Saccade Velocity Driven Oscillatory Network Model of Grid Cells

Area of Science:

• Computational Neuroscience
• Cognitive Neuroscience
• Neuroscience

Background:

• Grid cells and place cells in the hippocampus are crucial for spatial navigation during physical movement.
• Recent studies suggest grid-like representations may also occur during saccadic eye movements while viewing images.

Purpose of the Study:

• To present computational models explaining the formation of grid patterns on saccadic trajectories.
• To investigate the network-level mechanisms underlying grid cell formation using neurally plausible algorithms.
• To predict the emergence of place cells in saccadic space.

Main Methods:

• Developed two computational models: Saccade Velocity Driven Oscillatory Network-Direct PCA (SVDON-DPCA) and Saccade Velocity Driven Oscillatory Network-Network PCA (SVDON-NPCA).
• Utilized an attention model to generate saccade trajectories based on image saliency maps.
• Extended SVDON-NPCA with a Layered Attention Hebbian Network (LAHN) to predict place cells.

Main Results:

• Both models successfully generated grid patterns on saccadic trajectories, consistent with experimental findings.
• The models captured spatial characteristics of grid cells, including scale variation along the medial entorhinal cortex axis.
• The extended model predicted place cell activity in saccadic space.

Conclusions:

• This study provides the first computational models for grid and place cell formation from saccade trajectories.
• The findings suggest that similar neural mechanisms may underlie spatial representations during both physical navigation and visual scanning.
• The models offer a framework for future experimental investigation into the neural basis of spatial cognition during visual attention.