Dual-Self Model

Smoking is the leading preventable cause of premature death in the United States.¹ While approximately 70% of smokers indicate a desire to quit smoking, only 2% to 3% actually do so. This internal conflict between the desire and ability to stop smoking comes from the difficulty of prioritizing long-term benefits (i.e. improved health) over short-term benefits (i.e. the rush of dopamine from nicotine).² If you’ve ever felt like your long-term and short-term goals are at war, you’ve experienced the concept driving dual-self models.³ 

Dual-self models in behavioral science refer to the conflict that we feel when trying to balance our immediate desires with options that we know are more beneficial in the long-run.³  While classic economic theory predicts humans to be rational decision makers who prioritize maximum utility (also known as homo economicus), this assumption doesn’t correspond to reality. Smokers, for example, choose the momentary dopamine hit from a cigarette over the long-term health benefits of avoiding cigarettes, even if this behavior is not rational. Dual-self models thus address the paradox created by classic economic views of decision utility, by considering many different decision criteria.

Utility: The total satisfaction derived from consuming a good or service.

Homo economicus: The notion that humans make rational economic decisions under an ideal mathematical optimization framework. Any and all decisions should come easily, maximizing utility.

Homo sapiens: The notion that humans do more than try to maximize utility in their decision making. Rather, psychological, cultural, and social factors constrain a human’s rationality.

Myopic: In terms of decision making, myopia refers to short-sighted (and ultimately selfish) choices.

Temporal discounting: The tendency for rewards to decrease in subjective value as the time to receive the reward increases.

Risk aversion: The tendency for humans to prefer outcomes with low uncertainty over outcomes with high uncertainty.

According to classical economists, homo economicus always considered their various motivations through conscious deliberation.³ Yet, according to Adam Smith – often called the father of economics – homo sapiens frequently do not do this.^{5 6} By modelling humans as if they always maximize utility, our decision making processes are confined to a simple black box.³

Behavioral economists Richard Thaler and Hersh Shefrin developed the first economic dual-self model in their 1981 article, “An economic theory of self-control.”⁴ Inspiration for Thaler and Shefrin’s model came from the researchers who identified the paradoxical nature of the concept of self-control: unless it can be assumed that one’s consciousness contains more than one energy system, and that those energy systems are independent,⁷ it doesn’t make sense that people would impose constraints on their future behavior.⁴ As a result, Thaler and Shefrin’s model focused on what they called a “two-self economic man.”

Thaler and Shefrin’s model was developed in the context of behaviors around financial savings, based around the concepts of a far-sighted “planner” and a short-sighted “doer.”4 The planner and doer are in conflict with each other at any point in time, based on their polarizing preferences. The “planner” is concerned with lifetime utility, so the “planner” understands that every single decision should maximize utility. On the other hand, the “doer” only exists for fleeting moments and is completely selfish regarding those immediate desires.

A more recent dual-self model was published by American Economists Drew Fudenberg and David Levine in their 2006 article, “A dual-self model of impulse control.”⁸ Under this model, the economists considered economic concepts such as temporal discounting and risk aversion. Fudenberg and Levine proposed that many decision-making problems should be viewed as a game between a series of short-term impulsive selves and a long-term, patient self. This idea is consistent with evidence from magnetic resonance imaging studies, which suggested that short-term impulsive behavior is associated with different brain areas than long-term planned behavior.⁹ 

In Fudenberg and Levine’s model, the patient long-term self and a series of short-term myopic selves ultimately share the same preferences over stage-game outcomes.⁸ Stage games require players to consider the impact of their current actions on the future actions of other players, since there are repetitive turns. This means that both the patient long-term self and the short-term myopic selves want the same outcomes for the decision maker. However, they differ in how they regard the future. The short-term self has baseline preferences in the game that is dependent on the outcome of the current stage, which allows them to be completely myopic. On the other hand, the long-term self tries to maximize utility across time.

