Skip to content
Navigation
🏠Overview
Knowledge
🔬Scientific Foundation
🧠Critical Thinking
🤖AI and Technology
Debunking
🔮Esotericism and Occultism
🛐Religions
🧪Pseudoscience
💊Pseudomedicine
🕵️Conspiracy Theories
Tools
🧠Cognitive Biases
✅Fact Checks
❓Test Yourself
📄Articles
📚Hubs
Account
📈Statistics
🏆Achievements
⚙️Profile
Deymond Laplasa
  • Home
  • Articles
  • Hubs
  • About
  • Search
  • Profile

Knowledge

  • Scientific Base
  • Critical Thinking
  • AI & Technology

Debunking

  • Esoterica
  • Religions
  • Pseudoscience
  • Pseudomedicine
  • Conspiracy Theories

Tools

  • Fact-Checks
  • Test Yourself
  • Cognitive Biases
  • Articles
  • Hubs

About

  • About Us
  • Fact-Checking Methodology
  • Privacy Policy
  • Terms of Service

Account

  • Profile
  • Achievements
  • Settings

© 2026 Deymond Laplasa. All rights reserved.

Cognitive immunology. Critical thinking. Defense against disinformation.

  1. Home
  2. /Critical Thinking
  3. /Mental Errors
  4. /Cognitive Biases
  5. /Base Rate Neglect: Why 99% Test Accuracy...
📁 Cognitive Biases
✅Reliable Data

Base Rate Neglect: Why 99% Test Accuracy Can Mean 90% False Diagnoses

Base rate neglect is a cognitive bias where people ignore the statistical prevalence of a phenomenon, focusing only on specific information about a particular case. This leads to dramatic errors in medical diagnosis, legal decisions, cybersecurity, and risk assessment. Even a highly accurate test (99% accuracy) can produce 90% false positives if the tested condition is rare — but most people, including professionals, don't understand this. This article reveals the mathematical mechanism of the error, demonstrates the scale of the problem in real-world systems, and provides a self-assessment protocol.

📅
Published: February 22, 2026
⏱️
Reading time: 13 min

Neural Analysis

Neural Analysis
  • Topic: Base rate neglect (base rate fallacy) — a systematic error in probabilistic judgments where people ignore the statistical prevalence of a phenomenon
  • Epistemic status: High confidence — phenomenon replicated in experiments since the 1970s, has formal mathematical description through Bayes' theorem
  • Evidence level: Experimental studies, systematic reviews in decision-making psychology, applied research in medicine and cybersecurity (S009, S010, S011, S012)
  • Verdict: Base rate neglect is a proven cognitive bias with serious practical consequences. People systematically overestimate the probability of rare events when specific information is present, leading to errors in diagnosis, judicial decisions, and security threat assessment.
  • Key anomaly: Even professionals (physicians, judges, security analysts) make this error despite statistical training — intuition overrides formal logic
  • Test in 30 sec: When you see "test accuracy 99%" — ask: "What's the base rate of the phenomenon in the population?" If it's 1 in 10,000, most positive results will be false positives
Level1
XP0
🖤
Imagine: a test for a rare disease comes back positive. The test is 99% accurate. Your doctor says the error rate is only 1%. You panic and start searching for specialists. But the actual probability that you're truly sick might be just 9% — with a 91% chance the test was wrong. This isn't a quantum physics paradox or a philosophical puzzle. It's mathematical reality that even professionals ignore — and it leads to catastrophic decisions every day in medicine, criminal justice, cybersecurity, and risk assessment.

📌Base Rate Neglect: The cognitive blindness that turns precise instruments into error generators

Base rate neglect is a systematic cognitive bias in which people ignore the statistical prevalence of an event in the general population, focusing instead on specific information about a particular case (S001).

The Encyclopedia of Social Psychology defines this as an error in probabilistic judgment: individuals disregard information about event frequency in the population and rely on vivid but statistically less significant details. Learn more in the Sources and Evidence section.

The structure of error: Three components that create the illusion of precision

The classic structure includes three elements:

Base rate (prior probability)
The prevalence of an event in the population — for example, 0.1% of the population has disease X.
Test sensitivity (true positive rate)
The test correctly identifies 99% of sick individuals.
Test specificity (true negative rate)
The test correctly identifies 99% of healthy individuals.

The human mind intuitively focuses on sensitivity and specificity, perceiving "99% accuracy" as a guarantee, while completely ignoring the rarity of the disease itself.

