Verified by 
Key points
- Research shows that happiness depends on the discrepancy between the rewards we get and the rewards we expected to get.
- When learning is required to make good choices between potential rewards, learning and not reward is what matters for happiness.
- More unstable environments reduce happiness, especially in people with depressive symptoms.
This guest post was written by Bastien Blain, Ph.D., a postdoctoral fellow at UCL.
Our obsession with happiness isn’t as modern as it may seem. Philosophers from Aristotle to Jeremy Bentham have all argued for the importance of subjective wellbeing. Bentham suggested that “it is the greatest happiness of the greatest number that is the measure of right and wrong”. This approach informs the policies of many nations that deploy population measures of wellbeing.
But the goal of increasing societal happiness has proved difficult to achieve, partly because it is difficult to determine what factors are most relevant for happiness. For example, many people believe they would be happier if only they had more money, but winning the lottery or receiving a large pay raise often only has temporary effects on happiness. Instead, we recently discovered that learning may play an even more important role in happiness than reward.
Anyone can contribute to our ongoing research to understand the factors that determine happiness by playing the short games in our free smartphone app, The Happiness Project. Thousands of people have played our smartphone games, helping us develop mathematical equations to predict how happiness will change from minute to minute. In one study, we found that the main factor driving happiness is not actually the rewards themselves but instead whether rewards are better than expected. Receiving a pay raise will make you feel happier only if it was bigger than what you expected. This difference between expected and actual rewards is referred to as a reward prediction error.
Reward prediction errors play a key role in learning. That’s because they motivate people to repeat behaviors that led to unexpectedly large rewards. They can also be used to update beliefs about the world, which might be rewarding in itself. For example, if you get a bigger pay rise than you expected because you made a big effort to negotiate with your employer, you will realize that this is a helpful approach that you should stick with. It may also feel like you’ve earned it.
So could it be that reward prediction errors are associated with happiness not because of the rewards, but instead because they help us understand the world a little better than before?
An Experiment to Separate Learning From Reward
To test this idea, we designed a task in which the likelihood of receiving a reward was unrelated to the size of the reward, enabling us to separate out the contributions of learning and reward in determining happiness. Seventy-five participants played a game that involved deciding which of two cars would win a race. In the “stable” condition, one of the cars always had an 80 percent chance of winning. In the “volatile” condition, one car had an 80 percent chance of winning for the first 20 trials. The other car then had an 80 percent chance of winning for the next 20 trials. The volunteers were not told these probabilities in advance but had to figure it out by trial and error while playing the game.
On every trial, the volunteers were shown the reward they would receive if the car they chose went on to win. Potential prizes were randomly assigned to the two cars. Making good choices required considering both the potential rewards and the probability of winning (it might be worth it to choose a slow car if the prize was big enough). The key thing was that the size of rewards was not useful for learning which car was more likely to win in the future.
Every few trials, the volunteers were asked to move a cursor to answer the question, “How happy are you right now?” Not surprisingly, the volunteers were happier after winning than after losing. On average, they were also less happy in the volatile compared to the stable condition. This was especially true for volunteers who reported symptoms of depression.
The biggest surprise was that happiness did not depend at all on the size of rewards. Instead, momentary happiness depended on whether outcomes were better than expected. This helped participants update beliefs and they could ignore information about the size of rewards. In other words, it was the process of learning how the game worked that made people feel good rather than the amount of reward they win.
The Benefits of Learning
These results suggest that how we learn about the world around us can be more important for how we feel than rewards we receive directly. It makes sense when you consider that learning is often considered to be intrinsically rewarding, whether it is a language, historical facts, Sudoku, or a computer game. People seek out learning opportunities even if it does not clearly result in material gain, and no one enjoys playing very easy games or unsolvable games with little opportunity for learning. Instead, we enjoy challenging games that we can learn to master.
Finding regular opportunities to learn may therefore be important for well-being. In fact, research shows that the motivation to perform an intrinsically rewarding activity, such as solving a problem, can actually be undermined when payment is introduced. In the real world, rewards are often uncertain and infrequent, but the good news is that learning may nevertheless boost happiness.
Our study raises important questions about why some people, such as those with depression, are worse at dealing with uncertain situations than others. Further research is needed to understand why this might be the case. To that end, we developed a smartphone app, The Happiness Project, that anyone can download for free to contribute to scientific research on happiness. This huge citizen science project includes games in which you learn and make decisions as well as surveys to help understand the differences between people. When people play the games on multiple days, we can see how happiness changes over time and understand the differences in what determines happiness for everyone.
This post was adapted from a post that also appears on The Conversation.
References
Blain B, Rutledge RB (2020) Momentary subjective well-being depends on learning and not reward. eLife 9, e57977. https://doi.org/10.7554/eLife.57977
Rutledge RB, Skandali N, Dayan P, Dolan RJ (2014) A neural and computational model of momentary subjective well-being. Proceedings of the National Academy of Sciences USA 111, 12252-12257. https://doi.org/10.1073/pnas.1407535111
