Gaze when reaching to grasp a glass

Age effects on predictive eye movements for action

When interacting with the environment, humans typically shift their gaze to where information is to be found that is useful for the upcoming action. With increasing age, people become slower both in processing sensory information and in performing their movements. One way to compensate for this slowing down could be to rely more on predictive strategies. To examine whether we could find evidence for this, we asked younger (19-29 years) and older (55-72 years) healthy adults to perform a reaching task wherein they hit a visual target that appeared at one of two possible locations. In separate blocks of trials, the target could appear always at the same location (predictable), mainly at one of the locations (biased), or at either location randomly (unpredictable). As one might expect, saccades toward predictable targets had shorter latencies than those toward less predictable targets, irrespective of age. Older adults took longer to initiate saccades toward the target location than younger adults, even when the likely target location could be deduced. Thus we found no evidence of them relying more on predictive gaze. Moreover, both younger and older participants performed more saccades when the target location was less predictable, but again no age-related differences were found. Thus we found no tendency for older adults to rely more on prediction.

Oculomotor behavior can be adjusted on the basis of artificial feedback signals indicating externally caused errors

Whether a saccade is accurate and has reached the target cannot be evaluated during its execution, but relies on post-saccadic feedback. If the eye has missed the target object, a secondary corrective saccade has to be made to align the fovea with the target. If a systematic post-saccadic error occurs, adaptive changes to the oculomotor behavior are made, such as shortening or lengthening the saccade amplitude. Systematic post-saccadic errors are typically attributed internally to erroneous motor commands. The corresponding adaptive changes to the motor command reduce the error and the need for secondary corrective saccades, and, in doing so, restore accuracy and efficiency. However, adaptive changes to the oculomotor behavior also occur if a change in saccade amplitude is beneficial for task performance, or if it is rewarded. Oculomotor learning thus is more complex than reducing a post-saccadic position error. In the current study, we used a novel oculomotor learning paradigm and investigated whether human participants are able to adapt their oculomotor behavior to improve task performance even when they attribute the error externally. The task was to indicate the intended target object among several objects to a simulated human-machine interface by making eye movements. The participants were informed that the system itself could make errors. The decoding process depended on a distorted landing point of the saccade, resulting in decoding errors. Two different types of visual feedback were added to the post-saccadic scene and we compared how participants used the different feedback types to adjust their oculomotor behavior to avoid errors. We found that task performance improved over time, regardless of the type of feedback. Thus, error feedback from the simulated human-machine interface was used for post-saccadic error evaluation. This indicates that 1) artificial visual feedback signals and 2) externally caused errors might drive adaptive changes to oculomotor behavior.

Continuous use of visual information about the position of the moving hand

People generally look at a target when they want to reach for it. Doing so presumably helps them continuously update their judgments about the target's position and motion. But not looking at their hand does not prevent people from updating judgments about its position on the basis of visual information, because people do respond to experimental perturbations of visual information about the position of their hand. Here, we study such responses by adding jitter to the movement of a cursor that follows participants' fingers. We analyse the response to the jitter in a way that reveals how the vigour of the response depends on the moment during the movement at which the change in cursor position occurs. We compare the change in vigour to that for equivalent jitter in the position of the target. We find that participants respond to jitter in the position of a cursor in much the same way as they respond to jitter in the target's position. The responses are more vigorous late in the movement, when adjustments need to be made within less time, but similarly so for the cursor as for the target. The responses are weaker for the cursor, presumably because of the jitter-free kinaesthetic information about the position of the finger.

Looking at the Ebbinghaus illusion: differences in neurocomputational requirements, not gaze-mediated attention, explain a classic perception-action dissociation

Perceiving and grasping an object present an animal with different sets of computational problems. The solution in primates entails the specialization of separate neural networks for visual processing with different object representations. This explains why the Ebbinghaus illusion minimally affects the grasping hand's in-flight aperture, which normally scales with target size, even though the size of the target disc remains misperceived. An attractive alternative account, however, posits that grasps are refractory to the illusion because participants fixate on the target and fail to attend to the surrounding context. To test this account, we tracked both limb and gaze while participants made forced-choice judgments of relative disc size in the Ebbinghaus illusion or did so in combination with grasping or manually estimating the size of one of the discs. We replicated the classic dissociation: grasp aperture was refractory to the measured illusory effect on perceived size, while judgments and manual estimates of disc size were not. Importantly, the number of display-wide saccades per second and the percentage of total fixation time or fixations directed at the selected disc failed to explain the dissociation. Our findings support the contention that object perception and goal-directed action rely on distinct visual representations. This article is part of a discussion meeting issue 'New approaches to 3D vision'.

