Prevalence Effect¶
In repeated visual search, rarer targets are more likely to be missed when they are present, often as observers adjust target-declaration and search-stopping criteria.
Core Idea¶
The low-prevalence effect is the finding that people searching many cases are more likely to miss a target when it is present if targets rarely appear across the run. The measure is misses divided by actual target-present cases, not the total number of misses. In a controlled simulated baggage search, misses rose from 7% at 50% target prevalence to 30% at 1%. Later research found that target-declaration criterion and the point at which observers end an apparently empty search can change with prevalence. Those processes help explain the effect but must be measured rather than assumed in every setting.[ref-23a6d8f38542][ref-3d33910e8f57]
Scope of Application¶
The literal domain is repeated human visual search with defined target-present and target-absent trials. Laboratory baggage simulations directly manipulate prevalence, while expert mammography research has compared known cases read in routine low-prevalence workflow with the same set read later in a high-prevalence laboratory session. The mammography study found 30% versus 12% false negatives, but setting, timing and observer assignment also differed; prevalence was not perfectly isolated as the only cause.[ref-23a6d8f38542][ref-6b5ed92c5ea3]
Clarity¶
A base rate says how often a target exists. The prevalence effect says how searchers' conditional miss behavior changes as that rate changes. It is not base-rate neglect or positive-predictive-value arithmetic. Nor does a higher miss rate automatically mean poorer visual sensitivity: in a 2010 study, response criterion and target-absent time tracked prevalence while measured sensitivity did not systematically do so.[^ref-3d33910e8f57]
Manages Complexity¶
The effect directs measurement toward five roles: repeated observer, verified target, target frequency, possible response/stopping adjustments, and misses among present targets. This separates prevalence from target difficulty and prevents a few raw misses in a rare-target workload from being misread as high accuracy. It also keeps laboratory control distinct from clinical realism; the two studies answer related but not identical causal questions.[ref-23a6d8f38542][ref-6b5ed92c5ea3]
Abstract Reasoning¶
To test a claim, compare the share of target-present cases missed under lower and higher prevalence, while checking whether stimulus difficulty, observers and context also changed. A controlled prevalence manipulation with rising conditional misses supports the effect. Reaction times and signal-detection measures can then distinguish earlier quitting or a more conservative target criterion from changed sensitivity. In applied evidence, report design limitations before treating a difference as prevalence-caused. Neither a low base rate alone nor a single missed target proves the effect.[ref-23a6d8f38542][ref-3d33910e8f57][^ref-6b5ed92c5ea3]
Knowledge Transfer¶
The original simulated baggage task and expert mammography study both have successive cases, known targets, different prevalence contexts and conditional miss comparisons. In the baggage simulation, prevalence was experimentally varied and misses increased as targets became rare. In mammography, a similar pattern appeared in professional reading but with clinical-versus-laboratory differences that limit causal attribution. The broader live Signal Detection Theory helps analyze criterion versus sensitivity; Base Rate names target frequency; neither is identical to this behavioral effect.
[^ref-23a6d8f38542]: Jeremy M. Wolfe, Todd S. Horowitz and Naomi M. Kenner, “Rare targets are often missed in visual search,” Nature 435 (2005), 439–440, original author manuscript, Figure 2 and discussion. DOI 10.1038/435439a. [^ref-3d33910e8f57]: Jeremy M. Wolfe and Michael J. Van Wert, “Varying Target Prevalence Reveals Two Dissociable Decision Criteria in Visual Search,” Current Biology 20 (2010), 121–124, original author-hosted paper, Summary and Experiments 1–2. DOI 10.1016/j.cub.2009.11.066. [^ref-6b5ed92c5ea3]: Karla K. Evans, Robyn L. Birdwell and Jeremy M. Wolfe, “If You Don’t Find It Often, You Often Don’t Find It: Why Some Cancers Are Missed in Breast Cancer Screening,” PLOS ONE 8 (2013), e64366, original open-access paper, Results and Discussion. DOI 10.1371/journal.pone.0064366.
Neighborhood in Abstraction Space¶
Prevalence Effect sits in a sparse region of the domain-specific corpus (68th percentile for distinctiveness): few abstractions share its structure, so a faithful description tends to retrieve it precisely.
Family — Unclustered & Miscellaneous (2551 abstractions)
Nearest neighbors
- Diagnostic Method — 0.85
- Spectrum Bias — 0.84
- N-Back Task — 0.84
- Latent Learning — 0.83
- Face validity — 0.83
Computed from structural-signature embeddings · 2026-10-08