The pandemic has disproportionally affected adolescent mental health, exacerbating existing trends of increasing mental health problems observed over recent decades. In the present study, we examined whether age-related differences in the way in which individuals tolerate uncertainty—a key characteristic of the COVID-19 pandemic—and individuals’ ability to flexibly adapt their thoughts or behaviours in the context of adversity, may account for age-related differences in mental health problems across a year of the pandemic.

In contrast with our predictions, IU was greater in adolescents compared to adults, with IU peaking in late adolescence. The existing literature which has shown less uncertainty avoidance in adolescence compared to adulthood has studied behavioural responses to uncertainty [12, 13, e.g., 15]. Indeed, adolescents (11–17 years) reported higher IU compared to the adult normative scores using self-report measures. The uncertainty reflected in behavioural tasks may be less threatening (e.g., small monetary gains and losses) than real life uncertainties where the stakes are higher (e.g., social, health, and economic uncertainty). Indeed, task-based and self-report measures of IU show only small to moderate correlations in adults [37]. Alternatively, the discrepancy observed is a true reflection of conflicting goal states in adolescence. That is, while it is developmentally beneficial to explore the adult environment [17], our results suggest this may come at the cost of psychological distress due to high cognitive IU in adolescents. If replicated, this conflict may partially account for high levels of negative affect observed in adolescents compared to adults, especially during times of high uncertainty [31].

Further, adolescence is a period of both heightened threat sensitivity and heightened reward sensitivity, which may contribute to the observed age-related differences in response to uncertainty [19]. Specifically, threat sensitivity and reward sensitivity may be differentially activated in the face of uncertainty in adolescence, such that when uncertainty signals high potential gain (e.g., peer acceptance), adolescents tend to show high tolerance of uncertainty (e.g., increased risk-taking behaviours; [38]). In contrast, when uncertainty signals high potential threat/loss (e.g., social isolation), threat sensitivity may override reward sensitivity and lead to higher intolerance of uncertainty. This may be especially true for late adolescents (18–24 years), who typically experience important transitions in educational, employment, and/or housing contexts, which were likely disrupted by COVID-19.

Supporting our hypothesis, older adults reported the lowest IU. This is consistent with the socioemotional theory, which posits that people become less concerned with future-oriented goals such as reducing uncertainty as they age [39].

The observed increase in PF with age is consistent with the selection, optimisation, and compensation theory, which suggests that older adults compensate for diminished cognitive resources by flexibly pursuing new goals or relying more on previously under-used resources such as social support [40]. The lower PF in younger adolescents potentially reflects the ongoing development of its proposed cognitive substrate, affective control, which continues to develop into late adolescence [28].

We found that IU did not account for age-related differences in mental health. That is, the relationship between IU and mental health outcomes was the same for younger and older individuals. One reason for this lack of age-related differences may be that age-related variance in mental health problems might be tied to more specific domains of uncertainty (e.g., health, social, economic). For example, it has been proposed that adolescents may be especially impacted by social uncertainties [18], which may have been particularly salient in the context of the pandemic (e.g., when they will next see their friends).

In contrast to IU, PF did partially account for the association between age and anxiety, such that this association was stronger in adults compared to other age groups. If adolescents have low PF, and PF is also less protective against anxiety during adolescence compared to adulthood, this may partially explain high rates of adolescent anxiety during the pandemic [31]. The greater protective effect of PF in adults compared to older adults may be that other mechanisms such as social support are more predictive of mental wellbeing in older adults. Together, these findings suggest that improving PF and reducing IU, especially in young people, may benefit population mental health, especially during future pandemics, or periods of significant uncertainty.