Forecast gap · what you predict vs what you feel
Restraint Bias
The tendency for people to overestimate their capacity for impulse control, leading them to overexpose themselves to temptation and ultimately exhibit more impulsive behavior.
The forecast reaches for the dashed ceiling; reality settles at the solid line. The hatched shortfall is invisible in advance — you brace for the whole column and feel a fraction of it.
Why the forecast misses
Restraint bias operates through a 'cold-hot empathy gap' (Loewenstein, 1996): when in a calm, non-impulsive state, people underestimate the motivational force of visceral drives (hunger, craving, arousal, fatigue). This forecasting error produces overconfidence in self-restraint, which then leads people to choose more tempting environments, creating a self-fulfilling prophecy of impulsive behavior.
Competing account — The bias is mechanistically grounded in the broader empathy gap and projection bias literature; competing accounts emphasize present-state anchoring in affective forecasting rather than a unique bias about self-control per se.
Forecasts on record
Smokers in the high-control condition chose significantly higher temptation levels than those in the low-control condition, and were approximately 3 times more likely to actually smoke during the task.
Nordgren, van Harreveld, & van der Pligt · 2009 · n = 53 smokers
Satiated participants chose more tempting snacks than hungry participants, overestimating their future impulse control. Consequently, the satiated group was less likely to return the snack uneaten.
Nordgren, van Harreveld, & van der Pligt · 2009 · n = 79
Participants with more inflated impulse-control beliefs at baseline exposed themselves to more temptation over the follow-up period and had higher rates of smoking relapse at 4 months.
Nordgren, van Harreveld, & van der Pligt · 2009 · n = 55 recent quitters
Evidence — limited-evidence. The original paper reported four experiments supporting the restraint bias, but the concept has received limited independent direct replication. Related phenomena—especially the empathy gap and hot-cold empathy gaps in affective forecasting—are well-replicated. The smoking-relapse finding (Study 4) was longitudinal but has not been independently replicated to our knowledge. The bias is widely cited in applied and popular contexts but lacks meta-analytic validation.
Where the gap cost something
medicine · 2009
Smoking relapse among overconfident quitters
In Nordgren et al.'s Study 4 (2009), recent quitters who rated themselves as having high impulse control exposed themselves to more smoking-related temptation and showed higher rates of relapse at 4-month follow-up, demonstrating the real-world health consequences of restraint bias.
medicine · 2009
Obesity and unhealthy lifestyle prevalence
Nordgren and colleagues argued in their 2009 paper and press coverage that restraint bias contributes to societal-level obesity and addiction crises: people overestimate their ability to resist food and drug temptations, leading them to expose themselves to environments that promote unhealthy consumption.
To forecast honestly
Adopt a 'humility about willpower' strategy: avoid tempting situations entirely rather than testing your self-control, and pre-commit to choices when in a calm state.
Nordgren et al. (2009) found that participants who were led to have more realistic self-control beliefs chose lower levels of temptation and were less likely to succumb. The key intervention is situational avoidance rather than reliance on restraint.
Catch it in the act
Watch for moments when you think 'I can handle being around temptation' while in a calm state—such as a recovering smoker entering a smoky bar, a dieter walking past a bakery, or a tired student planning to cram the night before an exam.
Related forecasting errors
- Empathy GapparentRestraint bias stems from Loewenstein's empathy gap research: people in a 'cold' state fail to appreciate how a 'hot' impulsive state will influence their behavior.
- Projection BiassiblingBoth involve mispredicting future preferences or behavior based on current states; restraint bias is specific to predictions about impulse control.
- Overconfidence Effectmechanistically-linkedRestraint bias can be viewed as a domain-specific form of overconfidence applied to self-regulatory capacity.