocial choices, but only an interaction between perceived stress and SVO angle. This absence of a group effect may have resulted from several participants in the stress group not responding to the stressor and, conversely, those in the control group being stressed, for example, by the scanning environment itself (Noack et al., 2019). Heterogeneity in stress responses is a feature of all studies involving acute stress, as no stress induction protocol is completely effective, particularly within a scanning environment where successful stress induction (as measured by cortisol reactivity) ranges from 47.5% and 65.0% for the MIST (Noack et al., 2019).
One potential mechanism underlying a shift in prosocial tendencies under greater perceived stress could be a downregulation of the brain’s ‘executive control network’ (Hermans et al., 2014). The dlPFC is a key component of this network and, in our complementary whole brain analysis, this area showed reduced activation to SVother relative to SVself at increasing levels of perceived stress. Similar changes in dlPFC activation following stress have been reported during working memory tasks (Qin et al., 2009). Three meta-analyses (Bellucci et al., 2020; Cutler & Campbell-Meiklejohn, 2019; Rhoads et al., 2021) have highlighted the dlPFC as a key region involved in prosocial behaviour (for an overview, see Lamm & Forbes, in press) potentially due to its role in inhibiting selfish responses (Feng et al., 2015; Strang et al., 2015). An interesting avenue for future work is to test whether individual differences in executive functions, such as inhibition, predict changes in social behaviour following acute stress. This would be analogous to the finding in non-social domains, where greater working memory capacity protects against stress-induced changes in learning (Otto et al., 2013).
Individual differences in prosocial traits as measured by Social Value Orientation (Murphy et al., 2011) modulated the effect of perceived stress on effortful prosocial behaviour. Participants with a more individualistic SVO chose to exert effort more often for rewards for themselves compared to rewards for others at increasing levels of perceived stress. This relationship was not seen in individuals with a more prosocial SVO. Similar results have been reported when participants shared money with others under time constraints, whereby time pressure exaggerates existing prosocial tendencies (Chen & Krajbich, 2018) (but see Bouwmeester et al., 2017). Our findings lend further support to models of acute stress that emphasise a shift to automatic or habitual responding (Hermans et al., 2014; Starcke & Brand, 2012) and suggest that such general shifts in cognitive processing also apply within the social domain. The specific consequences these shifts to automatic processing have for social behaviour is likely to be person- and context-specific and understanding these factors should be a key aim for future work (Nitschke, Forbes, et al., 2022).
Our study extends previous findings demonstrating the importance of individual differences in empathy (Azulay et al., 2022), mentalising (Schulreich et al., 2022), and Social Value Orientation (Ying et al., 2022) in predicting prosocial responses under acute stress. However, it is not yet clear when acute stress accentuates or supresses existing prosocial tendencies. For example, Schulreich et al (2022) found that increases in cortisol were associated with reductions in donations in participants scoring high on mentalising, whereas Azulay et al. (2022) found that increases in cortisol predicted enhanced donations in participants with high trait empathy (see also Speer et al., 2022). Given the heterogeneity in behaviour in economic games under stress (Nitschke, Forbes, et al., 2022), future studies may benefit from manipulating the effort involved in helping, or other non-financial costs, to help us understand how the social behaviour of different individuals is affected by acute stress.
The neuroimaging data showed that participants with a more prosocial SVO showed reduced responses in the dACC to SV across both self and other trials at greater levels of perceived stress. This suggests that more prosocial individuals may become less sensitive to SV overall following stress, whilst the responses of more individualistic individuals to SV do not change under stress. Yet, it remains unclear how this relates to reduced prosocial behaviour under stress in these individuals. We also found that – independent of our stress manipulation – SVO angle predicted increased responses to SVother but not to SVself. This is in line with previous work showing that SVO modulates dmPFC/dACC activity depending on whether decisions are social or self-interested (Kuss et al., 2015).
