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

Reliable, rapid, and remote measurement of metacognitive bias

Metacognitive biases have been repeatedly associated with transdiagnostic psychiatric dimensions of 'anxious-depression' and 'compulsivity and intrusive thought', cross-sectionally. To progress our understanding of the underlying neurocognitive mechanisms, new methods are required to measure metacognition remotely, within individuals over time. We developed a gamified smartphone task designed to measure visuo-perceptual metacognitive (confidence) bias and investigated its psychometric properties across two studies (N = 3410 unpaid citizen scientists, N = 52 paid participants). We assessed convergent validity, split-half and test-retest reliability, and identified the minimum number of trials required to capture its clinical correlates. Convergent validity of metacognitive bias was moderate (r(50) = 0.64, p < 0.001) and it demonstrated excellent split-half reliability (r(50) = 0.91, p < 0.001). Anxious-depression was associated with decreased confidence (β =  - 0.23, SE = 0.02, p < 0.001), while compulsivity and intrusive thought was associated with greater confidence (β = 0.07, SE = 0.02, p < 0.001). The associations between metacognitive biases and transdiagnostic psychiatry dimensions are evident in as few as 40 trials. Metacognitive biases in decision-making are stable within and across sessions, exhibiting very high test-retest reliability for the 100-trial (ICC = 0.86, N = 110) and 40-trial (ICC = 0.86, N = 120) versions of Meta Mind. Hybrid 'self-report cognition' tasks may be one way to bridge the recently discussed reliability gap in computational psychiatry.

Some habits are more work than others: Deliberate self-regulation strategy use increases with behavioral complexity, even for established habits

OBJECTIVE

We tested the hypothesis that complex behaviors are commonly supported by self-regulation strategies, even when those behaviors are supported by strong instigation habits.

BACKGROUND

Goal-directed and habit-mediated processes arise from separable systems that have been suggested to seldomly interact.

RESULTS

Self-regulation strategy use was lower for habitually instigated simple behaviors compared to nonhabitually instigated simple behaviors. However, participants' use of self-regulation strategies increased with the increasing complexity of behaviors, even when complex behaviors were habitually instigated. The difference in the extent of strategy use between habitually and nonhabitually instigated actions was absent when behavioral complexity was particularly high.

CONCLUSION

These results point to a qualitative distinction-while simple behaviors may progress in a relatively automatic and unthinking manner, complex behaviors receive frequent support from self-regulation strategies, even if they are instigated habitually.

The temporal structure of goal-directed and habitual operant behavior

Operant behavior can reflect the influence of goal-directed and habitual processes. These can be distinguished by changes to response rate following devaluation of the reinforcing outcome. Whether a response is goal directed or habitual depends on whether devaluation affects response rate. Response rate can be decomposed into frequencies of bouts and pauses by analyzing the distribution of interresponse times. This study sought to characterize goal-directed and habitual behaviors in terms of bout-initiation rate, within-bout response rate, bout length, and bout duration. Data were taken from three published studies that compared sensitivity to devaluation following brief and extended training with variable-interval schedules. Analyses focused on goal-directed and habitual responding, a comparison of a habitual response to a similarly trained response that had been converted back to goal-directed status after a surprising event, and a demonstration of contextual control of habit and goal direction in the same subjects. Across experiments and despite responses being clearly distinguished as goal directed and habitual by total response rate, analyses of bout-initiation rate, within-bout rate, bout length, and bout duration did not reveal a pattern that distinguished goal-directed from habitual responding.

Reduction of Financial Health Incentives and Changes in Physical Activity

Importance

Governments and others continue to use financial incentives to influence citizen health behaviors like physical activity. However, when delivered on a population scale they can be prohibitively costly, suggesting more sustainable models are needed.

Objectives

To evaluate the association of incomplete financial incentive withdrawal ("schedule thinning") with physical activity after more than a year of incentive intervention and to explore whether participant characteristics (eg, app engagement and physical activity) are associated with withdrawal outcomes.

Design, Setting, and Participants

This case-control study with a pre-post quasi-experimental design included users of a government-funded health app focused on financial incentives. Eligible participants were residents in 3 Canadian provinces over 25 weeks in 2018 and 2019. Data were analyzed from July 2021 to December 2022.

Exposure

Due to fiscal constraints, financial incentives for daily physical activity goals were withdrawn in Ontario in December 2018 (case)-representing a 90% reduction in incentive earnings-but not in British Columbia or Newfoundland and Labrador (controls).

Main Outcome and Measures

The primary outcome was objectively assessed weekly mean daily step count. Linear regression models were used to compare pre-post changes in daily step counts between provinces (a difference-in-differences approach). Separate models were developed to examine factors associated with changes in daily step count (eg, app engagement and physical activity). Clinically meaningful initial effect sizes were previously reported (approximately 900 steps/d overall and 1800 steps/d among the physically inactive).

Results

In total there were 584 760 study participants (mean [SD] age, 34.3 [15.5] years; 220 388 women [63.5%]), including 438 731 from Ontario, 124 101 from British Columbia, and 21 928 from Newfoundland and Labrador. Significant physical activity declines were observed when comparing pre-post changes in Ontario to British Columbia (-198 steps/d; 95% CI, -224 to -172 steps/d) and Newfoundland and Labrador (-274 steps/d; 95% CI, -323 to -225 steps/d). The decrease was most pronounced for highly engaged Ontario users (-328 steps/d; 95% CI, -343 to -313 steps/d). Among physically inactive Ontario users, physical activity did not decline following withdrawal (107 steps/d; 95% CI, 90 to 124 steps/d).

Conclusions and Relevance

In this case-control study of incomplete financial incentive withdrawal, statistically significant daily step count reductions were observed in Ontario; however, these declines were modest and not clinically meaningful. Amidst substantial program savings, the physical activity reductions observed here may be acceptable to decision-makers working within finite budgets.

The Impact of Emotions on Habitual Inhibition

Emotional information prioritizes human behavior. How much emotions influence ongoing behavior critically depends on the extent of executive control functions in a given context. One form of executive control is based on stimulus-stop associations (i.e., habitual inhibition) that rapidly and effortlessly elicits control over the interruption of ongoing behavior. So far, no behavioral accounts have explored the emotional impact on habitual inhibition. We aimed to examine the emotional modulation on habitual inhibition and associated psycho-physiological changes. A go/no-go association task asked participants to learn stimulus-stop and stimulus-response associations during 10-day training to form habitual inhibition (without emotional interference). Probabilistic feedback guided learning with varying probabilities of congruent feedback, generating stronger versus weaker pairings. A reversal test measured habitual inhibition strength counteracted by emotional cues (high-arousal positive and negative stimuli compared with neutral ones). Our training protocol induced stable behavioral and psycho-physiological responses compatible with habitual behavior. At reversal, habitual inhibition was evident as marked by significant speed costs of reversed no-go trials for strongly associated stimuli. Positive and negative emotional cues produced larger impact on habitual inhibition. We report first evidence on a cognitive control mechanism that is vulnerable to emotional stimuli and suggest alternative explanations on how emotions may boost or counteract certain behavioral abnormalities mediated by habitual inhibition.