Visual search for arbitrary objects in real scenes

Wolfe, Jeremy M.; Alvarez, George A.; Rosenholtz, Ruth; Kuzmova, Y.; Sherman, A. M. · 2011 · OpenAlex-citations

DOI: 10.3758/s13414-011-0153-3

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 the efficiency of visual search for arbitrary objects within complex, real-world scenes, addressing the gap between laboratory studies using isolated stimuli and naturalistic perception. The authors aim to determine if search in scenes is genuinely efficient and to identify the mechanisms—specifically semantic and episodic guidance—that facilitate this efficiency. A central challenge addressed is the "problem of set size," as defining the number of items in a cluttered scene is ambiguous. To resolve this, the researchers used hand-labeled regions in 100 indoor photographic scenes as a surrogate for set size. The research comprised six experiments. In Experiment 1, observers searched for named objects (e.g., chairs, bowls) in novel scenes. Reaction times (RTs) were analyzed against the number of labeled regions. Results showed highly efficient search, with slopes of approximately 5 ms/item for target-present trials, comparable to simple feature searches. However, high error rates suggested observers used guessing strategies based on object typicality. Experiment 2 controlled for these guesses, resulting in slightly steeper slopes (~15 ms/item), while Experiment 3 demonstrated that search for arbitrary objects outside of a scene context was significantly less efficient (~40 ms/item). Experiments 4–6 examined repeated search through the same scenes to assess the role of episodic guidance (familiarity with specific scenes). Key findings indicate that search efficiency in scenes is driven by scene-specific guidance rather than simple feature pop-out. Semantic guidance, derived from knowledge of scene structure (e.g., knowing where a stove is likely located), reduces the "functional set size" by eliminating irrelevant regions. Episodic guidance further refines this process; while general familiarity with a scene had modest effects on RT, searching for a specific target item a second time in the same scene produced massive speeding of RTs. This suggests that memory of a specific item’s location persists and aids subsequent searches. Additionally, target size and eccentricity were significant predictors of RT, with larger and more central targets found faster. The significance of this work lies in its demonstration that visual search in natural environments is efficient due to top-down guidance mechanisms that constrain attention to plausible locations. The authors propose that scene structure and prior knowledge allow the visual system to ignore most of the visual field, focusing only on a small functional set of potential targets. This challenges traditional models based on random displays and highlights the importance of semantic and episodic information in understanding human attention and scene perception.

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
• inattentional change blindness