Future directions and implications
Participants earned money for either themselves or the next participant in the study who was not known to them. An exciting avenue for future work is to manipulate the social distance of the other recipient. This will determine whether participants under stress are more willing to put in effort for close others, such as family or friends (Margittai et al., 2015, 2018). Additionally, it will be important to explore whether we see similar reductions in prosocial motivation under stress when preventing harm in others (Hartmann et al., 2022) rather than gaining rewards for them.
Since there are known sex/gender differences in social behaviours following acute stress (Nitschke, Pruessner, et al., 2022; Tomova et al., 2014; Zhang et al., 2019), we only tested young men in the current study (see Materials and Methods). Age has also been shown to modulate effortful prosocial behaviour (Cutler et al., 2021; Lockwood et al., 2021). Thus, future studies should investigate how effortful prosocial behaviour in women and in older populations is affected by acute stress.
The experience of stress is a common everyday occurrence (Hassard et al., 2018) and the negative effects of prolonged stress for our physical and mental health are well-documented (Lecrubier, 2001). Our results show that acute stress is not only potentially detrimental for our own wellbeing but can have negative consequences for our behaviour towards others. Given the importance of effortful prosocial behaviour for relationships and, more generally, for social cohesion, this highlights the importance of tackling the major sources of stress, such as low income (Haushofer & Fehr, 2014) and inequality (Pickett & Wilkinson, 2010), as well as providing individuals with the resources to cope with stress when it does occur (e.g. Meichenbaum, 2017).
Our study demonstrates that participants under acute stress were less willing to exert effort for another person’s benefit at a low level of effort. The adverse effects of acute stress on prosocial effortful behaviour were most marked in more selfish individuals. These findings show using effort-based tasks and incorporating trait differences in prosocial tendencies could be key to understanding changes in prosocial behaviour under acute stress. Additionally, we found that the effects of stress were related to activation differences in the AI and dACC - areas heavily implicated in effort-based decisions and sensitive to stress effects. Thus, we provide valuable insight into the potential neural mechanisms underlying stress-induced changes in effortful prosocial behaviour. Moreover, the results raise the possibility that shifts to automatic or habitual responding under stress could also apply within the social domain (Hermans et al., 2014). Together, this emphasises the need to tackle the sources of stress, and provide individuals with the resources to cope with it, so that the potential detrimental effects of stress on prosocial behaviour are curtailed.
Materials and methods
We recruited healthy, right-handed, male students who were non-smokers or smoked less than 5 cigarettes per day. Only male participants were included as previous work has shown that they show a stronger salivary glucocorticoid response to laboratory stressors (Kirschbaum et al., 1999). We did not recruit participants who had studied or were studying psychology nor those who had taken part in previous stress induction studies. To achieve 80% power to detect a small-to-medium effect size (f = 0.15; η2 = 0.022) in a between-subject (Group: stress vs. control) by within-subject (Recipient: self vs. other) design (calculated using G*Power, 76), we required a sample size of at least 90 participants. To account for potential exclusions, we recruited 96 participants in total.
Three participants were excluded due to either a disruption during the scanning (n=1), having previously worked in the lab (n=1), or never choosing to work on other trials (n=1). This left a sample of 93 participants (mean age = 23.5 years, SD = 2.89, range = 18-31) for the behavioural analysis with 47 in the stress group and 46 in the control group. The study was approved by the local ethics committee and all participants provided written informed consent. Participants were paid 40 Euros for their participation plus the money they received from the prosocial effort task (between 6-8 Euros).
Participants came to the lab twice. In the first session, participants were familiarised with the scanning environment and completed a series of online questionnaires, including Social Value Orientation (Murphy et al., 2011; Murphy & Ackermann, 2014). Participants were told not to consume any alcohol, cigarettes, or medication, nor to engage in vigorous exercise 24 hours before the second session, and, two hours before the session, not to consume any food or drinks other than water. During the second session, participants completed the stress induction procedure (or a control procedure) and prosocial effort task within the MRI scanner. Saliva samples and visual analogue scales were used throughout the second session to measure cortisol and perceived stress. The second session was within approximately one week of the first session, lasted 160 minutes, and always took place between 12:00 and 18:00 to account for diurnal fluctuations in cortisol.