Why 99% accuracy can mean 90% false alarms

Concrete example: a disease occurs in 1 person out of 1,000 (base rate 0.1%). The test has 99% sensitivity and 99% specificity. We test 100,000 people.

Group Number Test result Number of people
Sick (0.1%) 100 True positives (99%) 99
Healthy (99.9%) 99,900 False positives (1%) 999
Total positive results 1,098
Probability that a person with a positive result is actually sick: 99 / 1,098 ≈ 9%. Probability of false alarm: 999 / 1,098 ≈ 91%.

Boundaries of the phenomenon: From individual error to systemic problem

Base rate neglect isn't just an individual judgment error. Professionals fall victim to this mistake: doctors misinterpret screening test results, judges overestimate probability of guilt based on expert testimony, cybersecurity specialists generate avalanches of false alarms (S001).

The phenomenon manifests in any situation requiring integration of base statistical information with specific case data — from risk assessment to medical diagnosis and security threat evaluation.

Visualization of the diagnostic accuracy paradox when ignoring disease base rate
🧪 The diagram shows how 99% test accuracy with a disease base rate of 0.1% leads to 91% false positive results — a visualization of the mathematical reality that human intuition cannot see

🧱Seven Arguments That Make Base Rate Neglect So Convincing and Dangerous

Base rate neglect is not a result of stupidity. It's a consequence of deep features of human cognition that work efficiently in most situations, but create systematic distortions in the context of probabilistic judgments. More details in the Epistemology Basics section.

⚠️ Argument 1: Concrete Information Is Psychologically More Vivid Than Abstract Statistics

The brain evolved to process concrete, tangible events, not abstract distributions. Information that "the test showed a positive result specifically for you" is perceived as more relevant than abstract information that "in the population this disease is rare" (S001).

The psychological vividness of the specific case suppresses the statistical context—this is not a logical error, but a feature of attention architecture.

⚠️ Argument 2: Representativeness Dominates Over Probability in Intuitive Judgments

People assess the probability of an event not by its statistical frequency, but by how "representative" it is—how well it matches a prototype or stereotype (S002). If symptoms or test results "look like" a disease, the brain automatically increases the probability estimate, ignoring the base rate.

The representativeness heuristic is a fast but systematically biased way of judging that works against you in rare events.

⚠️ Argument 3: Professional Expertise Creates the Illusion That Base Rates Are "Already Accounted For"

Physicians, lawyers, and security analysts often believe that their experience automatically compensates for the need to explicitly account for base rates. The expert thinks: "I know this is a rare disease, but the symptoms are so specific that the base rate doesn't apply."

This is an illusion—the mathematics of Bayes' theorem doesn't depend on expert opinion about the "specificity" of the case (S004).

⚠️ Argument 4: Training Systems Focus on Test Accuracy, Not on Interpreting Results

Medical education teaches how to evaluate the sensitivity and specificity of diagnostic tests, but rarely trains the skill of integrating these metrics with the base rate of disease in a specific population. Cybersecurity specialists are trained to configure intrusion detection systems for maximum sensitivity, but not to minimize false positives accounting for the actual frequency of attacks.

Educational systems reproduce the error at an institutional level.

⚠️ Argument 5: Asymmetry of Consequences Creates Motivation to Ignore Base Rates

In medicine, missing a rare but dangerous disease is perceived as a more serious error than causing panic with a false positive diagnosis. In cybersecurity, missing a real attack is more catastrophic than generating thousands of false alarms.

Domain False Negative False Positive System Pressure
Medicine Patient won't receive treatment for rare disease Patient will undergo unnecessary examination Increase sensitivity
Cybersecurity Real attack will go undetected False alarm will distract analysts Increase sensitivity

This asymmetry creates institutional pressure toward "playing it safe"—increasing system sensitivity without accounting for the fact that with low base rates, this leads to an avalanche of false positives.

⚠️ Argument 6: Cascading Effects in Decision Chains Amplify the Original Error

A base rate error at one stage becomes input data for the next stage. A physician who receives a false positive screening test result orders a more invasive examination, which itself carries risks and may produce new false positives.

A security analyst responding to a false alarm from an intrusion detection system may interpret normal activity as suspicious, creating a cascade of erroneous conclusions.

