CORTICAL MECHANISMS UNDERLYING HUMAN EYE-HAND COORDINATION: BEYOND SUMMATION OF SACCADE AND REACH NETWORKS

Eye-hand coordination underpins everyday behaviour, yet neuroimaging studies have yet to clearly separate eye and hand control regions or determine how they interact during eye-hand coordination. Most neuroimaging work has examined eye and hand movements in isolation and faces methodological challenges in dissociating sensory encoding from motor planning and execution. The aim of this study was to determine whether eye-hand coordination is simply the additive combination of saccade and reach-only neural processes, or whether it recruits distinct cortical mechanisms.
Twenty adults performed a cue-separation fMRI task, consisting of three distinct phases, isolating neural responses to visual input, movement planning, and execution of saccades, reaches, and eye-hand coordination (EHC). Behaviour was measured using MRI-compatible eye-tracker and LED-touchscreen. Analyses included, 1) nonparametric permutation testing (p < 0.05, FWE-corrected) to identify task-related activation, and 2) conjunction and region-of-interest (ROI) analyses to directly compare EHC activity to saccades and reaches.
Planning and execution revealed dissociable, effector-specific patterns.
Planning: Reaches and EHC engaged extensive occipito-parieto-frontal areas compared to saccades.
Execution: Saccades produced extensive medial / lateral frontal and occipito-parietal activation, while reaches / EHC shifted from frontal to more extensive primary sensorimotor areas, with hemispheric lateralization. Conjunction and ROI analyses demonstrated that planning and executing EHC required greater activity than saccades and reaches alone in visual, superior parietal, and premotor dorsal areas.
These findings demonstrate distinct temporal and spatial organization of eye and hand planning and execution signals, additional neural processes dedicated to eye-hand coordination, and highlight potential candidate mechanisms supporting coordinated motor control.