Participants in the stress group completed an adapted version of the Montreal Imaging Stress Test (MIST; Dedovic et al., 2005; Tomova et al., 2017) in which they completed challenging mental arithmetic questions under time pressure. During the MIST, participants saw their live and, seemingly, ‘below average’ performance on a scale at the top of the screen and a video stream of the experimenters observing their performance. This ensured social evaluation from the experimenters who also reminded participants that they could not use their data if their performance did not reach the group average. Throughout the scanning session, participants completed three six-minute blocks of the MIST and also completed a counting task three times in two-minute blocks (Kirschbaum et al., 1993). Here, participants were required to count backwards in steps of 13 or 17 from a large number (e.g. 2053, 2036, 2019, etc.). The experimenters asked participants to start the count again if they made a mistake or speed up if responses were not quick enough. Participants in the control group completed the same mental arithmetic questions in the MIST but without the time pressure and social evaluation. The experimenters wore white lab coats in the stress condition but not in the control condition. For the counting task, the control group counted silently in steps of either 5, 10 or 20 for 2 minutes and did not receive feedback from the experimenters. Participants completed a MIST block, a run of the prosocial effort task, the counting task, and another run of the prosocial effort task, and this sequence was repeated three times during the scanning period. This ensured that participants’ psychological and physiological stress responses were maintained throughout the scanning period (Figure 2).
Cortisol and stress measurements
To measure cortisol responses, saliva samples were collected throughout the experiment using oral swabs (Sarstedt Salivette®) which were placed in participants’ mouths for two minutes. Six saliva samples were collected (all times are relative to the start of stress induction): upon arrival (sample 1; -42 min), before the first block (sample 2; 0 min) of the MIST after the second MIST (sample 3; 38 mins) and during the third MIST (sample 4; 64 mins), at the end of the scan (sample 5; 94 mins), and then following debrief (sample 6; 119 mins). Following each session, saliva samples were frozen and stored at -20°C until analysis. During each saliva sample, we asked participants to complete a visual analogue scale to determine their perceived stress levels (“Right now I feel stressed”) on a 7-point scale from “not at all” (1) to “very much” (7). Additionally, participants completed these ratings following the first MIST (rating 3; 11 mins) and following the second counting task (rating 5; 51 minutes), although saliva samples were not collected at these time points (see Figure 2).
After thawing, the oral swabs were centrifuged at 3,000 rpm for 5 minutes, which resulted in a clear supernatant of low viscosity. Salivary concentrations were measured using commercially available chemiluminescence immunoassay with high sensitivity (IBL International, Hamburg, Germany). The intra- and inter-assay coefficients for cortisol were both below 9%. Three subjects were excluded from the cortisol analysis as they failed to provide sufficient saliva in five, three, and two (out of a possible six) saliva samples, respectively. One participant did not provide sufficient saliva for his fifth sample, but this value was interpolated using his fourth and sixth sample to allow area under the curve to be calculated. This created a final sample of 91 participants (stress = 46, control = 45) whose saliva samples were analysed. Three participants had a missing visual analogue scale stress rating and these were interpolated in the same manner to allow AUC to be calculated.
Prosocial effort task
We used the prosocial effort task developed by Lockwood et al. (2017). On each trial, participants had a choice between two options: work or rest. The rest option always involved receiving a low reward for no effort (one credit for no effort), whereas the work option involved putting in physical effort for a higher reward (more credits for more effort). The rest option was always the same, whereas the effort level and reward level for the work option differed on each trial. There were five reward levels - two, four, six, eight, or ten credits - and these credits were converted to money at the end of the study. There were five effort levels - 30, 40, 50, 60, or 70% of participants maximum voluntary contraction (MVC) - and the effort level was indicated by how filled the circle was, i.e., the more filled the circle, the more effort was required. Each participants’ MVC was collected at the start of the study by asking them to squeeze a MR compatible hand dynamometer (Current Designs Inc, Philadelphia, USA) as hard as possible. Thus, the effort level was calibrated to each participants’ individual strength. During practice, participants were familiarised with each effort level.
