Characterizing spatiotemporal population receptive fields in human visual cortex with fMRI.

Area of Science:

• Neuroscience
• Computational Neuroscience
• Visual Neuroscience

Background:

• Understanding the spatiotemporal dynamics of neural responses is crucial for deciphering brain function.
• Previous research using functional magnetic resonance imaging (fMRI) has primarily focused on spatial characteristics of population receptive fields (pRFs), with limited insight into their temporal properties due to the slow nature of fMRI signals.

Approach:

• Developed an image-computable framework integrating fMRI data and computational modeling to estimate spatiotemporal pRFs.
• Created simulation software to validate the recovery of millisecond-resolution spatiotemporal parameters from synthesized fMRI responses.
• Applied the framework and a novel stimulus paradigm to map spatiotemporal pRFs in individual voxels across the human visual cortex in 10 participants.

Key Points:

• A compressive spatiotemporal (CST) pRF model significantly outperforms conventional spatial pRF models in explaining fMRI responses across visual streams.
• Identified three organizational principles: progressive increases in spatial/temporal integration windows and nonlinearities from early to later visual areas.
• Observed diverging spatial and temporal integration window characteristics across visual streams in later areas.
• Demonstrated systematic increases in spatial and temporal integration windows with eccentricity in early visual areas (V1-V3).

Conclusions:

• The developed computational framework enables accurate estimation of millisecond-resolution spatiotemporal pRFs from fMRI data.
• The findings reveal fundamental organizational principles governing spatiotemporal receptive field properties across the human visual cortex.
• This approach opens new avenues for studying fine-grained neural dynamics in the brain using non-invasive fMRI techniques.