⚠️ Argument 7: Lack of Feedback Makes the Error Invisible to Practitioners

A physician who referred a patient with a false positive result for additional examination rarely learns the final diagnosis—the patient goes to another specialist. A security analyst doesn't receive systematic feedback about how many of their alarms were false.

  1. Without explicit feedback, professionals cannot calibrate their intuitive probability assessments.
  2. The error reproduces endlessly, becoming embedded in routine.
  3. The practitioner remains convinced of the correctness of their approach because they don't see the full picture of consequences.

This creates a closed loop: the error remains invisible, so it's not corrected, so it reproduces again.

🔬Evidence Base: What Empirical Research Shows About Base Rate Neglect

The phenomenon of base rate neglect was first systematically described in a series of experiments by Kahneman and Tversky in the 1970s and has since been replicated in hundreds of studies across various contexts—from laboratory experiments to analyses of real professional decisions (S001).

📊 Classic Experiments: How People Ignore Statistics Even When Explicitly Presented

In Kahneman and Tversky's original study, participants were presented with a problem: "In a city, 85% of cabs are green and 15% are blue. A witness to an accident claims to have seen a blue cab. The witness's reliability has been tested: they correctly identify the color 80% of the time. What is the probability that the cab was actually blue?" The correct answer by Bayes' theorem: approximately 41%. Typical participant response: 80%—they completely ignored the base rate (85% green cabs) and focused only on the witness's reliability (S001).

People don't integrate information. They substitute a complex calculation with a simple rule: "The witness is 80% reliable—therefore, the answer is 80%." This isn't a calculation error. It's a refusal to calculate.

📊 Medical Diagnosis: Doctors Make the Same Error as Non-Specialists

A study in which doctors were presented with a task of interpreting mammography results showed widespread base rate neglect. Participants were told: the base rate of breast cancer in the screening population is 1%, mammography sensitivity is 90%, and the false positive rate is 9%. Question: what is the probability of cancer given a positive result? Correct answer: approximately 9%. Median doctor response: 75%. Most doctors overestimated the probability of cancer by a factor of 8, ignoring the low base rate (S002).

This isn't a competence problem. Doctors know statistics. The problem is that the availability heuristic and the concreteness of the clinical case outweigh abstract numbers. More details in the Media Literacy section.

📊 Cybersecurity: Avalanche of False Alarms as a Consequence of Ignoring Attack Base Rates

A systematic review of intrusion detection system (IDS) applications in cybersecurity showed that ignoring the base rate of actual attacks leads to a catastrophic ratio of false to true alarms (S004). With a typical attack base rate of 0.01% (1 attack per 10,000 events) and IDS sensitivity of 99%, a system with a 1% false positive rate will generate 100 false alarms for every real attack.

Parameter Value Consequence
Attack base rate 0.01% Attacks are rare
IDS sensitivity 99% Catches 99% of real attacks
False positives 1% 100 false per 1 real

Security analysts systematically underestimate the scale of this problem, focusing on the system's "high accuracy" (99%) and ignoring the rarity of actual attacks (S004).

📊 Legal System: Expert Testimony and Overestimation of Guilt Probability

Analysis of the use of probabilistic expert testimony in legal proceedings (e.g., DNA matches, ballistic analysis) showed that jurors and judges systematically overestimate the probability of guilt, ignoring the base rate of crimes in the population. If an expert reports that "the probability of a random DNA match is 1 in a million," jurors interpret this as "the probability of innocence is 1 in a million," completely ignoring the prior probability that a random person from the population committed the crime (S002).

Prosecutor's Fallacy
Confusion between P(match | guilty) and P(guilty | match). The former is close to 1, the latter depends on the base rate of crimes and other suspects.
Why This Is Dangerous
An innocent person can be convicted if their DNA happens to match DNA at the crime scene, and the court ignores that millions of people in the population have similar DNA.

🧾 Meta-Analysis: Robustness of the Effect Across Different Populations and Contexts

Meta-analysis of base rate neglect studies showed that the effect is robust across different cultures, age groups, and education levels (S001). Effect size varies depending on how information is presented: when the base rate is presented as natural frequencies (e.g., "10 out of 1,000") instead of percentages (e.g., "1%"), the error decreases but doesn't disappear completely.