Each trial started with a variable jitter (2 - 3.5 s), after which participants were presented with the binary choice between work and rest. They had 3.2 seconds to make their decision by using their left hand to press the button box. After 3.2 seconds, their choice was highlighted with a yellow box for 0.3 seconds. If they chose to rest, then they did not do anything for the remainder of the trial and received one credit at the end of it. If they chose to work, then following a variable jitter (2 - 3.5 s), participants had to squeeze the hand dynamometer at the required effort level for one second during a three second window. During the squeezing period, a bar was displayed on the screen which showed the amount of force participants were currently exerting and a yellow line indicated the required effort level. If participants squeezed above the yellow line for at least one second, then following another variable jitter (1 - 2 s), the credits they received (or the next participants received) were displayed on the screen (0.5s). If participants failed to squeeze above the yellow line during the three second period, then they received no credits. If participants failed to respond within 3.2 seconds during the offer period, then “0 credits - Please respond quicker!” was displayed on the screen for the duration of the trial.
On half the trials, participants made decisions whether to put in effort to earn more money for themselves (self trials), and on the other half of trials, they could put in effort to earn money for the next participant in the study (other trials). On self trials, “You” (German: “Du”) appeared in the middle of the screen and the stimuli were red, whereas, on other trials “Thomas” (i.e., the name of the next participant) was displayed in the middle of the screen and the stimuli were presented in blue (Figure 1). At the end of the study, participants received the money they earned for themselves plus the money the previous participant had earned for them. Each combination of effort level and reward level was presented three times in self trials and three times in other trials. This created a total of 150 trials presented in a pseudorandomised order and split into six runs of 25 trials. The task was presented using the Cogent toolbox (http://www.vislab.ucl.ac.uk/cogent.php) in Matlab (MathWorks).
Computational modelling of choices
We used a model comparison approach (Lockwood et al., 2017) in which we compared models with a single discount parameter, K, to those with a separate discount parameter for self (Kself) and other (Kother), and models with a single noise parameter, β, to those with a separate noise parameter for self (βself) and other (βother) trials. We also compared models in which rewards were discounted linearly (subjective value = reward - effort * k), hyperbolically (subjective value = reward / [1 + [effort * k]]) and parabolically (subjective value = reward - effort2 * k). In total, these combinations (2x2x3) created 12 different models which we compared using the Bayesian information criterion (BIC) based on the log-likelihood. The discount parameter, k, was bounded from 0 to 1.5 to ensure an appropriate range.
FMRI acquisition and analysis
MRI data were collected using a 3 Tesla MRI scanner (Skyra, Siemens Medical) and a 32 channel-head coil. The structural scans were acquired using a magnetization prepared rapid gradient echo (MPRAGE) sequence with the following parameters: TR = 2300 ms; TE = 2.43 ms; flip angle = 8°, voxel size = 0.8 mm isotropic; field of view (FOV) = 240 x 240 mm. BOLD functional scans were acquired with a multiband accelerated EPI sequence with the following parameters: TR = 1200 ms; TE = 34 ms; flip angle = 66°; slices = 52; multiband acceleration factor = 4 (i.e. 13 excitations per TR); FOV = 210 x 210 mm, voxel size = 2 x 2 x 2 mm.
Data were preprocessed and analysed using SPM12 in Matlab (http://www.fil.ion.ucl.ac.uk/spm/). The functional images were slice-time corrected to the middle slice, realigned to the mean image, and smoothed with a Gaussian kernel (5 mm full-width at half maximum; FWHM). The structural scan was co-registered to the mean functional scan from the first run and then segmented into grey matter, white matter, cerebrospinal fluid, bone, soft tissue, and air. Diffeomorphic Anatomical Registration Through Exponentiated Lie Algebra (DARTEL; Ashburner, 2007) was used to normalise both the structural and functional scans to the Montreal Neurological Institute (MNI) template.
