# Prevalence effects in visual search: Theoretical and practical implications

> **NIH NIH R01** · BRIGHAM AND WOMEN'S HOSPITAL · 2024 · $480,432

## Abstract

Project Summary
Most of what we know about visual search comes from experiments where observers perform hundreds of
searches for the same things (e.g. several hundred searches for a vowel among other letters). However,
most searches in the world are not performed in blocks. People look for one thing and then another.
Emergency Department (ED) radiology where a radiologist might look for pneumonia in a chest X-ray,
followed by a search for a possible stroke in a head CT, and so on, represents a socially important real-world
example that we will study here. Decades of research have established rules of search, based on roughly
uniform blocks of searches. Do those rules continue to apply if searches are mixed? The importance of this
question can be seen if we consider the problem of when to end a search. Quit too soon and targets (like
that stroke) can be missed. Quit too late and time is wasted, perhaps dangerously (e.g. if a driver
perseverates on a search on the dashboard control panel). Models of search termination often propose an
adaptive process where responses on previous trials serve to adjust the quitting threshold on subsequent
trials. How can such an adaptive process work if the preceding trials come from many different tasks? We
hypothesize that multiple adaptive rules coexist, but that they may be less optimal in the mixed environment.
The answer to such questions will shape how we attempt to minimize error in settings like ED radiology.
There are four series of proposed experiments. In Exp 1, trials of several different tasks are run, either in
single-task blocks or mixed together. In Exp 2, the target remains the same on all trials but the distractors
(and, thus, the search task) change. This is intended to emulate situations like screening mammography
where the same target (e.g. a mass) is presented on visually different backgrounds (fatty vs dense breast
parenchyma). In Exp 3, multiple tasks are available simultaneously (as in driving or managing the control
panel of a complex machine). Here, the observer can choose, to some extent, whether to perform blocked or
mixed searches. Finally, in Exp 4, we compare mixed and blocked searches using radiologists as observers
in a version of the ED radiology task. Significantly, this will be the first medical image perception study of the
ED task. While we expect that many of the basic principles of search will be similar in mixed and blocked
conditions, we hypothesize that some rules, notably those pertaining to errors in search, will differ in mixed
and blocked situations. Understanding the rules of search when the specific search task changes from
moment to moment will be critical to reducing the errors that plague real world searches from the
inconsequential failure to find a pen to those with potential life and death consequences.

## Key facts

- **NIH application ID:** 10867909
- **Project number:** 2R01EY017001-16
- **Recipient organization:** BRIGHAM AND WOMEN'S HOSPITAL
- **Principal Investigator:** Jeremy M Wolfe
- **Activity code:** R01 (R01, R21, SBIR, etc.)
- **Funding institute:** NIH
- **Fiscal year:** 2024
- **Award amount:** $480,432
- **Award type:** 2
- **Project period:** 2007-04-01 → 2029-06-30

## Primary source

NIH RePORTER: https://reporter.nih.gov/project-details/10867909

## Citation

> US National Institutes of Health, RePORTER application 10867909, Prevalence effects in visual search: Theoretical and practical implications (2R01EY017001-16). Retrieved via AI Analytics 2026-05-24 from https://api.ai-analytics.org/grant/nih/10867909. Licensed CC0.

---

*[NIH grants dataset](/datasets/nih-grants) · CC0 1.0*