Erika Kirgios, PhD Candidate, Wharton, University of Pennsylvania
Resolving the conflict between extrinsic incentives and intrinsic motivation
Policy makers frequently reward people for behaviour that is good for them or for society. In the absence of these incentives these good behaviours might make people feel proud, signalling to themselves and others that they genuinely care about bettering themselves and the world. But what happens to these feelings if they’ve earned a financial reward for the behaviour, rather than completing it of their own initiative?
Well, these feelings of pride don’t seem to persist in the face of incentives, because financial incentives tend to be perceived as fundamentally in conflict with intrinsic motivation. As a result, actions that would otherwise signal, to others and to themselves, that they are “good people” appear to lose their signalling power when they incentivised. Indeed, both observers and the actors themselves judge those who accept incentives for good behaviour as motivated by money, not intrinsic rewards.
But how malleable are these perceptions after the effort has been made? Might people wish to disavow the financial incentives they have received, in order to signal to themselves that they did the right thing for the right (non-monetary) reason?
Research conducted with colleagues at Wharton and Booth proposes a novel intervention to resolve this common issue with incentive programs and allow people to walk away from these programs with both a boost to their incentivised behaviour and a self-image boost. This intervention is based on the insight that people who were promised rewards ex ante may choose to forgo those rewards ex post to retrospectively interpret their good behaviour as intrinsic.
We call this “motivation laundering”, inspired by research on “psychological money laundering”. Psychological money laundering refers to the idea that people will attempt to “cleanse” money earned through unethical behaviour by giving it up to a good cause or pooling it with ethical earnings. Here, we hypothesized that people would launder not their money, but their motivations, giving up some or all of their financial rewards for a completed behaviour to recast their original motivation as intrinsic, not financial. We tested this idea in two experiments, one online and one in the field.
We recruited 775 participants in exchange for $0.50. They were first asked to complete a short image classification task, irrelevant to the study. They were then given the option to take part in a second task for a $2 bonus. In all 763 took the option. They were asked to write a letter to give hope to a sick child who would be spending the holiday in hospital. All letters deemed appropriate would be sent to sick children in December 2020 via the “I See Me! Letters of Love Campaign.
After they had written their letters they were randomised into one of three experimental conditions and offered the opportunity to forgo some or all of their $2.00 incentive:
- Reminded of the intrinsic reward associated with the letter-writing. “Treat the joy and hope you’ve spread as your reward and forgo some or all of your monetary compensation”
- Active Control. Reminded of a less-valued reward. “Treat the letter-writing practice you’ve received as your reward and forgo some or all of your monetary compensation”.
- Baseline control. No additional prompts; participants were simply given the opportunity to forgo some or all of their incentive.
Participants gave up an average of 47.4¢ in the treatment condition, 35.6¢ in the active control condition, and 31.2¢ in the baseline control condition. The percentages choosing to give up anything at all were 40.6% in the treatment condition group, and 29% and 27.1% respectively in the active and baseline control condition groups. Testing these outcomes in a regression analysis and with a Wald test showed that assignment to the treatment condition significantly increased forgone earnings relative to both the control and active control conditions.
This experiment was conducted in partnership with the StepUp programme, a 28-day digital rewards programme designed to promote exercise amongst members of a large gym chain. Working with StepUp allowed us to test for motivation laundering in the context of a self-improvement behaviour instead of a pro-social behaviour. All participants earned monetary incentives during the program, averaging $2.04 (95% earned between $0.08 and $5.65). After completing the program, 17,968 StepUp participants were randomly assigned to either the treatment or control condition.
Everyone received an email inviting them to either claim their reward or donate it back to the program. Those in the treatment condition were given an additional prompt: “Treat the healthy habits you’ve kick-started as your reward” by donating the incentives back to the StepUp program. Participants who did not reply to the email were defaulted into donating because they did not claim their incentives, so donation rates in this experiment were high. These emails were sent in batches randomly between 3 days and 70 days after completing the program. The average gap between program completion and the email was around one month.
A total of 91.14% donated their earnings in the treatment condition, compared to 90.27% in the control condition. Again, a regression analysis was undertaken to predict whether participants would keep their earnings. Participants were significantly more likely to donate their reward in the treatment condition; the estimated increase in the likelihood to donate was 1.04% points. Whilst this may seem small, the test is fairly conservative; because the open rate on emails is low, the “treatment on treated” effect may be large (i.e. if the emails are only opened by 20% of recipients, the effects on those who do open and read it may be five times as large).
It was also possible to examine the treatment effect amongst those who had exercised more during the program, by tallying how many times a participant had visited the gym. We found that for every extra visit the impact of the treatment on the decision to donate increased by 99% (i.e. 1.03% points). Additionally, the length of delay between the end of the program and receipt of the email also had a fairly predictable effect; each passing week decreased the treatment effect by 47%. This suggests that motivation laundering opportunities are more attractive when they arise closer to completion of the incentivised activity.
These experiments show that, when reminded of the valued intrinsic rewards of incentivised actions (what might seem like the “right” motivation for a virtuous act), people are willing to forgo incentives for good behaviour in order to cleanse themselves of any perceptions of having impure (financial) motives.
Concerns that intrinsic motivation may be crowded out by financial incentives have long been debated in public policy, upsetting the “gift relationship”. These experiments suggest that there may be ways to resolve some of these tensions. It may be possible to use extrinsic rewards to kick-start good behaviour, then present people with opportunities to motivation launder (by forgoing their incentives) to boost self-image and reduce the cost of these programs.
This blog is based on: Kirgios EL, Chang EH, Levine EE, Milkman K, & Kessler JB “Forgoing earned incentives to signal pure motives” PNAS Latest Articles (Accessed 10 July 2020) https://www.pnas.org/content/pnas/early/2020/07/02/2000065117.full.pdf