Visual targets aren't irreversibly converted to motor coordinates: eye-centered updating of visuospatial memory in online reach control.

Counter to current and widely accepted hypotheses that sensorimotor transformations involve converting target locations in spatial memory from an eye-fixed reference frame into a more stable motor-based reference frame, we show that this is not strictly the case. Eye-centered representations continue to dominate reach control even during movement execution; the eye-centered target representation persists after conversion to a motor-based frame and is continuously updated as the eyes move during reach, and is used to modify the reach plan accordingly during online control. While reaches are known to be adjusted online when targets physically shift, our results are the first to show that similar adjustments occur in response to changes in representations of remembered target locations. Specifically, we find that shifts in gaze direction, which produce predictable changes in the internal (specifically eye-centered) representation of remembered target locations also produce mid-transport changes in reach kinematics. This indicates that representations of remembered reach targets (and visuospatial memory in general) continue to be updated relative to gaze even after reach onset. Thus, online motor control is influenced dynamically by both the external and internal updating mechanisms.

Allocentrically implied target locations are updated in an eye-centred reference frame.

When reaching to remembered target locations following an intervening eye movement a systematic pattern of error is found indicating eye-centred updating of visuospatial memory. Here we investigated if implicit targets, defined only by allocentric visual cues, are also updated in an eye-centred reference frame as explicit targets are. Participants viewed vertical bars separated by varying distances, and horizontal lines of equivalently varying lengths, implying a "target" location at the midpoint of the stimulus. After determining the implied "target" location from only the allocentric stimuli provided, participants saccaded to an eccentric location, and reached to the remembered "target" location. Irrespective of the type of stimulus reaching errors to these implicit targets are gaze-dependent, and do not differ from those found when reaching to remembered explicit targets. Implicit target locations are coded and updated as a function of relative gaze direction with respect to those implied locations just as explicit targets are, even though no target is specifically represented.

The coding and updating of visuospatial memory for goal-directed reaching and pointing.

In this review we discuss evidence from psychophysical, electrophysiological, and neuroimaging studies that demonstrates the updating of remembered visual space in a reference frame that is centred on the eye. We then extend these findings by discussing recent work from our lab. Specifically, we address eye-centred updating of visuospatial memory for arm movements following different types of eye movements, the role of retinal versus extraretinal information in such spatial updating, and the use of allocentric versus egocentric information in coding multiple targets. We provide a conceptual model to explain the relationships among these findings.

Locations of serial reach targets are coded in multiple reference frames.

Previous work from our lab, and elsewhere, has demonstrated that remembered target locations are stored and updated in an eye-fixed reference frame. That is, reach errors systematically vary as a function of gaze direction relative to a remembered target location, not only when the target is viewed in the periphery (Bock, 1986, known as the retinal magnification effect), but also when the target has been foveated, and the eyes subsequently move after the target has disappeared but prior to reaching (e.g., Henriques, Klier, Smith, Lowy, & Crawford, 1998; Sorrento & Henriques, 2008; Thompson & Henriques, 2008). These gaze-dependent errors, following intervening eye movements, cannot be explained by representations whose frame is fixed to the head, body or even the world. However, it is unknown whether targets presented sequentially would all be coded relative to gaze (i.e., egocentrically/absolutely), or if they would be coded relative to the previous target (i.e., allocentrically/relatively). It might be expected that the reaching movements to two targets separated by 5° would differ by that distance. But, if gaze were to shift between the first and second reaches, would the movement amplitude between the targets differ? If the target locations are coded allocentrically (i.e., the location of the second target coded relative to the first) then the movement amplitude should be about 5°. But, if the second target is coded egocentrically (i.e., relative to current gaze direction), then the reaches to this target and the distances between the subsequent movements should vary systematically with gaze as described above. We found that requiring an intervening saccade to the opposite side of 2 briefly presented targets between reaches to them resulted in a pattern of reaching error that systematically varied as a function of the distance between current gaze and target, and led to a systematic change in the distance between the sequential reach endpoints as predicted by an egocentric frame anchored to the eye. However, the amount of change in this distance was smaller than predicted by a pure eye-fixed representation, suggesting that relative positions of the targets or allocentric coding was also used in sequential reach planning. The spatial coding and updating of sequential reach target locations seems to rely on a combined weighting of multiple reference frames, with one of them centered on the eye.

Updating visual memory across eye movements for ocular and arm motor control.

Remembered object locations are stored in an eye-fixed reference frame, so that every time the eyes move, spatial representations must be updated for the arm-motor system to reflect the target's new relative position. To date, studies have not investigated how the brain updates these spatial representations during other types of eye movements, such as smooth-pursuit. Further, it is unclear what information is used in spatial updating. To address these questions we investigated whether remembered locations of pointing targets are updated following smooth-pursuit eye movements, as they are following saccades, and also investigated the role of visual information in estimating eye-movement amplitude for updating spatial memory. Misestimates of eye-movement amplitude were induced when participants visually tracked stimuli presented with a background that moved in either the same or opposite direction of the eye before pointing or looking back to the remembered target location. We found that gaze-dependent pointing errors were similar following saccades and smooth-pursuit and that incongruent background motion did result in a misestimate of eye-movement amplitude. However, the background motion had no effect on spatial updating for pointing, but did when subjects made a return saccade, suggesting that the oculomotor and arm-motor systems may rely on different sources of information for spatial updating.

The hand knows something that the eye does not: reaching movements resist the Müller-Lyer illusion whether or not the target is foveated.

Previous reports suggest that saccades are affected by the Müller-Lyer (ML) pictorial illusion, whereas reaching movements are not. It is unclear if the resistance of reaching to illusions depends on the concurrent engagement of the oculomotor system. Here we show that the endpoints and kinematics of reaching movements were unaffected by a peripherally viewed ML stimulus regardless of whether or not a concurrent saccade was carried out. Primary saccade endpoints were affected by the ML stimulus but secondary saccades were not. Perceptual judgments of target location were influenced by the ML stimulus in the expected direction. The resistance of reaching movements to pictorial illusions does not appear to depend on the concurrent engagement of the oculomotor system. Implications for models of oculomotor and upper limb control are discussed.