Normalcy bias is the tendency to underestimate the probability or severity of a threatening event, even when warning signs are obvious. It is sometimes described as the mental state people enter when they interpret danger signals through the lens of their everyday experience, concluding that things will continue as normal because they usually have. The bias shows up across a strikingly wide range of situations, from building fires and earthquakes to pandemics and heart attacks, and its effects can delay life-saving action by minutes or longer.
What Normalcy Bias Actually Looks Like
The classic image of someone panicking in an emergency turns out to be far less common than someone doing nothing. In building fires, when occupants encounter ambiguous cues like a faint smell of smoke or a muffled alarm, research on fire evacuation behavior has found that normalcy bias tends to persist longer while people remain inside the building, because they interpret the cues as part of everyday life rather than a genuine threat.1PubMed Central. Risk perception in fire evacuation behavior revisited: definitions, related concepts, and empirical evidence People might assume the alarm is a drill, or that the smell is coming from a kitchen on another floor. They wait. They look around. They continue eating lunch.
This same pattern emerges during disaster warnings. A study examining why residents failed to comply with disaster warnings found that cognitive factors, including prior experience and normalcy bias, were more critical to the decision not to evacuate than social factors alone.2Proceedings of the Human Factors and Ergonomics Society Annual Meeting. Evaluating the Interactive Social-Cognitive Model for Explaining Non-Compliance to a Disaster Warning In other words, people who had weathered a previous storm without serious damage often assumed the next one would be the same. Their personal history became a filter that screened out new information suggesting this time was different.
The term itself was framed alongside a counterpart. Disaster researcher Thomas Drabek described normalcy bias as one of two errors people make: while normalcy bias leads to underestimating disruption, its opposite, the “abnormalcy bias,” leads to overestimating how badly victims will cope.3PubMed. The continuity principle: a unified approach to disaster and trauma Emergency planners and outside observers often assume people will fall apart, while the people themselves assume nothing bad will happen. Both sides get it wrong, just in opposite directions.
Why Your Brain Defaults to “Everything Is Fine”
Part of the explanation is neurological. The brain has a built-in defense sequence that activates in stages when danger appears. Arousal comes first. Then active responses like fleeing or fighting kick in. But between arousal and action, there is a phase researchers call freezing, which is distinct from paralysis. During freezing, a specific brain region called the ventrolateral periaqueductal grey acts as a brake on the body’s threat-response systems, holding fight-or-flight reactions in pause. The body is actually primed for action during this phase: muscle tone increases and physiological arousal is high. But motor output is suppressed.4PubMed Central. Freeze for action: neurobiological mechanisms in animal and human freezing Think of it as the engine revving with the parking brake on. The system is ready to move, but something has to release the brake first.
This freezing response is one stage in what researchers describe as a defense cascade, a continuum of innate, automatically activated defense behaviors that all humans share. The sequence runs from arousal through flight-or-fight, then freezing, and in extreme cases, tonic immobility or collapse.5PubMed Central. Fear and the Defense Cascade: Clinical Implications and Management Normalcy bias maps onto the early stages of this cascade, the period where arousal has begun but the brain has not yet committed to active defense. The person senses that something is off but has not crossed the threshold into action.
There is also an evolutionary dimension to the brain’s approach to uncertainty. From a neurobiological perspective, the stress response is not something that gets switched on by danger; it is always running in the background and gets actively suppressed when you feel safe. The prefrontal cortex keeps the stress system under a kind of tonic inhibition as long as you perceive your environment as safe.6Journal of Anxiety Disorders. The default response to uncertainty and the importance of perceived safety in anxiety and stress: An evolution-theoretical perspective When that sense of safety wavers, the stress response is disinhibited, essentially unmasked rather than newly created. This means the brain’s default reaction to ambiguity is not alarm but rather a struggle between maintaining its usual “everything is safe” signal and acknowledging that the situation has changed. In an unfamiliar emergency, the “safe” signal can persist for a dangerously long time simply because nothing in your recent experience has taught your brain that this particular cue means danger.
How Other People Make It Worse
Normalcy bias does not happen in isolation. When you are surrounded by other people who also appear calm, your own bias gets reinforced. Research on behavioral responses to earthquake shaking has confirmed that people look to those around them for cues in novel or uncertain situations, including emergencies. Rather than acting immediately, people often hesitate because they are unsure how to respond given how rarely earthquakes occur, and in that hesitation they scan the room for social guidance.7Natural Hazards. Social influences on behavioural response to earthquake shaking: evidence from CCTV footage
This creates a feedback loop. Everyone waits to see what everyone else does. Everyone else is also waiting. The result is a room full of people who all look calm because nobody has started reacting yet, which each person individually reads as confirmation that the situation must be normal. Sociologists who study disasters have long noted this dynamic: the inaction of bystanders functions as a false signal that the threat is not real. One person breaking the cycle and moving toward an exit can shift the entire group’s behavior, which is part of why trained personnel are disproportionately valuable in emergencies.