  1. Natural frequencies reduce the error by 20–40%, but don't eliminate it
  2. Visualization (diagrams, graphs) helps more than text
  3. Even with optimal format, a significant portion of participants continue to ignore the base rate
  4. Education and experience weaken but don't cancel the effect

🔬 Neurocognitive Correlates: Which Brain Systems Are Involved in the Error

Neuroimaging studies showed that tasks requiring integration of base rate with specific information activate the dorsolateral prefrontal cortex—an area associated with working memory and cognitive control (S005). Participants who successfully account for the base rate demonstrate higher activation in this area, indicating that the correct solution requires suppression of the intuitive response and explicit analytical effort.

The correct answer requires cognitive resources that in real conditions are often unavailable due to load, stress, or time constraints. The error isn't stupidity. It's brain energy conservation that becomes dangerous in high-stakes situations.

This explains why groupthink amplifies base rate neglect: in groups, social pressure suppresses analytical effort even more strongly.

Comparative visualization of base rate errors among professionals and non-specialists
📊 The graph shows that doctors, lawyers, and security analysts commit base rate neglect errors at the same frequency as non-specialists—professional expertise does not protect against cognitive bias

🧠The Mechanism of Error: Why the Brain Systematically Ignores Base Rates

Base rate neglect is not a random error, but a systematic consequence of human cognitive architecture. Understanding the mechanism is critical for developing effective prevention strategies. More details in the section Cognitive Biases.

🧬 Representativeness Heuristic: Fast Judgment Instead of Slow Calculation

Kahneman and Tversky showed that people assess the probability of an event not through formal application of Bayes' theorem, but through the representativeness heuristic: "How much does A resemble B?" (S001). If a patient's symptoms "resemble" the typical presentation of a disease, the brain automatically increases the estimated probability of that disease, ignoring its rarity.

This heuristic works quickly and produces acceptable results in most situations, but systematically fails in situations with low base rates and highly specific information. Similarity to a prototype becomes stronger than statistical reality.

The brain asks: "What does this look like?" — not: "How often does this occur?"

🧬 System Competition: Intuitive System 1 vs. Analytical System 2

In terms of Kahneman's dual-system model, base rate neglect is the dominance of fast intuitive System 1 over slow analytical System 2. System 1 automatically generates an answer based on representativeness and information availability.

System 2 is capable of applying Bayes' theorem and accounting for base rates, but this requires explicit effort, time, and motivation. Under conditions of cognitive load, time pressure, or absence of an explicit signal for analytical thinking, System 2 does not activate, and the erroneous System 1 response dominates (S004).

System 1 (intuitive) System 2 (analytical)
Automatic, fast Requires effort, slow
Relies on similarity and availability Applies formal logic
Active by default Activates when explicitly needed
Ignores base rates Accounts for base rates

🔁 Framing Effect: How Information Format Modulates the Error

Research has shown that the format of probabilistic information presentation critically affects the frequency of base rate neglect (S002). When information is presented as percentages or probabilities ("1% of the population has the disease, the test is 99% accurate"), the error is maximal.

When the same information is presented as natural frequencies ("out of 1,000 people, 10 have the disease, the test correctly identifies 9 of them and falsely flags 10 healthy people"), the error significantly decreases. This indicates that the human brain is evolutionarily adapted to process frequencies, not abstract probabilities.

Natural Frequencies
Presenting information as concrete numbers from a population (e.g., "out of 1,000"). Activates System 2 and reduces base rate neglect by 50–70%.
Abstract Probabilities
Presenting as percentages or decimal fractions. Remains in System 1 mode, error is maximal.

🧬 Motivational Biases: When Desired Outcomes Influence Probability Assessment

Base rate neglect is amplified by motivational factors. If a person fears a particular disease, they tend to overestimate its probability even with low base rates and nonspecific symptoms.

If a security analyst is under pressure to "not miss an attack," they tend to interpret any anomaly as a threat, ignoring the low base rate of actual attacks. Motivation distorts not only information interpretation, but also the very willingness to engage analytical thinking (S001).

Fear and pressure don't just distort judgment — they shut down analytical thinking at its root.

The connection to the availability heuristic is direct here: motivationally significant events seem more frequent than they actually are, which further amplifies base rate neglect.

⚙️Conflicting Data and Confidence Boundaries: Where Evidence Diverges

The base rate effect is robust, but the conditions under which it manifests and methods to overcome it remain subjects of scientific disagreement. Three key debates reveal where evidence diverges and why no universal solution exists. More details in the Reality Check section.

Expertise: Shield or Illusion?

Experienced diagnostic physicians commit the base rate fallacy less often than novices (S009). But reframe the task in abstract terms—and the difference vanishes (S011).

