Gaze–grasp coordination in obstacle avoidance: differences between binocular and monocular viewing

Most adults can skillfully avoid potential obstacles when acting in everyday cluttered scenes. We examined how gaze and hand movements are normally coordinated for obstacle avoidance and whether these are altered when binocular depth information is unavailable. Visual fixations and hand movement kinematics were simultaneously recorded, while 13 right-handed subjects reached-to-precision grasp a cylindrical household object presented alone or with a potential obstacle (wine glass) located to its left (thumb’s grasp side), right or just behind it (both closer to the finger’s grasp side) using binocular or monocular vision. Gaze and hand movement strategies differed significantly by view and obstacle location. With binocular vision, initial fixations were near the target’s centre of mass (COM) around the time of hand movement onset, but usually shifted to end just above the thumb’s grasp site at initial object contact, this mainly being made by the thumb, consistent with selecting this digit for guiding the grasp. This strategy was associated with faster binocular hand movements and improved end-point grip precision across all trials than with monocular viewing, during which subjects usually continued to fixate the target closer to its COM despite a similar prevalence of thumb-first contacts. While subjects looked directly at the obstacle at each location on a minority of trials and their overall fixations on the target were somewhat biased towards the grasp side nearest to it, these gaze behaviours were particularly marked on monocular vision-obstacle behind trials which also commonly ended in finger-first contact. Subjects avoided colliding with the wine glass under both views when on the right (finger side) of the workspace by producing slower and straighter reaches, with this and the behind obstacle location also resulting in ‘safer’ (i.e. narrower) peak grip apertures and longer deceleration times than when the goal object was alone or the obstacle was on its thumb side. But monocular reach paths were more variable and deceleration times were selectively prolonged on finger-side and behind obstacle trials, with this latter condition further resulting in selectively increased grip closure times and corrections. Binocular vision thus provided added advantages for collision avoidance, known to require intact dorsal cortical stream processing mechanisms, particularly when the target of the grasp and potential obstacle to it were fairly closely separated in d