It’s easy to see the similarities between Thaler and Shefrin’s 1981 two-self model, and Fudenberg and Levine’s 2006 model of impulse control: Thaler and Shefrin’s “planner” is similar to Fudenberg and Levine’s patient long-term self; Thaler and Shefrin’s “doer” is similar to Fudenberg and Levine’s myopic short-term self.^{4 8} So what’s the difference?

Ultimately, the difference lies in Fudenberg and Levine’s emphasis on different criteria that influence one’s short-term selves and long-term self: concepts like temporal discounting and risk aversion.⁸ The economists factored in the fact that humans prefer choices that seem more certain, compared to uncertain choices. This risk aversion can make us more inclined to satisfy our short-term selves’ desires, since immediate choices are often more certain than choices that occur in the future. Similarly, for temporal discounting, our long-term self hopes to maximize utility, which includes satisfaction. If smaller, immediate rewards grant us more gratification than larger, delayed rewards, then our myopic selves have another leg to stand on, in their fight with our patient long-term self.

In the case of smoking, for example, Fudenberg and Levine’s model acknowledges the fact that quitting smoking does not guarantee immediate health benefits, especially for those who do not believe smoking is detrimental to their health, in the first place. Compared to the immediate gratification that comes with smoking, improved health as a result of smoking cessation is much more delayed. Considering these factors of risk aversion and temporal discounting, we can better understand why the short-term myopic selves may overpower the patient long-term self.

Dual-self models connect to research on emotions and motivational drive.¹⁶ In Adam Smith’s The Theory of Moral Sentiments, for example, Smith argued that homo sapiens’ behaviors were driven by struggles between passions and what Smith called the “impartial spectator.”⁶ Passions included things like hunger and sex drives, and emotions such as fear, anger, and pain. On the other hand, “impartial spectators” do not have such impulsive passions, which allows them to enjoy immediate or delayed pleasures equally. As a result, the lack of passions allow people to maximize utility and prioritize long-term benefits.¹⁶

Though not an explicit dual-self model in behavioral economics, these researchers proposed a two-system model where people’s behaviors result from an interaction between:¹⁶

1. Their affective system, which includes emotions like anger and fear, and motivational drives such as those involving hunger and sex; and,
2. Their deliberative system, which assesses options with a broad, goal-based perspective.

Increasing retirement savings by connecting with future selves

Researchers have turned their attention to how they can resolve the conflict between short-term and long-term goals presented in dual-self models by making people feel more connected to their future selves.¹⁷ They believe that this connectedness varies based on age difference: the sooner we meet our future self, the more we care. We feel more connected to a one-year older future self-compared to a self in thirty or forty years, for example.

People are expected to live for approximately ¹⁶ years in retirement, which has increased since the 19th century.¹⁷ Naturally, increased time spent in retirement increases the risk that people will outlive their savings or undergo a decrease in a quality of life. Since many people fail to save enough retirement money, research on temporal discounting has suggested removing the appeal of immediate rewards by having people pre-commit to future decisions. However, pre-commitment plans have mostly shown success in the context of large corporations offering retirement commitment programs for their employees. What about increasing savings for those who work in environments that do not offer such programs?

To assess the retirement savings problem in the context of dual-self models, a group of American researchers exposed people to their future selves in the form of age-progressed avatars.¹⁷ As these participants felt more connected to, and witnessed their future selves in a virtual reality dependent on their retirement savings, they allocated twice as much more money to a retirement account, compared to participants who were not exposed to virtual versions of their future selves. Thus, being exposed to one’s future self led to lower discounting of future rewards. These findings encourage more researchers to find ways to make people feel connected to their future selves.

The two-period problem of self-control

Self-control, or the lack thereof, is a problem for our decision making processes.¹⁸ Self-control is even more difficult given that we prioritize immediate goals more than long-term goals, as per the dual self model. One way of thinking about people who fail to exert self-control to override their selfish preferences is to envision an alter-ego who prefers immediate gratification, being evaluated by a long-term decision maker.