Expertise works only if the professional has an explicit mental model for integrating base rates, and that model is activated by context. In atypical situations, experience offers no protection.

A physician accustomed to diagnostic protocols in their specialty may automatically account for disease prevalence. But if the task is framed as an abstract logical puzzle, their brain switches to "novice" mode—and the error returns.

Training: An Effect That Doesn't Hold

Brief training sessions on Bayes' theorem improve performance on subsequent tasks, but the effect doesn't transfer to new contexts and fades over time (S011). Intensive programs with repeated practice and feedback show more durable results, but require significant resources (S010).

Intervention Type Immediate Effect Transfer to New Contexts Durability Over Time
Brief training (explanation + examples) Present Weak Fades
Intensive program (practice + feedback) Present Stronger More durable

The problem: the brain learns context, not principle. Teaching someone to calculate using Bayes in a laboratory doesn't mean they'll do it in a doctor's office or when assessing risk at work.

Data Format: Natural Frequencies—Not a Panacea

Presenting information as natural frequencies (e.g., "10 out of 1,000" instead of "1%") consistently reduces the base rate fallacy (S011). But even with optimal formatting, 30–40% of participants continue to ignore base rates.

Natural Frequencies
A format that facilitates intuitive understanding of probabilities (e.g., "50 out of 10,000 patients"). Works better than percentages, but isn't universal.
Real-World Context
In medical protocols, safety reports, and financial documents, information is often presented as percentages or probabilities. Changing format requires systemic changes in documentation and training.

Even if you reformat data perfectly, the system in which that data circulates may work against you. A physician receives test results as natural frequencies, but the electronic health record requires input as percentages—and the cycle closes.

These three debates point to one thing: there's no universal cure. Each solution works under specific conditions and requires ongoing support. Ignoring base rates isn't simply a cognitive error that can be fixed with a single intervention. It's a systemic problem embedded in how we learn, how information is organized, and how we make decisions under pressure.

🧩Cognitive Anatomy of Manipulation: Which Biases Does Base Rate Neglect Exploit

Base rate neglect not only leads to unintentional errors, but can also be deliberately exploited to manipulate risk perception and decision-making. More details in the section Pharmaceutical Company Data Concealment.

⚠️ Exploitation Through Selective Presentation of Test Accuracy

Manufacturers of diagnostic tests, security systems, or machine learning algorithms often advertise "99% accuracy" or "high sensitivity" while remaining silent about the base rate of the event (S001). This is not an error—it's a strategy.

When the probability of an event is low (rare disease, rare breach), high test accuracy becomes an illusion of reliability. The consumer hears "99%" and ignores the context in which most positive results are false positives.

Manipulation works not because the information is false, but because it is incomplete. The fact remains a fact, but without the base rate it becomes a weapon.

🎯 Three Mechanisms of Exploitation

  1. Selective disclosure. They report sensitivity (proportion of true positives), but not specificity or positive predictive value.
  2. Emotional anchoring. "99% accuracy" sounds like a guarantee, activating the availability heuristic—a vivid number displaces statistical context.
  3. Social proof. When most people believe in test reliability (due to base rate neglect), groupthink reinforces the illusion (S004).

🔗 Connection to Other Cognitive Biases

Base rate neglect rarely operates in a vacuum. It intertwines with false dichotomy (the test either works or it doesn't), with confirmation bias (we seek facts supporting the initial positive result), and with disinformation, which deliberately conceals base rates.

The result: a system in which accurate instruments become error generators, and people become victims of their own inability to integrate statistical context.

⚔️

Counter-Position Analysis

Critical Review

⚖️ Critical Counterpoint

The Bayesian approach to diagnosis is a powerful tool, but its application requires understanding its boundaries. This is where the article's logic meets reality.

Overestimating the Universality of the Error

Ignoring the base rate is not an absolute cognitive defect. If the base rate is unknown, outdated, or irrelevant for a specific patient (for example, from a high-risk group), focusing on specific information may be rational. Critics of the Bayesian approach rightly point out: in real-world conditions, the "true" base rate is often unavailable or disputed.

Underestimating the Adaptiveness of Heuristics

The article presents ignoring the base rate as a pure defect, but researchers of ecological rationality (Gigerenzer) show that such heuristics can be adaptive in environments with high uncertainty or variable distributions. Focus on concrete information is often faster and more robust under conditions of limited time and resources.