The study of disaster warning non-compliance noted that while normalcy bias and other cognitive factors were the primary drivers of people’s decisions to stay put, social factors did offer opportunities for modifying the perceived context in ways that motivated compliance.8Proceedings of the Human Factors and Ergonomics Society Annual Meeting. Evaluating the Interactive Social-Cognitive Model for Explaining Non-Compliance to a Disaster Warning Translation: you might not be able to talk someone out of their normalcy bias through argument alone, but if you change the social environment they are reading cues from, such as by having a trusted authority figure visibly take action, you can shift the equation.
The Panic Myth
Pop culture has trained people to expect stampedes, hysteria, and mass disorder in emergencies. The actual research tells a different story. A study of public behavior across five terrorist incidents found that people are not prone to panic. Instead, they tend to modify their behavior in ways that are roughly proportionate to the perceived risk, such as avoiding unnecessary travel or taking precautionary measures.9PubMed. Terrorism and dispelling the myth of a panic prone public
This matters for understanding normalcy bias because the two concepts are often confused. People assume the main danger in emergencies is that crowds will lose control. In reality, the far more common and often more lethal problem is that people will underreact. They will stay seated, continue conversations, wait for official instructions, or convince themselves the threat is not serious. The real risk is not panic but paralysis dressed up as calm.
This mismatch between expectation and reality has consequences for how emergency plans are designed. If planners assume the public will panic, they may withhold information or delay warnings to avoid “causing a stampede.” But if the actual danger is inaction and underestimation, then clearer, earlier, more alarming warnings are exactly what is needed. The panic myth and normalcy bias work together in a destructive way: officials hold back information because they fear panic, and the public interprets the absence of alarm as confirmation that things are fine.
Slow-Moving Threats and Optimism Bias
Normalcy bias is easiest to spot in sudden emergencies like fires and earthquakes, but it operates just as powerfully in slow-moving crises where the threat builds gradually. The COVID-19 pandemic offered a massive natural experiment. A study across European populations during the first wave found that people displayed what researchers call comparative unrealistic optimism: they believed they were less likely than the average person to be infected. That optimism was not harmless. It was negatively associated with the adoption of protective behaviors like hand hygiene, mask wearing, and social distancing.10PubMed Central. Are People Optimistically Biased about the Risk of COVID-19 Infection? Lessons from the First Wave of the Pandemic in Europe
Optimism bias and normalcy bias are not identical, but they overlap heavily in practice. Normalcy bias says “this threat probably is not as bad as they say.” Optimism bias adds “and even if it is, it probably won’t happen to me.” Together, they produce a mindset that resists behavioral change even when the evidence for risk is abundant. During the pandemic, this showed up as people acknowledging the virus existed and was dangerous to others while simultaneously concluding they personally did not need to alter their routines.
The same dynamic appears in financial markets, where investors collectively underreact to early signs of a downturn, and in climate-related planning, where communities in flood zones or fire corridors resist evacuation or preparation because previous seasons passed without major damage. In each case, the mechanism is the same: past normalcy is treated as evidence that future normalcy is guaranteed.
When Minutes Count in a Medical Emergency
One of the most consequential settings for normalcy bias is the onset of a medical emergency, particularly a heart attack. Chest pain, shortness of breath, and radiating discomfort are symptoms most people have at least heard about, yet the average time between symptom onset and calling for emergency help is often far longer than doctors would like. Part of this delay is normalcy bias. People interpret early symptoms as indigestion, muscle strain, or stress. They wait to see if it passes. They do not want to “overreact” or feel embarrassed by a false alarm.
A recent case report documented an interesting workaround. A patient experiencing symptoms of a heart attack consulted a publicly available generative AI system, which responded with a clear and unambiguous instruction to contact emergency medical services immediately. The patient did so and received timely treatment. The case report suggests that AI-generated language can help cut through normalcy bias by delivering direct, confident guidance that removes the ambiguity people otherwise use to justify waiting.11PubMed Central. Overcoming Normalcy Bias in Acute Myocardial Infarction: A Case Report of Generative AI as a Behavioral Catalyst for Emergency Care Seeking This is a single case, not a controlled trial, so it would be premature to draw broad conclusions. But it points to something practical: normalcy bias thrives on ambiguity, and tools that reduce ambiguity, whether human or digital, can help people act faster.
