Donning your exercise attire and making your way to the fitness center can initially seem laborious. In due time, you may establish a routine of hitting the gym and effortlessly attending your Zumba class or hopping on the treadmill for a run. Recent research conducted by social scientists at Caltech has revealed that it takes approximately six months on average to cultivate the habit of going to the gym.

The same study also looked at how long it takes health care workers to get in the habit of washing their hands: an average of a few weeks.

Anastasia Buyalskaya (PhD '21), currently an assistant professor of marketing at HEC Paris, asserts that there is no fixed timeframe for habit formation. The study, published in the Proceedings of the National Academy of Sciences, was conducted by Buyalskaya, along with Colin Camerer, Robert Kirby Professor of Behavioral Economics and director and leadership chair of the T&C Chen Center for Social and Decision Neuroscience at Caltech, and researchers from the University of Chicago and the University of Pennsylvania. Additionally, Xiaomin Li (MS '17, PhD '21), who previously served as a graduate student and postdoctoral scholar at Caltech, is also listed as an author.

Camerer explains that the commonly held belief that it takes 21 days to form a habit lacks scientific basis. "Our research reinforces the notion that the timeline for habit formation varies depending on the specific behavior and various other factors," Camerer states.

This groundbreaking study marks the first time machine learning tools have been utilized to investigate habit formation. The researchers leveraged machine learning to analyze extensive datasets from tens of thousands of individuals who were either swiping their badges to access the gym or engaging in hand-washing during hospital shifts. Collaborating with 24 Hour Fitness for the gym research and a company utilizing radio frequency identification (RFID) technology to monitor hand-washing in hospitals, the data sets encompassed over 30,000 gymgoers tracked over four years and more than 3,000 hospital workers across nearly 100 shifts.

Buyalskaya elaborates on the power of machine learning in this context, stating, "Through machine learning, we can capture and analyze numerous contextual variables that may have predictive value for behavioral execution. It allows us to identify relevant variables without necessarily starting with a predefined hypothesis. The machine learning algorithms do the heavy lifting in identifying key factors for us."

Machine learning also let the researchers study people over time in their natural environments; most previous studies were limited to participants filling out surveys.

The study's findings revealed that certain variables, such as time of day, did not significantly impact habit formation when it came to going to the gym. However, other factors, such as an individual's past behavior, played a crucial role. For instance, for 76 percent of gymgoers, the duration of time that had elapsed since their last gym visit was a key predictor of whether they would return. In other words, the longer the gap between gym visits, the less likely the individual was to establish a habit of going to the gym. Additionally, 69 percent of gymgoers were more inclined to visit the gym on consistent days of the week, with Monday and Tuesday being the most frequented days.

In the hand-washing segment of the study, the researchers analyzed data from healthcare workers who were provided with new RFID badges to record their hand-washing activity as per new requirements. Buyalskaya notes, "While it's possible that some healthcare workers already had established hand-washing habits prior to our observation, we treated the introduction of the RFID technology as a 'shock' and assumed that they may need to rebuild their habit from the moment they started using the technology." This approach allowed the researchers to account for any potential changes in habit formation due to the implementation of the new technology.

“Overall, we are seeing that machine learning is a powerful tool to study human habits outside the lab,” Buyalskayasays. (PB/Newswise)