The Problem of Choosing a Reference Group

The article assumes that the base rate in the general population is the correct prior. In reality, the choice of reference group is subjective: for a patient with symptoms, the relevant rate is not the general population frequency, but the frequency among people with such symptoms. "Ignoring the base rate" often means using a different, more specific base rate, rather than completely ignoring it.

Limitations of Laboratory Experiments

Most base rate neglect research has been conducted on artificial tasks with explicitly stated probabilities. In real-world conditions, people have additional information, experience, and context that may justify deviation from formal Bayesian logic. The ecological validity of classical experiments remains disputed.

Risk of the Reverse Error

Excessive focus on base rates leads to the reverse error—ignoring strong specific evidence. If a doctor sees clear symptoms of a rare disease but refuses the diagnosis due to low base rate, this is also an error. The Bayesian approach requires balance, not absolutization of the prior, and the article may underestimate this risk.

Context vs. Formal Logic

Real diagnosis is not a probabilistic task in a vacuum. A doctor works with medical history, physical examination, laboratory data, and clinical experience, which often contain information that cannot be formalized. Mechanical application of Bayesian logic without accounting for this context can lead to errors no less serious than ignoring the base rate.

Knowledge Access Protocol

FAQ

Frequently Asked Questions

It's a thinking error where you ignore how common something is overall and focus only on information about a specific case. For example, a test for a rare disease comes back positive — people think "I have the disease," ignoring that the disease occurs in 1 out of 10,000 people and the test gives false positives. As a result, even with 99% test accuracy, the actual probability of having the disease may be less than 10%. This is called base rate neglect — ignoring the baseline statistical frequency of a phenomenon in the population (S009, S011).
Because it leads to mass false diagnoses and unnecessary treatment. When a doctor or patient sees a positive result from a highly accurate test, intuition says "diagnosis confirmed." But if the disease is rare (low base rate), most positive results will be false — even with 99% test accuracy. This leads to unnecessary biopsies, chemotherapy, and psychological stress. Research shows that even doctors systematically overestimate the probability of disease after a positive test, ignoring the base rate (S011, S012).
Through Bayes' theorem. Probability of disease after positive test = (base rate × test sensitivity) / [(base rate × sensitivity) + (1 - base rate) × (1 - specificity)]. Example: disease in 0.01% of population (1 in 10,000), test is 99% accurate. Out of 10,000 people: 1 is sick and gets a positive result, 9,999 are healthy, but 1% (≈100 people) get false positives. Total 101 positive results, only 1 true — probability of disease with positive test ≈1%. People ignore the denominator of the formula — the number of healthy people who will also test positive (S009, S010, S011).
Yes, this is experimentally proven. Research shows that doctors, judges, and security analysts systematically ignore base rates, even with statistical training. In a classic experiment, doctors were given a problem with a rare disease and accurate test — most overestimated disease probability by 10-50 times. The reason: human intuition works poorly with conditional probabilities, the brain focuses on vivid specific information (test result) and ignores abstract statistics (base rate). This isn't about intelligence — it's an architectural feature of the cognitive system (S011, S012).
In cybersecurity, the judicial system, terrorist threat assessment, and financial risks. In cybersecurity: intrusion detection systems (IDS) with high accuracy generate enormous false alarms if real attacks are rare — analysts drown in false positives (S010). In courts: jurors overestimate the significance of DNA matches or other evidence, ignoring the base rate of innocence. In terrorism assessment: the rarity of actual terrorists makes most "suspicious signals" false, but security systems ignore this. Wherever a rare event is detected by an imperfect method, base rate neglect creates an avalanche of false positives (S009, S010).
Partially yes, but it requires conscious effort and formal tools. Research shows: when information is presented as natural frequencies (e.g., "out of 10,000 people, 1 is sick and 100 will get false positives") instead of percentages, people better understand actual probabilities. Visualization through probability trees or contingency tables helps. But the intuitive error remains — even trained people slip into base rate neglect under time pressure or emotions. Solution: checklists and Bayes calculators for critical decisions (S011, S012).