Toward an Attentive Robotic Architecture: Learning-Based Mutual Gaze Estimation in Human–Robot Interaction

Social robotics is an emerging field that is expected to grow rapidly in the near future. In fact, it is increasingly more frequent to have robots that operate in close proximity with humans or even collaborate with them in joint tasks. In this context, the investigation of how to endow a humanoid robot with social behavioral skills typical of human–human interactions is still an open problem. Among the countless social cues needed to establish a natural social attunement, this article reports our research toward the implementation of a mechanism for estimating the gaze direction, focusing in particular on mutual gaze as a fundamental social cue in face-to-face interactions. We propose a learning-based framework to automatically detect eye contact events in online interactions with human partners. The proposed solution achieved high performance both in silico and in experimental scenarios. Our work is expected to be the first step toward an attentive architecture able to endorse scenarios in which the robots are perceived as social partners.

The response to background motion: Characteristics of a movement stabilization mechanism

When making goal-directed movements toward a target, our hand deviates from its path in the direction of sudden background motion. We propose that this manual following response arises because ongoing movements are constantly guided toward the planned movement endpoint. Such guidance is needed to compensate for modest, unexpected self-motion. Our proposal is that the compensation for such self-motion does not involve a sophisticated analysis of the global optic flow. Instead, we propose that any motion in the vicinity of the planned endpoint is attributed to the endpoint's egocentric position having shifted in the direction of the motion. The ongoing movement is then stabilized relative to the shifted endpoint. In six experiments, we investigate what aspects of motion determine this shift of planned endpoint. We asked participants to intercept a moving target when it reached a certain area. During the target's motion, background structures briefly moved either leftward or rightward. Participants’ hands responded to background motion even when each background structure was only briefly visible or when the vast majority of background structures remained static. The response was not restricted to motion along the target's path but was most sensitive to motion close to where the target was to be hit, both in the visual field and in depth. In this way, a movement stabilization mechanism provides a comprehensive explanation of many aspects of the manual following response.

Grasping performance depends upon the richness of hand feedback

Although visual feedback of the hand allows fast and accurate grasping actions, little is known about whether the nature of feedback of the hand affects performance. We investigated kinematics during precision grasping (with the index finger and thumb) when participants received different levels of hand feedback, with or without visual feedback of the target. Specifically, we compared performance when participants saw (1) no hand feedback; (2) only the two critical points on the index finger and thumb tips; (3) 21 points on all digit tips and hand joints; (4) 21 points connected by a "skeleton", or (5) full feedback of the hand wearing a glove. When less hand feedback was available, participants took longer to execute the movement because they allowed more time to slow the reach and close the hand. When target feedback was unavailable, participants took longer to plan the movement and reached with higher velocity. We were particularly interested in investigating maximum grip aperture (MGA), which can reflect the margin of error that participants allow to compensate for uncertainty. A trend suggested that MGA was smallest when ample feedback was available (skeleton and full hand feedback, regardless of target feedback) and when only essential information about hand and target was provided (2-point hand feedback + target feedback) but increased when non-essential points were included (21-point feedback). These results suggest that visual feedback of the hand affects grasping performance and that, while more feedback is usually beneficial, this is not necessarily always the case.

Looking Precisely at Your Fingertip Requires Visual Guidance of Gaze

People often look at objects that they are holding in their hands. It is therefore reasonable to expect them to be able to direct their gaze precisely with respect to their fingers. However, we know that people make reproducible idiosyncratic errors of up to a few centimetres when they try to align a visible cursor to their own finger hidden below a surface. To find out whether they also make idiosyncratic errors when they try to look at their finger, we asked participants to hold their finger in front of their head in the dark, and look at it. Participants made idiosyncratic errors of a similar magnitude to those previously found when matching a visual cursor to their hidden finger. This shows that proprioceptive position sense of finger and gaze are not aligned, suggesting that people rely on vision to guide their gaze to their own finger.