The medical setting also highlights how normalcy bias interacts with identity and social expectations. People do not want to be the person who calls an ambulance for heartburn. That social fear layers on top of the cognitive bias, creating a double barrier. The patient is already telling themselves “it’s probably nothing,” and on top of that, they are imagining the embarrassment if it turns out to be nothing. This combination is a significant contributor to delayed treatment in cardiac events, stroke, and other time-sensitive emergencies.
What Actually Helps People Snap Out of It
If normalcy bias is a default cognitive mode rather than a character flaw, the question becomes how to interrupt it. The evidence points to a few approaches that work, and one popular intuition that largely does not.
The approach that does not work well is simply providing more information. Telling people about the statistical likelihood of a disaster or explaining the science behind a threat rarely changes behavior on its own. People do not fail to evacuate because they lack data. They fail to evacuate because their brain is actively filtering that data through a lens that says “this is probably normal.”
What does help:
- Prior experience with real events: People who have lived through a genuine emergency, not a drill but an actual event where they were personally affected, tend to respond faster the next time. The disaster warning study found that prior experience was one of the most critical cognitive factors in determining whether someone complied with a warning.12Proceedings of the Human Factors and Ergonomics Society Annual Meeting. Evaluating the Interactive Social-Cognitive Model for Explaining Non-Compliance to a Disaster Warning Their previous experience has recalibrated the brain’s sense of what “normal” includes.
- Clear, specific, and personalized warnings: Vague messages like “be prepared” are easy to dismiss. Warnings that name a specific threat, describe what will happen, and tell you exactly what to do right now are harder to fold into a narrative of normalcy. The more specific the instruction, the less room the brain has to reinterpret the situation as routine.
- Visible action by others: Because social cues play such a powerful role in emergency behavior, seeing someone else act, especially someone perceived as authoritative, can break the feedback loop of collective inaction. This is one reason trained floor wardens in office buildings and community emergency responders are so effective: they provide a visible model of appropriate behavior that short-circuits the “everyone else seems calm” signal.
- Pre-commitment and mental rehearsal: People who have mentally walked through what they would do in an emergency, visualizing the steps rather than just reading a checklist, tend to act faster when the real event arrives. This works because it pre-loads a behavioral script that competes with the default “do nothing” response.
The common thread is that normalcy bias is not overcome through willpower or intelligence. It is overcome through experiences, cues, and preparation that shift the brain’s calculation of what counts as normal. If your brain has a memory, a social signal, or a rehearsed plan that says “this is the kind of situation where you act,” the freezing phase shortens and action begins sooner.
Risk Perception Is Personal and Emotional
One reason normalcy bias is so stubborn is that risk perception itself is not a purely rational calculation. A review of fire evacuation research characterized risk perception as the personalization of risk related to a current event, influenced by emotions and prone to cognitive biases.13PubMed Central. Risk perception in fire evacuation behavior revisited: definitions, related concepts, and empirical evidence In other words, knowing that a building is on fire and feeling personally threatened by that fire are two different psychological states, and only the second one reliably produces action.
This gap between knowing and feeling is where normalcy bias lives. You can be fully aware that a warning has been issued, that the sirens are sounding, that other people are leaving. But if the threat has not become emotionally real to you, if it has not crossed from abstract knowledge to felt danger, the bias persists. This is why the most effective emergency communications are not just informative but vivid and concrete. A message that says “flooding may occur” is easy to dismiss. A message that says “water will reach the second floor of your home within two hours” is much harder to fold into a story about everything being fine.
The processing of threat-related information itself occurs in stages, starting with a rapid, automatic detection of hazard information followed by a slower, conscious integration of that information.14PubMed Central. Are We Sensitive to Different Types of Safety Signs? Evidence from ERPs Normalcy bias can intervene at the second stage. Your brain may detect the threat signal quickly, but the conscious processing that follows can reinterpret that signal as non-threatening if your mental model of “normal” is strong enough. The automatic detection says “something is wrong.” The conscious interpretation says “but everything has always been fine, so this is probably nothing.” In the time it takes for that internal debate to resolve, critical minutes can pass.
This two-stage process also explains why some people seem immune to normalcy bias in their professional domains but not in their personal lives. A firefighter entering a burning building processes smoke cues differently than a civilian, not because the firefighter’s eyes work differently, but because years of training and experience have reshaped the conscious interpretation stage. The automatic detection is roughly the same for everyone. The difference is in what the brain does with that detection next.