Drawing from this concept, two economics professors broke down self-control into two periods.¹⁸ In the first period, an individual must decide on feasible choices from which they will select a final choice. In the second period, the individual either chooses a superior alternative  or they stick with their first choice. To this end, the researchers proposed that the conflict between two selves only occurs in the second period, when one’s alter-ego potentially takes over to re-evaluate the initial decision.

The probability of one’s alter-ego taking over depends on which self was responsible for the first period. If the long-term self was dominant in the first period because the choice set was both appealing and associated with long-term goals, then choices could still be evaluated according to long-term preferences. However, if the short-term self was dominant in the first period, then self-control would require the alter-ego to strike a compromise between the short-term self’s preferences and long-term goals.¹⁸

Hyperbolic discounting

For a deeper dive into hyperbolic discounting, including why we choose smaller, immediate rewards over larger, delayed rewards, why it’s important to understand and how to avoid it, read through this article.

Increasing the pull of the future self

Intertemporal tradeoffs between present and future benefits are at the forefront of dual-self models, and PhD candidate Sarah Molouki explores how we can address this tradeoff. In fact, the recommended strategies include increasing our concern for our future selves, similar to the virtual aging experiment described above. Read this piece to discover how we can increase the pull of our future selves.

1. Volpp, K. G., Troxel, A. B., Pauly, M. V., Glick, H. A., Puig, A., Asch, D. A., Galvin, R., Zhu, J., Wan, F., DeGuzman, J., Corbett, E., Weiner, J., & Audrain-McGovern, J. (2019). A randomized, controlled trial of financial incentives for smoking cessation. The New England Journal of Medicine, 360, 699-709.
2. Benowitz, N. L. (2009). Pharmacology of nicotine: Addiction, smoking-induced disease, and therapeutics. Annual Review of Pharmacology and Toxicology, 49, 57-71.
3. Embrey, I. (2019). A review of economic dual-self modelling.
4. Thaler, R. H., & Shefrin, H. M. (1981). An economic theory of self-control. Journal of Political Economy, 89(2), 392-406.
5. Sharma, R. (2020, February 16). Adam Smith: The father of economics. https://www.investopedia.com/updates/adam-smith-economics/
6. Smith, A. (1759). The Theory of Moral Sentiments. Andrew Millar.
7. McIntosh, D. (1969). The Foundations of Human Society. University of Chicago Press.
8. Fudenberg, D., & Levine, D. K. (2006). A dual-self model of impulse control. The American Economic Review, 96(5), 1449-1476.
9. McClure, S. M., Laibson, D. I., Loewnstein, G., & Cohen, J. D. (2004). Separate neural systems value immediate and delayed monetary rewards. Science, 306(5696), 503-507.
10. Duigan, B. (2020, September 8). Richard Thaler. Encyclopedia Britannica. https://www.britannica.com/biography/Richard-Thaler
11. Hersh Shefrin. (n.d.). Forbes. https://www.forbes.com/sites/hershshefrin/?sh=74ed38e26033
12. Drew Fudenberg, (n.d.). MIT Economics. http://economics.mit.edu/faculty/drewf
13. Frankenfield, J. (2021, January 26). General equilibrium theory. https://www.investopedia.com/terms/g/general-equilibrium-theory.asp
14. Evolutionary game theory. (2021, April 24). Stanford Encyclopedia of Philosophy. https://plato.stanford.edu/entries/game-evolutionary/
15. Levine, D. K. (n.d.). About David Levine. Economic and Game Theory by David K. Levine. http://www.dklevine.com/david.htm
16. Ashraf, N., Camerer, C., & Loewenstein, G. (2005). Adam Smith, behavioral economist. The Journal of Economic Perspectives, 19(3), 131-145.
17. Hershfield, H. E., Goldstein, D. G., Sharpe, W. F., Fox, J., Yeykelis, L., Carstensen, L. L., & Bailenson, J. N. (2011). Increasing saving behavior through age-progressed renderings of the future self. Journal of Marketing Research, 48, S23-S37.
18. Chatterjee, K., & Krishna, R. V. (2005). Menu choice, environmental cues and temptation: A “dual self” approach to self-control.