Elsevier

Neuropsychologia

Volume 38, Issue 11, 1 October 2000, Pages 1473-1481
Neuropsychologia

Binocular cues are important in controlling the grasp but not the reach in natural prehension movements

https://doi.org/10.1016/S0028-3932(00)00065-8Get rights and content

Abstract

Binocular cues are typically considered to be pre-eminent in the control of reaching and grasping behaviour. However, in the absence of such information prehension movements can still be accurate and reliable. The present study therefore was designed to assess further the contribution of binocular information in the control of human reaching and grasping movements. Participants reached for and picked up objects under binocular and monocular viewing, both in the absence of a visible scene around the target objects (complete darkness with ‘self-illuminated’ objects and hand), and under normal (fully illuminated) viewing. Analysis of kinematic parameters indicated that the removal of binocular information did not significantly affect the major indices of the transport component, although it did affect the grasp component. In contrast, the kinematic parameters in the unlit conditions revealed that both the transport component and the grasp component of the reach were severely disrupted whether binocular cues were available or not. Our results suggest that binocular information may be more important for the control of grasp formation than for the control of the transport component. Elimination of the surrounding scene and normal visual feedback affects both the transport and the grasp. It is concluded that in normal viewing conditions, reaching and grasping movements are less dependent on binocular information than has previously been thought.

Introduction

In reaching to grasp an object in space, visual information is required to specify both the extrinsic and intrinsic properties of the object [4], [12]. The term extrinsic information refers to properties such as distance and orientation while the term intrinsic information refers to properties such as an object’s size, shape and possibly weight. The division between extrinsic/intrinsic information is useful as it maps on to the division between the relatively independent, although temporally coupled, components of a prehensile movement: the transport component and the grasp component [11], [12], [19]. Obviously to transport the hand, in the correct orientation, in a particular direction and for a particular distance, extrinsic object properties are required, whereas intrinsic properties are required to control the grip aperture and to select the most appropriate grasp points.

In normal, everyday visual scenes, both the extrinsic and intrinsic aspects of objects, and the layout of their supporting surfaces, are specified by a multitude of different visual cues. Binocular cues however, are typically considered to be paramount in the control of reaching and grasping behaviour [13], [17], [23]. One reason for this is that angle of convergence and binocular disparity once suitably scaled, can be used to specify the full metric properties of the visual scene,including absolute distance, which is necessary in the planning and control of many typical reaches [3], [6], [18], [23]. There is also strong evidence from physiological, neurological and behavioural studies to suggest that binocular vision is important for optimal prehensile movements. Sakata and colleagues, for example, have shown that many disparity sensitive cells in the posterior parietal cortex of primates, which are involved in control of manipulation-related activity, are also selective for 3-D surface orientation and for object 3-D axis-orientation [20], [21], [25]. In humans, Dijkerman, Milner and Carey [2] reported patient DF, who despite being a profound visual agnosic, could perform equally efficient prehension movements when compared to ‘normal’ controls. When only monocular information was available however, her performance deteriorated markedly relative to the controls. There is also evidence to suggest binocular information is important in the control of prehensile movements in normal participants. In their seminal paper Servos, Goodale and Jakobson [23] compared kinematic indices of movements made to grasp objects placed at different distances with either binocular or monocular (one eye occluded) viewing. They found that ‘monocular reaches’ showed lower peak wrist velocities, longer overall movement times and a longer deceleration phase than comparable ‘binocular reaches’ (see also [9]). They also found that the maximum grip aperture was consistently smaller with monocular viewing (but see [9]). The clear effects of the elimination of binocular viewing in these experiments augments the view that binocular cues are pre-eminent in the control of efficient reaching movements. In a slightly different approach, Mon-Williams and Dijkerman [15] selectively perturbed the angle of convergence using prisms, which did not alter the disparity information, and found that it influenced the transport but not the grasp component of the reach. In most of these experiments however, only a small number of object/distance combinations were used and these were presented repeatedly over many trials. The possibility exists, therefore, that participants quickly learned the object’s shape and/or distance as an invariant function of its angular size or its projected binocular disparity. Indeed, Haffenden and Goodale [7] demonstrated that such learning is characteristic of human reaching performance. The initial purpose of the present experiment therefore was to review the relative performance of participants to reach and grasp for objects viewed monocularly and binocularly when such strategies were prevented. To do this, a set of objects was carefully selected so that some object–distance combinations projected the same binocular disparity (front–back) and other object–width combinations projected the same angular size at each of the viewing distances used. Object height was randomised between trials. Therefore neither the width, height, depth or binocular disparity of any object could be used to cue uniquely the object’s identity or its distance.

