Visual similarity is stronger than semantic similarity in guiding visual search for numbers

Godwin, Hayward J.; Hout, Michael C.; Menneer, Tamaryn · 2014 · Crossref

DOI: 10.3758/s13423-013-0547-4

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 relative influence of visual and semantic similarity on guiding visual search, specifically addressing how attention is directed toward distractors during the search for numerical targets. While classic models of visual search emphasize guidance by low-level visual features (e.g., color, shape), recent research suggests that high-level semantic information also plays a role. The authors aimed to determine whether search is guided more strongly by visual resemblance or by semantic relatedness (numerical distance) by employing a controlled stimulus set of single-digit numbers (0–9). This approach allowed for the independent manipulation and quantification of both visual and semantic properties, extending previous work that focused on limited stimuli. The experimental design involved two phases. First, 21 participants completed a multidimensional scaling (MDS) task to generate a psychologically tractable metric of visual similarity among the digits. Participants arranged the digits in a spatial layout where proximity indicated perceived visual similarity; the resulting two-dimensional MDS solution provided distance estimates between all digit pairs. Second, 30 participants performed an eye-tracking visual search task. They searched for a specific target digit among 11 distractors on a grid. Eye movements were recorded using an Eyelink 1000 system. The study analyzed the probability of fixating each distractor as a function of its visual similarity (derived from MDS distances) and semantic similarity (defined as the inverse of numerical distance from the target) to the target. A linear mixed-effects model was used to analyze the binary fixation data, accounting for random variations across participants, targets, and distractors. The results demonstrated that both visual and semantic similarity significantly influenced the probability of fixating distractors. Distractors that were visually similar to the target were more likely to be fixated, as were distractors that were numerically close to the target. However, the magnitude of these effects differed substantially. The slope for visual similarity in the model was approximately 0.48, whereas the slope for semantic similarity was only 0.05. This indicates that the effect of visual similarity on fixation probability was roughly nine times stronger than that of semantic similarity. Additionally, the influence of visual similarity was modulated by target presence, with a steeper slope observed in target-absent trials compared to target-present trials. There was no significant correlation between the MDS-derived visual similarity ratings and semantic similarity, confirming that participants could distinguish between the two dimensions. The findings conclude that while semantic information does guide visual search for numbers, visual similarity is the dominant factor in directing attention. This result clarifies the interplay between low-level perceptual features and high-level semantic knowledge in search guidance. The study also validates a novel methodological approach combining MDS with eye-tracking, providing a reliable metric for quantifying visual similarity that is agnostic to specific featural assumptions. This framework offers a stable tool for future research investigating search behavior in more complex displays, where multiple sources of information converge to guide attention.

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