VoxEU Column Frontiers of economic research

Overcoming naiveté about self-control

Naiveté about one’s lack of self-control can result in costly mistakes. In order to shed his or her naiveté, an individual needs to learn from his or her past lapses in self-control. This column examines whether people are able to draw the correct inferences from their past behaviour. It reports on experimental evidence that people learn well from their past effort on a task and are able to transport what they learn to new environments. However, they appear to underappreciate how much self-knowledge experience with a task will provide.

Insufficient exertion of self-control is a pervasive problem that hinders individuals from achieving their intended objectives. When deciding how much to work next week, we tend to set ambitious goals for ourselves, but when the time arrives, we often come up short. Self-control problems have been documented in effort provision in the workplace (Kaur et al. 2015), in consumption-savings decisions (Ashraf et al. 2006), in alcohol consumption (Schilbach 2019), in smoking (Giné et al. 2010), and in laboratory experiments (Augenblick et al. 2015, Augenblick and Rabin 2018). 

But, at least in principle, they are not insurmountable. When Ulysses sailed home from the Trojan War, he famously tied himself to the mast of his ship in wise anticipation of his inability to resist the alluring songs of the Sirens. We can, similarly, leverage commitment devices in order to achieve self-control (Ahn et al. 2018). Illiquid bank accounts can help us not to overspend (Laibson et al. 2003), gym memberships with no per-visit fees can help us exercise more often (DellaVigna and Malmendier 2006), and publicly announcing our goals can give us extra motivation to avoid the embarrassment of failure.

Despite the promise commitment devices hold, their use is not observed very often in everyday life (Laibson 2015). An important reason might be that we are frequently naive about our self-control problems. When we start a new project or exercise regimen, we are invariably optimistic about our ability to see it through. Hence, we see no reason to commit to a more virtuous future course of action. Ulysses shed his naiveté with the help of the sorceress Circe, who told him how to withstand the temptation of the Sirens. But without having the luxury to rely on the advice of a sorceress, people generally have to learn about their self-control from observing self-control lapses in their past. As a result, understanding and effectively counteracting persistent naiveté requires a better understanding of whether and how people (fail to) learn from their past behaviour.

Evidence on learning about self-control

We conducted an experiment in which participants were given repeated opportunities to increase their payoffs by completing effortful tasks (Le Yaouanq and Schwardmann 2019). The tasks were intentionally chosen to be uninteresting so that their completion would require participants to exert self-control: tasks either involved repeatedly positioning sliders on a target value or counting the number of zeroes in a table of ones and zeros. 

In line with previous experiments (e.g. Augenblick et al. 2015), we find that participants exhibit present bias: they choose a higher workload when they choose their effort for a future date than when the future date arrives and they choose their effort on the spot. Participants are also naive in that they overestimate the likelihood with which they will complete a high workload in the future (Augenblick and Rabin 2018).

Our main interest is in how participants’ predictions of future effort evolve with experience. Are they able to learn that they suffer from a self-control problem after an initial failure to stick to their goal? Or do they remain unrealistically optimistic about their future effort against all evidence, like the individual who, year after year, sets and fails to achieve the same New Year’s resolution? 

As a first step, we needed to establish that participants’ past behaviour is indeed informative for predictions of future behaviour because only then is there something to learn from. Unlike in standard experiments on learning biases where the experimenter controls the informativeness of a signal, the informativeness of past behaviour needs to be inferred from the data. We find that behaviour is highly correlated across time periods and that there is scope for learning. 

Furthermore, our results show that subjects learn rather well. More specifically, the beliefs participants arrive at after gaining experience with the task are consistent with their prior naiveté being updated according to the true informativeness of their experience. We also see that participants’ individual-level predictions of future effort get more precise. Taken together, these results allow us to reject the hypothesis that people are intrinsically unable or reluctant to draw lessons from their past behaviour.

While participants appear to be good at learning about their self-control problem in a given task, it might be the case that they struggle to transport their new knowledge to different environments. For example, do people realise that a tendency to file their taxes late could be indicative of procrastination in other areas of their life? 

To test this idea, we varied whether participants were exposed to the same or a different task after their initial experience. We find that participants are able to use the information acquired in a given task to predict their behaviour in a related but different task. 

Overall, our results suggest that the prevalent and persistent naiveté observed in the field must be explained by factors that are absent from our experimental setting. This points to two candidate explanations to be investigated in future research. First, it may be that individuals make correct inferences in the short run but unlearn these lessons over time because they have a better recall of their successes than of their failures (Zimmermann 2019). Second, the greater complexity people encounter in everyday environments may provide greater scope for the misattribution of failures to circumstances rather than to their intrinsic type.

Another interesting feature of our data is that they allow us to ask whether subjects foresee the learning opportunities that the exposure to a task will offer. This type of forecast has not received much attention in the literature on learning biases, which has mostly focused on retrospective learning (i.e. learning from realised information), as opposed to prospective learning (i.e. making contingent plans based on future information). We find that participants vastly underestimate how much they will learn from gaining experience with the decision problem. This suggests that we might not make the most of it when an opportunity to learn about ourselves presents itself. Experimenting with, for instance, different work arrangements, dietary choices, or exercise regimens is likely to be valuable in understanding what makes us more productive, healthier, and happier. But if we systematically underestimate how much information this experimentation can yield, then we are likely to miss out on many worthwhile opportunities to learn about ourselves.


Ahn, D, R Iijima, Y Le Yaouanq and T Sarver (2018), “Behavioral characterizations of naivete for time-inconsistent preferences”, Review of Economic Studies, forthcoming.

Ashraf, N, D Karlan and W Yin (2006), “Tying Odysseus to the mast: Evidence from a commitment savings product in the Philippines”, Quarterly Journal of Economics 121(2).

Ariely, D, and K Wertenbroch (2002), “Procrastination, deadlines, and performance: Self-control by precommitment”, Psychological Science 13(3): 219–224.

Augenblick, N, M Niederle and C Sprenger (2015), “Working over time: Dynamic inconsistency in real effort tasks”, Quarterly Journal of Economics 130(3): 1067–1115.

Augenblick, N, and M Rabin (2018), “An experiment on time preference and misprediction in unpleasant tasks”, Review of Economic Studies, forthcoming.

DellaVigna, S, and U Malmendier (2006), “Paying not to go to the gym”, American Economic Review 96(3): 694-719.

Giné, X, D Karlan and J Zinman (2010), “Put your money where your butt is: A commitment contract for smoking cessation”, American Economic Journal: Applied Economics 2(4): 213–235.

Kaur, S, M Kremer and S Mullainathan (2015), “Self-control at work”, Journal of Political Economy 123(6): 1227–1277.

Laibson, D, A Repetto and J Tobacman (2003), “A debt puzzle”, in P Aghion, R Frydman, J Stiglitz, M Woodford (eds.), Knowledge, Information, and Expectations in Modern Economics: In Honor of Edmund S. Phelps, Princeton: Princeton University Press, p. 228-266.

Laibson, D (2015), “Why don’t present-biased agents make commitments?”, American Economic Review 105(5): 267-72.

Le Yaouanq, Y, and P Schwardmann (2019), “Learning about one’s self”, CEPR Discussion Paper 13510.

Schilbach, F (2019), “Alcohol and self-control: A field experiment in India”, American Economic Review 109(4): 1290–1322.

Zimmermann, F (2019), “The dynamics of motivated beliefs”, American Economic Review, forthcoming.

3,674 Reads