Network Architecture of Object Recognition: Investigating Integration Between Dorsal and Ventral Visual Pathways

Poster Presentation: Monday, May 19, 2025, 8:30 am – 12:30 pm, Pavilion
Session: Object Recognition: Neural mechanisms

Claire Simmons1 (), Vlad Ayzenberg2, Marlene Behrmann3; 1Carnegie Mellon University, 2University of Pennsylvania, 3University of Pittsburgh

While mounting evidence indicates that both the dorsal and ventral visual pathways participate in object perception, their relationship remains controversial. Previous work has identified specific instances of pathway interaction, but whether these reflect a single integrated network or distinct systems with context-dependent coupling remains unclear. We investigated this question using multiple network analyses of fMRI data acquired in response to viewing objects. Eighteen participants viewed 96 object images while undergoing 3T fMRI scanning (3 runs, TR=2s). Images were presented in blocks of 15 (800ms display) with interleaved fixations. Functional connectivity (FC), psychophysiological interactions (PPI), and Granger causality analysis (GCA) were computed between regions of interest in both pathways, defined using probabilistic parcels registered to native space. Results revealed both distinct processing and integrated function. FC analysis showed widespread connectivity between pathways, with stronger patterns in the left hemisphere. PPI analysis demonstrated task-specific connectivity changes during object processing, with lateral occipital (LO) regions showing stronger modulation than posterior intraparietal sulcus (pIPS). This directional influence contrasts with previous tool-specific findings. Direct pathway comparisons through subtraction analysis revealed complementary functional connectivity patterns but distinct task-modulated interactions. GCA uncovered asymmetric information flow, with stronger dorsal-to-ventral directed connectivity in response to objects processing (mean GCA value ~7.5 for left pIPS to LO) compared to scrambled displays (mean ~1.0). Our results suggest that object recognition relies on an integrated network with hierarchical organization, rather than purely independent pathways. While regions maintain distinct processing roles, they show coordinated activity through specific spatial and temporal patterns. This work further develops a network-level understanding of how dorsal and ventral pathways interact during object recognition, reconciling previous findings of both pathway independence and interaction within a broader framework of coordinated but specialized processing.

Acknowledgements: National Eye Institute, NIH (RO1 EY027018) to M.B and NIGMS T32GM142630 to CS. MB acknowledges support from P30 CORE award EY08098 from the National Eye Institute, NIH, and unrestricted supporting funds from The Research to Prevent Blindness Inc, NY, and the Eye & Ear Foundation of Pittsburgh.