Judging harmful actions as worse than equally harmful inactions, preferring to do nothing even when acting would be better.
Imagine two kids at a pool. One kid pushes another kid into the water, and the other kid sees someone falling into the water but doesn't grab their hand to help. The kid in the water gets equally wet both times, but almost everyone thinks the pusher is the 'bad' one, even though the kid who just watched and did nothing let the same bad thing happen.
Omission bias describes the systematic preference for harm caused by inaction over equal or lesser harm caused by action. People evaluate someone who actively does something harmful as more blameworthy, more causal, and more immoral than someone who passively allows the same harm to occur by doing nothing. This asymmetry persists even when the outcomes are objectively identical or when inaction produces demonstrably worse consequences. The bias operates in both self-interested decision-making (where people avoid acting to escape potential regret) and in moral judgments of others (where observers assign less blame to those who failed to act than to those who acted).
The same glitch looks different depending on the terrain. Finance, medicine, a relationship, a team — same mechanism, different costume.
Investors hold onto declining assets rather than actively selling at a loss, because the realized loss from selling feels like a self-inflicted wound, while the paper loss from inaction feels like an external event happening to them. This contributes to the disposition effect and portfolio inertia.
Physicians and patients systematically prefer watchful waiting over active treatment when both carry equivalent risk, because adverse outcomes from treatment are perceived as iatrogenic harm while adverse outcomes from non-treatment are perceived as the natural course of disease. This is particularly pronounced in vaccine hesitancy, where parents judge vaccine side effects as far worse than equivalent disease complications.
Teachers may avoid intervening when they notice a student being socially excluded, because actively restructuring group dynamics risks making things worse and being blamed for it, while passively allowing the exclusion to continue feels like a less culpable position.
People avoid having difficult but necessary conversations — about boundaries, unmet needs, or dealbreakers — because actively raising the issue risks causing a fight they'd feel responsible for, while letting the relationship slowly deteriorate through inaction feels like something that 'just happened.'
Product teams resist removing underperforming features or sunsetting legacy products because actively discontinuing something users rely on feels more harmful than passively allowing technical debt and fragmented user experiences to accumulate. Default opt-in settings exploit this bias, knowing users rarely take the active step of opting out.
Managers delay firing poor performers, restructuring teams, or canceling failing projects because taking action that visibly harms someone feels worse than allowing the broader harm of inaction to continue invisibly. Performance review systems that require active ratings ('rate this person') face more resistance than those with default categories.
Legislators vote against policies that would save more lives than they cost (e.g., mandatory safety regulations) because the visible victims of regulation feel like the legislator's fault, while the statistical victims of non-regulation are attributed to circumstance. Media coverage reinforces this by extensively covering harms caused by government action while under-reporting harms caused by government inaction.
Ilana Ritov and Jonathan Baron (1990) first studied the phenomenon empirically in vaccination decisions. Mark Spranca, Elisa Minsk, and Jonathan Baron formalized the concept as 'omission bias' in their 1991 paper in the Journal of Experimental Social Psychology.
In ancestral environments, novel actions carried unpredictable physical risks — trying an unknown food, confronting a predator, or altering a shelter. Inaction preserved the status quo, which had at least allowed survival thus far. A generalized heuristic of 'when in doubt, do nothing' reduced exposure to novel dangers. Furthermore, small-group social dynamics punished visible harmful actions more than passive failures, so a reputational incentive to avoid commissions over omissions was reinforced.
AI systems trained on human moral judgments inherit omission bias, generating recommendations that systematically favor inaction over action even when action would produce better outcomes. Language models asked to advise on moral dilemmas tend to endorse doing nothing more strongly than human participants. In content moderation, AI may be calibrated to over-penalize active harmful speech while under-detecting harmful silence or failure to correct misinformation, reflecting the human tendency to treat commissions as worse than omissions.
50 cards are free to preview. Buyers unlock the rest of the deck plus the interactive training — Spot-the-Bias Quiz unlimited, Swipe Deck with spaced repetition, My Blindspots, Decision Pre-Flight, the Printable Deck + Cheat Sheets, and the Field Guide e-book. $29.50$59.
