Target-present guessing as a function of target prevalence and accumulated information in visual search

Peltier, Chad; Becker, Mark W. · 2017 · Crossref

DOI: 10.3758/s13414-017-1297-6

archive: archived pipeline: cataloged verified

Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)

Summary

This study investigates how target prevalence and accumulated visual information influence guessing strategies in visual search tasks. The research addresses a discrepancy in existing models, specifically the Multiple Decision Model (MDM), which posits that observers always respond "target-absent" if they terminate a search without detecting a target. However, prior evidence suggests that observers frequently make "target-present" guesses, particularly in high-prevalence conditions, implying they use prevalence rates to inform educated guesses after search termination. The authors aimed to determine whether observers integrate both the base rate of target prevalence and the amount of information gathered during the search to make these decisions. The experiment involved 40 undergraduate participants performing a visual search task using eye-tracking technology. Stimuli consisted of arrays of 24 items (rotated Ls and Ts) presented on a grid. The study manipulated two key variables: target prevalence (10%, 50%, or 90%) and the amount of information accumulated before a response was required. To control for accumulated information, trials were terminated either by the participant’s response or automatically after five or ten fixations. This design allowed the researchers to isolate the effect of inspected items on guessing behavior, distinguishing between guesses based on prevalence and those resulting from misidentification errors. The results confirmed standard prevalence effects in self-terminating trials, where lower prevalence led to higher miss rates and faster reaction times for target-absent trials. Crucially, the analysis of false alarms revealed that guessing rates were significantly influenced by both prevalence and the number of fixations allowed. False alarm rates increased with higher prevalence but decreased as the number of inspected items increased. For instance, in the 90% prevalence condition, false alarms were highest when only five fixations were allowed and lowest when unlimited fixations were permitted. This interaction indicates that observers weight the prevalence rate by the amount of evidence they have already ruled out. A computational model of ideal guessing, based on the probability of a target being in uninspected items, closely matched observed behavior, though participants were slightly more conservative than the model predicted, likely due to confidence in early search guidance. The findings challenge the MDM’s assertion that all non-detection responses are target-absent judgments. Instead, the study demonstrates that observers employ a sophisticated guessing strategy that integrates prior knowledge of target prevalence with real-time accumulated evidence. As observers inspect more items, the influence of prevalence on their guessing diminishes, suggesting they update their probabilistic estimates based on the specific display contents. This implies that visual search decisions are not solely driven by perceptual thresholds or fixed criteria but involve dynamic, statistically driven inferences about the likelihood of a target’s presence in uninspected regions.

Provenance

The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed.

Summary generated by qwen3.6-27b-prismaquant on 2026-06-25; verification: verified.

Topics

Ranked by relevance to this paper. Hover a topic for its definition.

• visual search