The present paper, therefore, first builds on the design of Servos et al. [23] by excluding such potential artefacts so that any difference in monocular and binocular performance can be determined unambiguously. One aspect of this design is that when binocular cues are subtracted from a normal rich scene there are often sufficient sources of information left to support the behaviour of interest due to the redundancy of the visual information in the original scene. This is exactly what happens, of course, in the case of monocularly guided reaching. A simple experiment at one’s desk will demonstrate that efficient prehension movements are still possible under monocular viewing. Indeed, Servos et al. [23] only reported systematic biases produced by the elimination of binocular information — variability did not increase — which suggests that reliable information about the extrinsic and intrinsic properties of objects could be specified by monocular information: only certain kinematic indices (such as peak velocity and acceleration) were affected. Whether this reflects the pre-eminence of binocular cues in the control of reaching or a switch to a more conservative strategy due to the loss of the normal range of visual cues remains unclear. Obviously other visual information can be used to specify the extrinsic and intrinsic properties to control behaviour as performance persists despite the perturbation.

The second aim of the present experiment, therefore, was to address this issue in more detail by assessing reaching performance in impoverished viewing conditions where the rich array of monocular cues is not available. To accomplish this, participants reached for self-illuminated objects, viewed monocularly or binocularly, in an otherwise dark room. In this condition many of the monocularly available visual cues are eliminated and so the influence of binocular cues on prehension can be examined in near-isolation (cf. the logic of random–dot–stereograms). Not only should this condition highlight the role of binocular information but it also enables us to determine the relative contribution of scene-based information in the control of prehension. Scene-based information refers to both the visual cues from the surrounding scene and to the rich visual information regarding body position and ego-movement which are drastically reduced in the unlit conditions (feedback about finger and thumb position was provided).

As Servos et al. [23] showed, the kinematic indices of a typical prehensile movement can reveal clear differences in the effects of visual information on the transport and grasp components. The transport component has been found to vary primarily as a function of extrinsic information such as egocentric object distance and can be described in terms of a ‘velocity profile’ of the wrist throughout a movement. Typically, this consists of a smooth bell shaped curve, with a slow movement phase as the hand makes a final approach to the object. The shape of this curve has been found to vary systematically with egocentric object distance, with ‘markers’ such as peak velocity and peak deceleration increasing reliably with increasing object distance [12], [22], [23]. The grasp component has been found to vary primarily as a function of intrinsic cues such as object size. When preparing to grasp, the hand is typically first opened wider than the size of the target object, then closed to meet the object during the final phase of the movement. The peak opening of the grasp has been found to scale reliably with object size along the dimension of the opposable grasp points [5], [12], [23]. Each of these indices are determined here for monocular and binocular viewing and in both the lit and unlit conditions.

In summary, the present study investigated the contribution of binocular viewing to the control of prehension movements. Four viewing conditions were employed which varied monocular/binocular viewing in lit/unlit conditions. In one condition neither binocular nor scene-based information was available and so the principal cues to distance and size were retinal size, height in the visual field and accommodation. Therefore performance in this condition might be expected to be severely disrupted and the addition of binocular information may be found to have a critical effect on performance. The subtraction of binocular information from normal lit viewing (cf. [23]) however, may yield only rather subtle effects, as observed previously, due to the remaining monocularly available pictorial depth cues and rich cues regarding ego-motion. However, in the present experiments we can exclude categorically that any preservation of performance is due to learned associations between the object–distance combinations and their projected angular size or binocular disparity.

Section snippets

Participants

Ten right-handed adult volunteers (eight males, and two females; mean age=29.2 years) participated in the experiment. All had normal or corrected to normal vision and stereoacuities of at least 40 arcsec as assessed by TNO test (Alfred Poll Inc., NY). Nine participants were right-eye dominant and one was left-eye dominant.

Apparatus

Participants sat at a matt black table approximately 80×80 cm. Head position was maintained using a chin rest. The start point for each trial was a 2 cm diameter microswitch

Results

Individual mean values were calculated for each object by distance combination for each of the four viewing conditions. For every dependent variable, the means for each participant were then entered into a 4×3×3×3 (viewing condition×object distance×object width×object depth) repeated measures analysis of variance. Table 1 shows the overall mean and standard error for each dependent measure for each of the experimental conditions, as well as the F values for the main effect of viewing condition.

Discussion

The present study was designed to assess the role of binocular information in the control of reaching movements in the presence and absence of scene-based information. Particular attention was paid in the design of the experiment in order to control extraneous visual information which could cue object size or distance information so that the influence of binocular cues could be clearly discerned. An interesting and clear dissociation of the role of binocular information in the transport and

Acknowledgements

We would like to acknowledge the contributions of Richard Eagle, Paul Hibbard and Rob van der Willigen to this work. This research was supported by grants from the Wellcome Trust and The Royal Society. S. J. Watt is supported by a BBSRC Special Research Studentship.

References (27)

  • P. Haggard et al.

    Assessing and reporting the accuracy of position measurements made with optical tracking devices

    Journal of Motor Behavior

    (1990)
  • S.R. Jackson et al.

    A kinematic analysis of goal-directed prehension movements executed under binocular, monocular, and memory-guided viewing conditions

    Visual Cognition

    (1997)
  • L.S. Jakobson et al.

    Factors affecting higher-order movement planning: a kinematic analysis of human prehension

    Experimental Brain Research

    (1991)
  • Cited by (0)

    View full text