Creatures of Habit

Activities can be thought of as habit forming – even addictive – when their value today becomes greater the more one has done them in the past. Many of our behaviors exhibit this sort of momentum, including good ones like hand washing and bad ones like substance abuse.

Habits tie our past, current, and future selves together.

When designing a nurturing environment for families, habits cannot be ignored. The baggage that people bring with them to a community will influence how each person acts there. And the habits that are nurtured or discouraged in the community will impact people long into the future.

This post will introduce a couple models from economics that facilitate thinking about habits and discuss my takeaways.

First, Becker and Murphy (1988) introduce a mathematical model of Rational Addiction. It illustrates the conditions in which even a perfectly planning robot might pursue a harmful addiction. As long as each subsequent hit of a drug is valued more highly than the future unhappiness it causes, the addiction will continue. (Conversely, with something good like handwashing, the habit will build as long as the inconvenience of each wash is outweighed by the future benefits.) Here’s one of the authors explaining the basic model of rational addiction in a 9-minute YouTube video.

This type of theory is useful not because we think people are automatons, but because it demonstrates how much of problematic behavior can be explained by ordinary reasoning. That turns out to be a lot. A despondent robot might pursue a strategy of, essentially, suicide by drug addiction. If the robot’s estimates of prices or benefits are adjusted mid-addiction, we expect to observe drastic consumption adjustments like binges and cold-turkey quitting. Addiction itself is portrayed as a choice that is only problematic if the agent’s predictions about the future are wrong.

The opposite extreme from Rational Addiction theory is to think that addicts have completely lost control over their choices. In formal terms, a myopic addict has time-inconsistent preferences. One moment, a person has chosen to quit their addiction and, the next, they lose control and relapse. Thinking Fast and Slow (Kahneman, 2011) is a good reference from behavioral economics to characterize why this might happen.

Both these models are useful for a community designer to understand if you think that everybody plans, but imperfectly. First, they lay out a mathematical story for how and why people’s past baggage matters. Second, they suggest external incentive structures and education that can be put in place to encourage our rational minds to pursue healthy habits. Third, they show how better internal decision-making habits can lead to better behaviors.

It’s important to note here that economics is obviously not the main discipline with expertise in habit formation. Economics introduces mathematical reasoning that can help one appreciate the importance of accounting for habits and outline high-level contours of the topic. But medical professionals should play a large role in developing real policies and community design choices.

As a final note, a lot of what we do online is habit forming. As people continually shift more of their activities online, where everything is built to ensnare attention, intentional and explicit focus on habit formation will become more important than ever. Kwon et al. (2016) do a good job presenting how to think about addiction to social media and online gaming.

When choosing to engage on a new social media platform, I know the experience will be habit forming and hard to stop once I start. Thus, I must carefully predict ahead of time whether the habit will provide me a net benefit. Once I’ve spent some time on the app and learned about the costs and benefits, I should summon all my rationality to determine whether to continue. Perhaps a pre-arranged accountability partner can help me make that determination. If it turns out I should quit the new app, strategies and resources are available to help me do it.

Written on February 24, 2023