Because real attacks are extremely rare relative to total traffic volume — the base rate of attacks is very low. Even if an intrusion detection system (IDS) has 99% accuracy, it will generate enormous false alarms: out of a million events, 10 are real attacks, 9,990 healthy events will trigger false alarms at 1% error rate. Analysts receive 10,000 alerts, of which only 10 are true — 99.9% of their work is spent on false positives. This leads to "alert fatigue" and missing real threats. A 2022 article (S010) calls for deep analysis of false positives, not just true attacks, and using offline attack graphs to reduce noise.
Base rate neglect is a violation of Bayesian belief updating. Bayes' theorem shows how to correctly update the probability of a hypothesis when receiving new data: P(H|E) = P(E|H) × P(H) / P(E), where P(H) is the base rate (prior), P(E|H) is the probability of observation given true hypothesis, P(E) is the overall probability of observation. People focus on P(E|H) (test accuracy) and ignore P(H) (base rate) and P(E) (how many false positives). As a result, they don't update beliefs correctly — they overestimate the posterior probability P(H|E). This is a formal description of the cognitive error through Bayesian statistics (S009, S011).
Doctors, judges, security analysts, financial risk managers, journalists. Anywhere you need to assess the probability of rare events based on imperfect indicators. Doctors overestimate the probability of rare diseases after positive tests. Judges and jurors overestimate the significance of evidence (DNA, witness testimony), ignoring the base rate of innocence. Security analysts drown in false alarms from detection systems. Risk managers overestimate the probability of defaults or fraud. Journalists inflate rare events (terrorist attacks, plane crashes) into "epidemics," ignoring their statistical rarity. Common thread: all work with low base rates and high noise levels (S009, S010, S011).
Yes, use the "three-question rule." 1) What's the base rate of the phenomenon in the population? (How many out of 10,000 people have this?) 2) What's the accuracy of the indicator/test? (How many false positives per 10,000?) 3) How many false positives will there be per true positive? If you can't answer the first question — you're ignoring the base rate. Practical example: "Drug test is 95% accurate, showed positive result." Ask: how many out of 10,000 employees use drugs? If 100 (1%), then 100 will give true positives, and 9,900 healthy people will give 495 false positives (5% of 9,900). Total 595 positives, only 100 true — probability of actual drug use ≈17%, not 95%. This destroys intuition in 30 seconds (S009, S011).
Because evolution optimized the brain to work with concrete, vivid, emotionally significant events, not abstract statistics. The base rate is abstract knowledge about a population ('1 in 10,000'), while a test result is concrete information about you personally ('your test is positive'). The brain weighs concrete information more heavily than abstract information—this is the availability and representativeness heuristic. Additionally, working with conditional probabilities requires working memory and the prefrontal cortex—a slow, energy-intensive system. Intuition (Kahneman's fast System 1) ignores base rates because they don't 'feel' real. This isn't a bug—it's a feature that worked in small-group environments but breaks down in a world of large numbers and statistics (S011, S012).
Use natural frequencies instead of percentages and conditional probabilities. Instead of saying 'the test is 99% accurate, base rate is 0.01%,' say: 'Out of 10,000 people, 1 is sick. The test will find them. But the test will also give a false positive for 100 healthy people. If your test is positive, you're one of 101 people with a positive result, and only 1 of them is actually sick.' This is called the 'natural frequency format'—research shows it dramatically improves understanding. Visualizations also help: 2×2 tables (sick/healthy × test +/−), probability trees, pictograms (100 people icons, 1 red, 10 yellow). The key: make the base rate concrete and visual, not abstract (S011, S012).
Deymond Laplasa
Deymond Laplasa
Cognitive Security Researcher

Author of the Cognitive Immunology Hub project. Researches mechanisms of disinformation, pseudoscience, and cognitive biases. All materials are based on peer-reviewed sources.

★★★★★
Author Profile
Deymond Laplasa
Deymond Laplasa
Cognitive Security Researcher

Author of the Cognitive Immunology Hub project. Researches mechanisms of disinformation, pseudoscience, and cognitive biases. All materials are based on peer-reviewed sources.

★★★★★
Author Profile
// SOURCES
[01] Base rates: Both neglected and intuitive.[02] Presentation and content: The use of base rates as a continuous variable.[03] Molecular Dynamics Simulation of the Kinetics of Spontaneous Micelle Formation[04] The role of causality in judgment under uncertainty.[05] Decorrelation of Neural-Network Activity by Inhibitory Feedback[06] IMPACT OF SNAKE BITES AND DETERMINANTS OF FATAL OUTCOMES IN SOUTHEASTERN NEPAL[07] Direct numerical simulation of turbulent transport with uniform wall injection and suction[08] On near-wall turbulent flow modelling

💬Comments(0)

💭

No comments yet