Next, we looked at head motion using framewise displacement (FD; Power et al., 2012) and excluded four participants who showed FD greater than 0.5 mm in over 35% of scans in two or more runs. Two subjects showed an FD over 0.5 mm in over 35% of scans in run 6. These runs were excluded but the two participants remained in the analysis. Data from run 2 was missing for one participant due to a technical issue during the scan so this run was also not included in the neuroimaging analysis. Finally, as outlined above, one subject in the stress group never chose to exert effort for the other so was excluded from the analysis. This left a final sample of 89 participants (45 control; 44 stress) for the imaging analysis with three out of these 89 participants missing one run from the six.
To create our design matrix regressors were constructed for three events in the trial - the offer phase, force phase, and outcome phase (Lockwood et al., 2022) - and these were convolved with SPM’s canonical haemodynamic response function. Each regressor had an associated parametric modulator: for the offer phase this was the subjective value of the chosen option relative to subjective value of the non-chosen option; for the force phase this was the chosen effort level (0:5), and for the outcome phase this was the reward level (i.e., the number of credits) received at the end of the trial. We split the trials according to the recipient of the reward so there were separate parametric regressors for self and other trials. If a participant did not respond quickly enough when making their choice (within 3.2 s), then this trial was labelled as a missed trial and was included as an additional regressor.
Six head motion parameters were included in the general linear model (GLM) as were scrubbing regressors (Power et al., 2014). Functional scans were individually “scrubbed” if they showed an FD above 0.5 mm and these were included as nuisance regressors in the first level analysis. Data were high-pass filtered with a 128 s cutoff as is standard in SPM and all six runs were combined into one GLM.
For each of the parametric modulators, we created first level images for the self > other (1 -1) and other > self (-1 1) contrast. This allowed us to look at regions which responded more on self trials compared to other trials and vice versa. These images were then inputted into our second-level analyses in which we conducted one-sample t-tests with perceived stress (AUC) as a covariate. This whole brain analysis used a statistical threshold of p<0.05 family-wise error (FWE) corrected at the cluster-level having thresholded at p<0.001 across the whole brain. We used the REX toolbox (http://web.mit.edu/swg/software.htm) to extract parameter estimates from two regions of interest (ROIs), the dACC and AI, which were taken from Lockwood et al. (2022).
Data, Material and Software Availability
Data and analysis code are available on the Open Science Framework (https://osf.io/4bqf7/).
We thank Livia Tomova for guidance in designing the stress induction protocol and Jonas Nitschke for providing extensive feedback on a previous version of the manuscript. This work was supported by the Austrian Science Fund (FWF I3381, awarded to C. L). I.C.W was supported by the Austrian Science Fund (FWF P 34775-B). M.A.J.A was supported by a Biotechnology and Biological Sciences Research Council David Phillips Fellowship (BB/R010668/1). P.L.L was supported by Medical Research Council Fellowship (MR/ P014097/1 and MR/P014097/2), a Sir Henry Dale Fellowship funded by the Wellcome Trust, the Royal Society (223264/Z/21/Z), and a Jacobs Foundation Research Fellowship.
