Direct visuomotor mapping for fast visually-evoked arm movements

Abstract

In contrast to conventional reaction time (RT) tasks, saccadic RT’s to visual targets are very fast and unaffected by the number of possible targets. This can be explained by the sub-cortical circuitry underlying eye movements, which involves direct mapping between retinal input and motor output in the superior colliculus. Here we asked if the choice-invariance established for the eyes also applies to a special class of fast visuomotor responses of the upper limb. Using a target-pointing paradigm we observed very fast reaction times (<150 ms) which were completely unaffected as the number of possible target choices was increased from 1 to 4. When we introduced a condition of altered stimulus–response mapping, RT went up and a cost of choice was observed. These results can be explained by direct mapping between visual input and motor output, compatible with a sub-cortical pathway for visual control of the upper limb.

Introduction

Many situations require fast visually-guided limb movements, for example when catching a ball or stepping over an obstacle. Human behavioral studies demonstrate that during such tasks, hand or foot trajectories can be adjusted within a remarkably short latency of a visual stimulus (<150 ms; Carlton, 1981, Soechting and Lacquaniti, 1983, Day and Lyon, 2000, Reynolds and Day, 2005, Reynolds and Day, 2007, Pruszynski et al., 2010, Brenner and Smeets, 1997, Prablanc and Martin, 1992). The speed of the visuomotor pathway suggests that it may bypass the cerebral cortex and evidence from a split-brain patient supports this idea (Day & Brown, 2001). However, the notion of a sub-cortical pathway for visually-guided reaching is contentious, and there is some evidence to the contrary (Desmurget et al., 1999, Grea et al., 2002, Pisella et al., 2000).

By contrast, the existence of a sub-cortical pathway for oculomotor control is well established both at a behavioral and anatomical level. Saccadic reaction times (RT) to visual stimuli are typically within 250 ms (Fischer and Ramsperger, 1984, Saslow, 1967). Furthermore, the cost of choice which affects most RT tasks (formalized as Hick’s law; Hick, 1952) does not apply; saccadic RT remains unaffected even as the number of possible target locations increases from 1 to 8 (Kveraga and Hughes, 2005, Kveraga et al., 2002). This violation of Hick’s law can be explained when we consider the basic nature of the underlying circuitry. The brainstem circuitry for visually-guided saccades involves a direct pathway from the retina to the oculomotor nuclei via the superior colliculus (SC). Visual afferents terminate in a retinotopic fashion within the SC, lying in spatial register with premotor neurons which form a motor map encoding saccade direction and amplitude (Wurtz and Albano, 1980, Gandhi and Katnani, 2011). Thus the anatomical arrangement involves direct mapping between sensory input and motor output. This explains why no additional processing is required to map required eye movements onto additional targets, thus accounting for the violation of Hick’s law. If the relationship between target and response is rendered arbitrary (e.g., during ‘anti-saccades’) then the direct anatomical mapping is no longer available and a cost of choice becomes apparent (Kveraga et al., 2002).

If visual guidance of the upper limb is sub-served by a similar pathway then it should exhibit the same behavioral properties as saccades. Specifically we should expect to see fast RT’s and no effect of choice, but the evidence thus far is equivocal. Studies which have varied target choice from 2 upwards have demonstrated minimal or no increase in RT with choice but, unlike saccades, the transition from 1 to 2 choices imposes a clear increase in RT (Wright et al., 2007, Dassonville et al., 1999, Berryhill et al., 2005, Kveraga et al., 2006). However, these studies have employed indirect reaching paradigms in which a joystick or manipulandum is used. We therefore set out to determine if visually-guided pointing is genuinely impervious to choice using a target-jump paradigm where the task involves direct pointing reactions with the index finger. For comparison, we also included a condition of altered stimulus–response compatibility involving a 90° rotation. The results clearly show that direct, but not orthogonal pointing, displays exactly the same properties as saccadic eye movement, i.e., fast reaction times completely unaffected by choice, even between 1 and 2 choices. Furthermore, analysis of pointing errors during orthogonal pointing shows that this fast visuomotor process competes for control of the arm with a slower, more flexible system. The results suggest two distinct control pathways for visual guidance of the arm; a sub-cortical pathway which is fast but inflexible, and a cortical pathway which is slower but capable of dealing with arbitrary stimulus–response mappings. As the number of choices increases, interference between these two systems becomes apparent.