Somatosensory target information is used for reaching but not for saccadic eye movements

A systematic investigation of contributions of different somatosensory modalities (proprioception, kinesthesia, tactile) for goal-directed movements is missing. Here we demonstrate that while eye movements are not affected by different types of somatosensory information, reach precision improves when two different types of information are available. Moreover, reach accuracy and gaze precision to unseen somatosensory targets improve when performing coordinated eye-hand movements, suggesting bidirectional contributions of efferent information in reach and eye movement control.

Depth cue reweighting requires altered correlations with haptic feedback

Depth cue reweighting is a feedback-driven learning process that modifies the relative influences of different sources of three-dimensional shape information in perceptual judgments and or motor planning. In this study, we investigated the mechanism supporting reweighting of stereo and texture information by manipulating the haptic feedback obtained during a series of grasping movements. At the end of each grasp, the fingers closed down on a physical object that was consistent with one of the two cues, depending on the condition. Previous studies have shown that this style of visuomotor training leads to cue reweighting for perceptual judgments, but the time course has never been documented for a single training session, and many questions remain regarding the underlying mechanism, such as the pattern of feedback signals required to drive reweighting. We address these issues in two experiments, finding short-term changes in the motor response consistent with cue reweighting: the slope of the grip aperture with respect to the reliable cue increased, whereas the slope with respect to the unreliable cue decreased. Critically, Experiment 2 shows that slope changes do not occur when one of the cues is rendered with a constant bias; the grip aperture simply becomes uniformly larger or smaller. Our findings support a model of cue reweighting driven by altered correlations between haptic feedback and individual cues, rather than simple mismatches, which can be resolved by other mechanisms such as sensorimotor adaptation or cue recalibration.

On the Visuomotor Behavior of Amputees and Able-Bodied People During Grasping

Visual attention is often predictive for future actions in humans. In manipulation tasks, the eyes tend to fixate an object of interest even before the reach-to-grasp is initiated. Some recent studies have proposed to exploit this anticipatory gaze behavior to improve the control of dexterous upper limb prostheses. This requires a detailed understanding of visuomotor coordination to determine in which temporal window gaze may provide helpful information. In this paper, we verify and quantify the gaze and motor behavior of 14 transradial amputees who were asked to grasp and manipulate common household objects with their missing limb. For comparison, we also include data from 30 able-bodied subjects who executed the same protocol with their right arm. The dataset contains gaze, first person video, angular velocities of the head, and electromyography and accelerometry of the forearm. To analyze the large amount of video, we developed a procedure based on recent deep learning methods to automatically detect and segment all objects of interest. This allowed us to accurately determine the pixel distances between the gaze point, the target object, and the limb in each individual frame. Our analysis shows a clear coordination between the eyes and the limb in the reach-to-grasp phase, confirming that both intact and amputated subjects precede the grasp with their eyes by more than 500 ms. Furthermore, we note that the gaze behavior of amputees was remarkably similar to that of the able-bodied control group, despite their inability to physically manipulate the objects.

Some binocular advantages for planning reach, but not grasp, components of prehension

Proficient (fast, accurate, precise) hand actions for reaching-to-grasp 3D objects are known to benefit significantly from the use of binocular vision compared to one eye alone. We examined whether these binocular advantages derive from increased reliability in encoding the goal object’s properties for feedforward planning of prehension movements or from enhanced feedback mediating their online control. Adult participants reached for, precision grasped and lifted cylindrical table-top objects (two sizes, 2 distances) using binocular vision or only their dominant/sighting eye or their non-dominant eye to program and fully execute their movements or using each of the three viewing conditions only to plan their reach-to-grasp during a 1 s preview, with vision occluded just before movement onset. Various kinematic measures of reaching and grasping proficiency, including corrective error rates, were quantified and compared by view, feedback and object type. Some significant benefits of binocular over monocular vision when they were just available for pre-movement planning were retained for the reach regardless of target distance, including higher peak velocities, straighter paths and shorter low velocity approach times, although these latter were contaminated by more velocity corrections and by poorer coordination with object contact. By contrast, virtually all binocular advantages for grasping, including improvements in peak grip aperture scaling, the accuracy and precision of digit placements at object contact and shorter grip application times preceding the lift, were eliminated with no feedback available, outcomes that were influenced by the object’s size. We argue that vergence cues can improve the reliability of binocular internal representations of object distance for the feedforward programming of hand transport, whereas the major benefits of binocular vision for enhancing grasping performance derive exclusively from its continuous presence online.