- Hypothalamic blood flow correlates positively with stress-induced cortisol levels in subjects with social anxiety disorderPsychosomatic Medicine 68:859–862
- A fast diffeomorphic image registration algorithmNeuroimage 38:95–113
- Empathy Modulates the Effect of Stress Reactivity on Generous GivingFrontiers in Neuroscience 16https://doi.org/10.3389/fnins.2022.814789
- Random effects structure for confirmatory hypothesis testing: Keep it maximalJournal of Memory and Language 68:255–278
- The valuation system: A coordinate-based meta-analysis of bold> fMRI experiments examining neural correlates of subjective valueNeuroImage 76:412–427https://doi.org/10.1016/j.neuroimage.2013.02.063
- Lme4: Linear mixed-effects models using Eigen and S4. R package version 1.1–7. 2014
- Neural signatures of prosocial behaviorsNeuroscience & Biobehavioral Reviews 118:186–195https://doi.org/10.1016/j.neubiorev.2020.07.006
- Acute stress selectively reduces reward sensitivityFrontiers in Human Neuroscience 7
- The brain under stress—A systematic review and activation likelihood estimation meta-analysis of changes in bold> signal associated with acute stress exposureNeuroscience & Biobehavioral Reviews 124:89–99https://doi.org/10.1016/j.neubiorev.2021.01.001
- Acute psychosocial stress increases cognitive-effort avoidancePsychological Science 32:1463–1475
- Registered Replication Report: Rand, Greene, and Nowak (2012)Perspectives on Psychological Science 12:527–542https://doi.org/10.1177/1745691617693624
- Perturbations in Effort-Related Decision-Making Driven by Acute Stress and Corticotropin-Releasing FactorNeuropsychopharmacology 41:2147–2159https://doi.org/10.1038/npp.2016.15
- Behavioural studies of strategic thinking in gamesTrends in Cognitive Sciences 7:225–231
- Acute stress impairs reward learning in menBrain and Cognition 147https://doi.org/10.1016/j.bandc.2020.105657
- Specific Configuration of Dendritic Degeneration in Pyramidal Neurons of the Medial Prefrontal Cortex Induced by Differing Corticosteroid RegimensCerebral Cortex 17:1998–2006https://doi.org/10.1093/cercor/bhl108
- Biased sequential sampling underlies the effects of time pressure and delay in social decision makingNature Communications 9https://doi.org/10.1038/s41467-018-05994-9
- Neurocomputational mechanisms underlying subjective valuation of effort costsPLOS Biology 15https://doi.org/10.1371/journal.pbio.1002598
- Effort shapes social cognition and behaviour: A neuro-cognitive frameworkNeuroscience & Biobehavioral Reviews 118:426–439https://doi.org/10.1016/j.neubiorev.2020.08.003
- Effort-Based Cost-Benefit Valuation and the Human BrainJournal of Neuroscience 29:4531–4541https://doi.org/10.1523/JNEUROSCI.4515-08.2009
- A comparative fMRI meta-analysis of altruistic and strategic decisions to giveNeuroImage 184:227–241https://doi.org/10.1016/j.neuroimage.2018.09.009
- Older adults across the globe exhibit increased prosocial behavior but also greater in-group preferencesNature Aging 1:880–888https://doi.org/10.1038/s43587-021-00118-3
- The Montreal Imaging Stress Task: Using functional imaging to investigate the effects of perceiving and processing psychosocial stress in the human brainJournal of Psychiatry and Neuroscience 30:319–325
- Neural correlates of processing stressful information: An event-related fMRI studyBrain Research 1293:49–60https://doi.org/10.1016/j.brainres.2009.06.044
- Evidence of Conjoint Activation of the Anterior Insular and Cingulate Cortices during Effortful TasksFrontiers in Human Neuroscience 8https://doi.org/10.3389/fnhum.2014.01071
- Why stress and hunger both increase and decrease prosocial behaviourCurrent Opinion in Psychology 44:49–57https://doi.org/10.1016/j.copsyc.2021.08.023
- G* Power 3: A flexible statistical power analysis program for the social, behavioral, and biomedical sciencesBehavior Research Methods 39:175–191
- Neural signatures of fairness-related normative decision making in the ultimatum game: A coordinate-based meta-analysisHuman Brain Mapping 36:591–602
- Stress and decision making: Neural correlates of the interaction between stress, executive functions, and decision making under riskExperimental Brain Research 232:957–973https://doi.org/10.1007/s00221-013-3808-6
