Maria Guadalupe Yanez-Ramos¹ (), Gabriela Ojeda Valencia¹, Harvey Huang², Nick Gregg³, Zeeshan Qadir¹, Morgan Montoya¹, Kendrick Kay⁵, Greg Worrell³, Kai Miller⁴, Dora Hermes¹; ¹Department of Physiology and Biomedical Engineering, Mayo Clinic, Rochester, MN, ²Mayo Clinic Medical Scientist Training Program, Rochester, MN, ³Department of Neurology, Mayo Clinic, Rochester, MN, ⁴Department of Neurological Surgery, Mayo Clinic, Rochester, MN, ⁵Center for Magnetic Resonance Research, Department of Radiology, University of Minnesota, Minneapolis, MN

Neural processing in areas along the visual pathways is influenced by inputs from connected visual areas. Although the connectivity between these areas has been studied using post-mortem dissection and diffusion MRI, little is known about the directionality of these connections in the living human brain. To understand the strength and directionality of human brain connectivity, capturing the inputs to one region when stimulating another is necessary. Cortico-cortical evoked potentials (CCEPs) recorded via stereo-electroencephalography electrodes are particularly suited for this purpose. A previous study using cortico-cortical spectral responses showed that visual feedforward connectivity was larger compared to feedback connectivity. Here, we leverage CCEPs to quantify the strength of the connectivity between visual pathways. We recorded CCEPs in 11 patients and used ‘Canonical Response Parametrization’ to capture CCEPs independent of their waveform. Connectivity strength was quantified in each direction by average response reliability, measured as the coefficient of determination (CoD), and connectivity sparsity as the percentage of significant connections from all possible connections. First, feedforward connections from V1-V3 to higher-order visual areas were more reliable and less sparse compared to feedback connections, similar to previous studies. Across 702 possible feedforward connections, the average CoD was 0.164 (SEM = 0.010), with 19-49% significant connections, whereas across 708 possible feedback connections, the CoD was lower (0.081 ± 0.007), with only 5-17% significant connections. Second, between pathways, ventral to dorsal connectivity was more reliable and less sparse compared to dorsal to ventral connectivity (98 possible connections each way, reliability: CoD = 0.102 ± 0.020 vs. 0.0543 ± 0.012, sparsity: 23% vs. 2%). These findings show that connectivity is sparse and not always reciprocal. Moreover, they suggest that visual feedforward influences are stronger compared to feedback influences. Finally, the ventral pathway may more strongly modulate the dorsal pathway compared to the reverse.

Acknowledgements: This work was supported by NIH R01MH122258 and R01EY035533.