Related Experiment Video
Updated: Jun 13, 2025
03:31
Author Spotlight: Enhancement of Salient Object Detection for Smart Grid Applications
Published on: December 15, 2023
485
Semantic influences on object detection: Drift diffusion modeling provides insights regarding mechanism.
Jingming Xue1, Robert C Wilson1, Mary A Peterson2,3
1School of Psychology, Georgia Institute of Technology, Atlanta, Georgia, United States of America.
Plos Computational Biology
|June 11, 2025
Summary
Semantic information significantly influences object detection. This study used drift diffusion modeling to show that valid labels enhance evidence accumulation, while invalid labels create uncertainty, impacting decision-making processes in object recognition.
Area of Science:
- Cognitive Psychology
- Neuroscience
- Computational Modeling
Background:
- Semantics, activated by words, plays a crucial role in object detection.
- Previous research indexed object detection using correct reports of object locations in bipartite displays.
Purpose of the Study:
- To elucidate the cognitive mechanisms underlying object detection influenced by semantic information.
- To investigate how valid and invalid labels affect evidence accumulation during object recognition tasks.
Main Methods:
- Utilized drift diffusion modeling (DDM) to analyze behavioral data from object detection experiments.
- Employed bipartite displays with familiar objects and manipulated labels (Valid and Invalid) prior to test displays.
Main Results:
- Valid labels increased the drift rate (signal-to-noise ratio) towards correct decisions.
- Invalid labels, particularly those from the same superordinate category, increased the decision threshold, indicating a need for more evidence.
- Invalid labels diminished the facilitative effect of valid labels on evidence accumulation.
Conclusions:
- Semantic networks are actively engaged during object detection.
- The findings support a model where semantic context influences the efficiency and accuracy of visual object recognition.