- Placebo Analgesia Reduces Costly Prosocial Helping to Lower Another Person’s PainPsychological Science 0https://doi.org/10.1177/09567976221119727
- The cost of work-related stress to society: A systematic reviewJournal of Occupational Health Psychology 23
- On the psychology of povertyScience 344:862–867https://doi.org/10.1126/science.1232491
- Dynamic adaptation of large-scale brain networks in response to acute stressorsTrends in Neurosciences 37:304–314https://doi.org/10.1016/j.tins.2014.03.006
- Medial prefrontal cortex and right insula activity predict plasma ACTH response to trauma recallNeuroImage 47:872–880https://doi.org/10.1016/j.neuroimage.2009.05.088
- Impact of Gender, Menstrual Cycle Phase, and Oral Contraceptives on the Activity of the Hypothalamus-Pituitary-Adrenal AxisPsychosomatic Medicine 61:154–162
- The ‘Trier Social Stress Test’–a tool for investigating psychobiological stress responses in a laboratory settingNeuropsychobiology 28:76–81
- Psychosocial versus physiological stress—Meta-analyses on deactivations and activations of the neural correlates of stress reactionsNeuroImage 119:235–251https://doi.org/10.1016/j.neuroimage.2015.06.059
- Neuronal correlates of social decision making are influenced by social value orientation—An fMRI studyFrontiers in Behavioral Neuroscience 9https://doi.org/10.3389/fnbeh.2015.00040
- Neurobiology of ProsocialityHandbook of Prosociality: Development, Mechanisms, Promotion
- The burden of depression and anxiety in general medicineJournal of Clinical Psychiatry 62:4–11
- Emmeans: Estimated marginal means, aka least-squares means. R package version 1.4. 7. 2020
- Aging Increases Prosocial Motivation for EffortPsychological Science 32:668–681https://doi.org/10.1177/0956797620975781
- Prosocial apathy for helping others when effort is requiredNature Human Behaviour 1https://doi.org/10.1038/s41562-017-0131
- Distinct neural representations for prosocial and self-benefiting effortCurrent Biology
- The neural basis of effort valuation: A meta-analysis of functional magnetic resonance imaging studies [Preprint]https://doi.org/10.1101/2021.01.08.425909
- A friend in need: Time-dependent effects of stress on social discounting in menHormones and Behavior 73:75–82
- Dissociable roles of glucocorticoid and noradrenergic activation on social discountingPsychoneuroendocrinology 90:22–28https://doi.org/10.1016/j.psyneuen.2018.01.015
- Stress inoculation training: A preventative and treatment approachThe Evolution of Cognitive Behavior Therapy
- The impact of stress in decision making in the context of uncertaintyJournal of Neuroscience Research 93:839–847https://doi.org/10.1002/jnr.23521
- Social Value Orientation: Theoretical and Measurement Issues in the Study of Social PreferencesPersonality and Social Psychology Review 18:13–41https://doi.org/10.1177/1088868313501745
- Measuring social value orientationJudgment and Decision Making 6:771–781
- Does Stress Make Us More—Or Less— Prosocial? A Systematic Review and Meta-Analysis of the Effects of Acute Stress on Prosocial Behaviours Using Economic Games
- Stress and Stress-Induced Glucocorticoids Facilitate Empathic Accuracy in Men but Have No Effects for WomenPsychological Science
- Imaging stress: An overview of stress induction methods in the MR scannerJournal of Neural Transmission 126:1187–1202https://doi.org/10.1007/s00702-018-01965-y
- Working-memory capacity protects model-based learning from stressProceedings of the National Academy of Sciences 110:20941–20946
- Why not try harder? Computational approach to motivation deficits in neuro-psychiatric diseasesBrain 141:629–650https://doi.org/10.1093/brain/awx278
- The spirit level: Why equality is better for everyone
- Spurious but systematic correlations in functional connectivity MRI networks arise from subject motionNeuroimage 59:2142–2154
- Methods to detect, characterize, and remove motion artifact in resting state fMRINeuroimage 84:320–341
- Separate Valuation Subsystems for Delay and Effort Decision CostsJournal of Neuroscience 30:14080–14090https://doi.org/10.1523/JNEUROSCI.2752-10.2010
- Deactivation of the Limbic System During Acute Psychosocial Stress: Evidence from Positron Emission Tomography and Functional Magnetic Resonance Imaging StudiesBiological Psychiatry 63:234–240https://doi.org/10.1016/j.biopsych.2007.04.041
