Base-Rate Neglect

Base-rate neglect is a cognitive bias in which people ignore prior probabilities — the "base rates" — when making judgments under uncertainty, relying instead on individuating information. The classic demonstration is the Linda problem: subjects are given base-rate information about a population and then ignore it when evaluating a specific case. The bias is typically attributed to the representativeness heuristic — the tendency to judge probability by similarity to a prototype rather than by statistical logic. But from the perspective of ecological rationality, base-rate neglect may be a rational response to environments where individuating information is genuinely more diagnostic than population statistics. The question is not whether minds ignore base rates but whether the environment rewards or punishes doing so.

The numerical cognition literature suggests that base-rate neglect is amplified when base rates are presented in formats that do not match natural frequency representations. When statistics are presented as relative frequencies ("1 in 20") rather than as conditional probabilities ("5%"), base-rate neglect is reduced. This implies that the bias is not purely cognitive but is partially a framing effect — a consequence of how information is presented rather than of how minds process it.

Base-Rate Neglect as a Systems Phenomenon

The standard treatment of base-rate neglect locates the error in individual cognition: the human mind is a flawed probability calculator, systematically overweighting vivid particulars and underweighting dry statistics. This framing is not wrong, but it is incomplete. Base-rate neglect is also a systems phenomenon — a property of the coupling between cognitive architecture, information environment, and social structure.

Consider medical diagnosis. A physician who ignores the base rate of a disease and overweights a patient's symptoms is committing base-rate neglect. But the physician operates in a system that systematically distorts base rates: rare diseases are overrepresented in medical training because they are "interesting," the media amplifies disease outbreaks beyond their statistical significance, and malpractice law rewards false positives more than false negatives. The physician's neglect is not merely cognitive. It is ecologically rational — a response to a system that has made base rates unreliable guides to action.

The same pattern appears in financial markets. Investors who chase recent performance — ignoring the base rate of long-term market returns — are routinely diagnosed with recency bias. But the financial system is designed to make recent performance salient: quarterly earnings reports, daily price movements, real-time portfolio tracking. The base rate — the historical distribution of returns over decades — is structurally invisible. The investor does not ignore it because of a cognitive flaw. The investor ignores it because the system has made it cognitively unavailable.

This reframing has consequences for intervention. If base-rate neglect is purely cognitive, the solution is training: teach people Bayes' theorem, teach them to seek base rates, teach them to de-bias. If base-rate neglect is systemic, the solution is redesign: restructure information environments so that base rates are salient, restructure incentives so that base-rate adherence is rewarded, restructure institutions so that base-rate information is generated and distributed.

The Base Rate in Collective Intelligence

Base-rate neglect is not only an individual phenomenon. It operates at the level of groups, organizations, and societies. A scientific community that overweights novel, surprising findings and underweights the base rate of false positives in exploratory research is committing collective base-rate neglect. The replication crisis in psychology and medicine is, in part, a consequence of this collective neglect: a literature dominated by statistically significant but base-rate-ignorant findings produces a field in which most published research findings are false.

The wisdom of crowds literature offers a partial solution: aggregating independent judgments can recover base-rate information that individuals neglect. But aggregation works only when the errors are independent. When information environments are shared — when everyone reads the same news, uses the same search engine, consults the same algorithmic recommender — the errors become correlated, and the crowd becomes no wiser than its most influential member. The filter bubble is a base-rate neglect machine: it systematically suppresses the statistical background and amplifies the individuating particular.

Base Rates and the Structure of Uncertainty

The deepest systems-theoretic question raised by base-rate neglect is whether base rates exist in the relevant sense for the judgments people actually make. A base rate is a stable frequency in a well-defined reference class. But many of the judgments that matter — Will this startup succeed? Is this policy effective? Will this relationship last? — involve reference classes that are ill-defined, dynamically changing, or causally heterogeneous. The "base rate" of startup success depends on which sector, which year, which funding environment, which team composition. The "base rate" of policy effectiveness depends on which jurisdiction, which implementation, which political context.

In these cases, base-rate neglect may not be a bias at all. It may be a rational recognition that the apparent base rate is not the relevant one — that the reference class is too broad, too heterogeneous, or too unstable to provide actionable guidance. The expert who overrides the base rate with case-specific information may not be ignoring statistics. She may be using a finer-grained model that the experimenter's "base rate" does not capture.

This does not mean base-rate neglect is a myth. It means that the diagnosis of base-rate neglect requires a prior judgment about what the correct reference class is — a judgment that is itself contestable and often underdetermined by the evidence. The bias is not purely in the head. It is in the gap between the statistical information available and the causal structure of the problem at hand.

Base-rate neglect is the signature of a system that has optimized for the salient over the significant, the particular over the general, the vivid over the valid. The cognitive error is real. But the cognitive error is the output of a system whose information architecture makes it inevitable.