Mindset effects on the regulation of thinking time in problem-solving

Abstract

AbstractUnderstanding time investment while solving problems is central to metacognitive research. By the Diminishing Criterion Model (DCM), time regulation is guided by two stopping rules: a confidence criterion that drops as time is invested in each problem and the maximum time to be invested. This combination generates curved confidence–time associations. We compared the belief that intelligence is malleable, a growth mindset, to the belief that intelligence is fixed, and to neutral control groups. We hypothesized that a growth mindset leads people to selectively invest time in problems carrying the hope of improvement. This extra time makes the curved DCM pattern curvier. In two experiments, participants primed with growth, fixed, or control mindsets solved analogies (Experiment 1) and compound remote associates (Experiment 2). As expected, in both experiments a growth mindset exhibited a curvier confidence–time pattern, while the fixed mindset and control groups replicated previous confidence–time associations. Most additional time was invested in problems with intermediate difficulty levels, suggesting strategic time allocation. The study offers useful measures for delving into factors that affect thinking time allocation.Keywords: Growth mindsetproblem-solvingmeta-reasoningmetacognitiontime regulation AcknowledgmentsWe thank Meira Ben-Gad for editorial assistance.Disclosure statementNo potential conflict of interest was reported by the author(s).Notes1 In Ackerman, Yom-Tov, et al. (2020), the confidence scale appeared on a separate page. The rest of the solving procedure was identical.2 We used this phrase as the low end of the scale consistently across the experiments, rather than a phrase conveying guessing by chance. This was done so that the scale would start at zero in both experiments, regardless of whether the task format was multiple-choice or open-ended.3 For addressing the lack of remaining self-report difference in mindset at the end of the task, we made several additional analyses. First, we analyzed the two experiments together, examining the tendency towards a fixed mindset by an Analysis of Variance (ANOVA) with experiment and group as two between-participant factors. We found a main effect of the group, F(2, 304) = 4.95, MSE = 23.47, p = .008, η2p = .032. A Tukey post-hoc test revealed that there was a significant difference between fixed and growth mindsets, p = .005, while other differences were not significant, both ps > .09. Importantly, there was no interactive effect with experiment, F(2, 304) = 0.574, MSE = 2.72, p = .564, η2p = .004, suggesting on comparable manipulation effect on both experiments. Second, we examined order effects within the set of problems each participant solved on the central analysis for this study, of confidence-time association. Indeed, the curve difference was found in the second half of the task, β = -1.34, SE = 0.39, t(2137) = 3.47, p < .001, 95% CI [-1.44, -0.33], indicating that the effect remained well into the later stages of this demanding task.Additional informationFundingThis work was supported by the Israel Science Foundation.