Related Experiment Video
Updated: Aug 20, 2025
09:15
The Attentional Set Shifting Task: A Measure of Cognitive Flexibility in Mice
Published on: February 4, 2015
27.5K
Look twice: A generalist computational model predicts return fixations across tasks and species
Mengmi Zhang1,2,3, Marcelo Armendariz1,2,4, Will Xiao5
1Boston Children's Hospital, Harvard Medical School, Boston, Massachusetts, United States of America.
Plos Computational Biology
|November 22, 2022
Summary
Primates frequently revisit previously viewed locations using return fixations, a common behavior across species and tasks. A novel neural network model explains these return fixations, balancing exploration with detailed scrutiny.
Area of Science:
- Neuroscience
- Computational Vision
- Cognitive Science
Background:
- Primates utilize saccadic eye movements for visual exploration, but also frequently revisit locations.
- Return fixations are a ubiquitous behavior observed across different tasks, species (monkeys and humans), and viewing conditions (static images or natural behaviors).
Purpose of the Study:
- To systematically analyze the properties of return fixations in primate vision.
- To develop and validate a biologically-inspired computational model that explains the mechanisms underlying return fixations.
Main Methods:
- Analysis of a large dataset of 44,328 return fixations from 217,440 total fixations in monkeys and humans.
- Development of a neural network model integrating image features, saliency, task relevance, inhibition-of-return, and saccade constraints.
Main Results:
- Return fixations exhibit consistent locations across subjects, short temporal offsets, and a 180-degree turn in saccade direction.
- The computational model successfully replicated universal properties of return fixations without task-specific parameter tuning.
Conclusions:
- Return fixations are a fundamental aspect of primate visual behavior, crucial for balancing new information gathering with detailed examination of previously viewed areas.
- The proposed model offers a mechanistic explanation for return fixations, contributing to understanding visual attention and exploration strategies.
