Dataset Information

What can machine learning teach us about habit formation? Evidence from exercise and hygiene.

ABSTRACT: We apply a machine learning technique to characterize habit formation in two large panel data sets with objective measures of 1) gym attendance (over 12 million observations) and 2) hospital handwashing (over 40 million observations). Our Predicting Context Sensitivity (PCS) approach identifies context variables that best predict behavior for each individual. This approach also creates a time series of overall predictability for each individual. These time series predictability values are used to trace a habit formation curve for each individual, operationalizing the time of habit formation as the asymptotic limit of when behavior becomes highly predictable. Contrary to the popular belief in a "magic number" of days to develop a habit, we find that it typically takes months to form the habit of going to the gym but weeks to develop the habit of handwashing in the hospital. Furthermore, we find that gymgoers who are more predictable are less responsive to an intervention designed to promote more gym attendance, consistent with past experiments showing that habit formation generates insensitivity to reward devaluation.

SUBMITTER: Buyalskaya A

PROVIDER: S-EPMC10151500 | biostudies-literature | 2023 Apr

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

What can machine learning teach us about habit formation? Evidence from exercise and hygiene.

Buyalskaya Anastasia A Ho Hung H Milkman Katherine L KL Li Xiaomin X Duckworth Angela L AL Camerer Colin C

Proceedings of the National Academy of Sciences of the United States of America 20230417 17

We apply a machine learning technique to characterize habit formation in two large panel data sets with objective measures of 1) gym attendance (over 12 million observations) and 2) hospital handwashing (over 40 million observations). Our Predicting Context Sensitivity (PCS) approach identifies context variables that best predict behavior for each individual. This approach also creates a time series of overall predictability for each individual. These time series predictability values are used t ...[more]

PMID: 37068252

Similar Datasets

Project description:Simple Summary Horse training exposes horses to many difficult learning challenges, including tasks that run counter to their natural behaviour and research from neuroscience can be used to understand how horses respond to these learning challenges. Horses can also quickly develop habits, however if the habit is undesirable, the horse may be subjected to retraining where it must learn to suppress the unwanted behaviour and learn a new response. This is called cognitive flexibility and is underpinned by complex neural processes and can be impaired by exposure to repeated or chronic stress. We trained horses to acquire habit-like responses in one of two industry-style aversive instrumental learning scenarios (moving away from the stimulus-instinctual-IS or moving towards the stimulus-non-instinctual-NIS) and evaluated the effect of repeated stress exposures on their cognitive flexibility in a reversal task. We measured heart rate, salivary cortisol and serum brain derived neurotrophic factor to infer possible neural correlates of the learning outcomes. Horses trained in the NIS task took longer to learn task than the horses trained in the IS task, however they were quicker to learn the IS task during the reversal. There was no significant effect of the stress exposures on cognitive flexibility, however the stress did make horses more sensitive to the aversive training stimuli. These results provide guidance for trainers to adapt their practices when training tasks that are difficult for horses to learn and during retraining. In particular, even when horses have been trained to approach aversive stimuli they can quickly learn to avoid them. This has implications for retraining unwanted responses during aversive situations such as trailer loading. Abstract Horse training exposes horses to an array of cognitive and ethological challenges. Horses are routinely required to perform behaviours that are not aligned to aspects of their ethology, which may delay learning. While horses readily form habits during training, not all of these responses are considered desirable, resulting in the horse being subject to retraining. This is a form of cognitive flexibility and is critical to the extinction of habits and the learning of new responses. It is underpinned by complex neural processes which can be impaired by chronic or repeated stress. Domestic horses may be repeatedly exposed to multiples stressors. The potential contribution of stress impairments of cognitive flexibility to apparent training failures is not well understood, however research from neuroscience can be used to understand horses’ responses to training. We trained horses to acquire habit-like responses in one of two industry-style aversive instrumental learning scenarios (moving away from the stimulus-instinctual or moving towards the stimulus-non-instinctual) and evaluated the effect of repeated stress exposures on their cognitive flexibility in a reversal task. We measured heart rate as a proxy for noradrenaline release, salivary cortisol and serum Brain Derived Neurotrophic Factor (BDNF) to infer possible neural correlates of the learning outcomes. The instinctual task which aligned with innate equine escape responses to aversive stimuli was acquired significantly faster than the non-instinctual task during both learning phases, however contrary to expectations, the repeated stress exposure did not impair the reversal learning. We report a preliminary finding that serum BDNF and salivary cortisol concentrations in horses are positively correlated. The ethological salience of training tasks and cognitive flexibility learning can significantly affect learning in horses and trainers should adapt their practices where such tasks challenge innate equine behaviour.

Project description:BackgroundInterventions aimed at promoting physical activity (PA) behavior through habit formation pathways are gaining popularity, as they differ from conventional interventions that rely on intention pathways. Past research has established a positive correlation between PA habits and behavior. However, the efficacy of current interventions designed to form PA habits and improve PA automaticity is not yet fully ascertained. Additionally, the intervention components that significantly impact the effectiveness of these interventions are yet to be determined.MethodsThis systematic review was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines. We conducted a search of three databases (PubMed, Embase, and Cochrane Library) from January 2000 to December 2022, with a focus on interventions for developing PA habits. Two independent authors conducted paper selection, quality assessment, data extraction, and coding of behavior change techniques (BCTs). The effect size of interventions was calculated using standardized mean difference. Subgroup analyses were carried out based on follow-up duration, delivery method, sample characteristics, and theory. Furthermore, we employed meta-regression to investigate the association between BCTs and PA habits.ResultsTen eligible studies with relatively high quality were included in the final data set. Characteristics of studies varied in intervention sample and delivery way. The habit formation interventions significantly increased PA habit (SMD = 0.31, 95% CI 0.14-0.48, P < .001) compared to the control groups. Subgroup analysis demonstrated that the duration of follow-up ≤ 12 weeks have a higher effect size on PA habit than the duration > 12 weeks. Meta-regression revealed that problem solving has a significant positive association with effectiveness improvement (β = 0.36, 95% CI 0.17-0.55), while social reward is linked with a reduction in effectiveness (β = -0.40, 95% CI -0.74-0.06).ConclusionsOur findings reveal that habit formation interventions are effective in fostering PA habit. Future studies could leverage the insights form this study to optimize the intervention design and achieve better effectiveness.

Dataset Information

What can machine learning teach us about habit formation? Evidence from exercise and hygiene.

Publications

What can machine learning teach us about habit formation? Evidence from exercise and hygiene.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets