An inter-item similarity model unifying feature and conjunction search

Area of Science:

• Cognitive Psychology
• Computational Neuroscience
• Visual Perception

Background:

• Visual search models often struggle to integrate attentional deployment with item similarity effects.
• Existing theories may not fully explain both efficient and inefficient search patterns.

Purpose of the Study:

• To propose and validate a unified model of visual search.
• To reconcile serial attention deployment with inter-item similarity effects.
• To account for variance in search performance across different set sizes and similarity conditions.

Main Methods:

• Systematic variation of target-distractor and distractor-distractor similarity across 85 set type-size conditions.
• Inclusion of univariate feature and bivariate conjunction search tasks.
• Development of a power function model incorporating similarity and set size.

Main Results:

• The proposed model, a power (square root) function, explained 98% of the variance in type-size means.
• The model effectively integrated target-distractor and distractor-distractor similarity with set size.

Conclusions:

• A unified theory of visual search is proposed, integrating item similarity as a key factor.
• Both efficient and inefficient visual search can be explained by a single theoretical framework.
• The model provides a robust account for performance variations in visual search tasks.