- Two formulas for computation of the area under the curve represent measures of total hormone concentration versus time-dependent changePsychoneuroendocrinology 28:916–931https://doi.org/10.1016/S0306-4530(02)00108-7
- Acute Psychological Stress Reduces Working Memory-Related Activity in the Dorsolateral Prefrontal CortexBiological Psychiatry 66:25–32https://doi.org/10.1016/j.biopsych.2009.03.006
- Reversibility of apical dendritic retraction in the rat medial prefrontal cortex following repeated stressExperimental Neurology 196:199–203https://doi.org/10.1016/j.expneurol.2005.07.008
- A feature-based network analysis and fMRI meta-analysis reveal three distinct types of prosocial decisionsSocial Cognitive and Affective Neuroscience 16:1214–1233
- Stress increases the feeling of being looked atPsychoneuroendocrinology 37:292–298
- Separate neural pathways process different decision costsNature Neuroscience 9:1161–1168
- Altruism under Stress: Cortisol Negatively Predicts Charitable Giving and Neural Value Representations Depending on Mentalizing CapacityThe Journal of Neuroscience 42:3445–3460https://doi.org/10.1523/JNEUROSCI.1870-21.2022
- Acute Stress Induces Selective Alterations in Cost/Benefit Decision-MakingNeuropsychopharmacology 37:2194–2209https://doi.org/10.1038/npp.2012.69
- Stress influences environmental donation behavior in menPsychoneuroendocrinology 63:311–319
- Speer, S., Martinovici, A., Smidts, A., & Boksem, M. (2022). The effect of acute stress on dishonesty: The moderating role of moral default. PsyArXiv. 10.31234/osf.io/ef23ghttps://doi.org/10.31234/osf.io/ef23g
- Decision making under stress: A selective reviewNeuroscience & Biobehavioral Reviews 36:1228–1248https://doi.org/10.1016/j.neubiorev.2012.02.003
- Be nice if you have to—The neurobiological roots of strategic fairnessSocial Cognitive and Affective Neuroscience 10:790–796
- Biobehavioral responses to stress in females: Tend-and-befriend, not fight-or-flightPsychological Review 107
- Personality and prosocial behavior: A theoretical framework and meta-analysisPsychological Bulletin 146:30–90https://doi.org/10.1037/bul0000217
- Increased neural responses to empathy for pain might explain how acute stress increases prosocialitySocial Cognitive and Affective Neuroscience 12:401–408https://doi.org/10.1093/scan/nsw146
- Acute stress alters neural patterns of value representation for othersNeuroImage 209https://doi.org/10.1016/j.neuroimage.2019.116497
- Is stress affecting our ability to tune into others? Evidence for gender differences in the effects of stress on self-other distinctionPsychoneuroendocrinology 43:95–104https://doi.org/10.1016/j.psyneuen.2014.02.006
- Acute stress does not affect economic behavior in the experimental laboratoryPloS One 16
- Time-dependent changes in altruistic punishment following stressPsychoneuroendocrinology 38:1467–1475
- The Social Dimension of Stress Reactivity: Acute Stress Increases Prosocial Behavior in HumansPsychological Science 23:651–660https://doi.org/10.1177/0956797611431576
- The effects of acute stress and stress hormones on social cognition and behavior: Current state of research and future directionsNeuroscience & Biobehavioral Reviews 121:75–88https://doi.org/10.1016/j.neubiorev.2020.11.026
- Asymmetric effects of acute stress on cost and benefit learningPsychoneuroendocrinology 138https://doi.org/10.1016/j.psyneuen.2021.105646
- Functional Specialization within Medial Frontal Cortex of the Anterior Cingulate for Evaluating Effort-Related DecisionsJournal of Neuroscience 23:6475–6479https://doi.org/10.1523/JNEUROSCI.23-16-06475.2003
- Perfusion functional MRI reveals cerebral blood flow pattern under psychological stressProceedings of the National Academy of Sciences 102:17804–17809https://doi.org/10.1073/pnas.0503082102
- Social Value Orientation Moderated the Effect of Acute Stress on Individuals’ Prosocial BehaviorsFrontiers in Psychology 13https://doi.org/10.3389/fpsyg.2022.803184
- How Cortisol Reactivity Influences Prosocial Decision-Making: The Moderating Role of Sex and Empathic ConcernFrontiers in Human Neuroscience 13https://doi.org/10.3389/fnhum.2019.00415