RESEARCH ARTICLE

Dissociation between vergence and binocular disparity cuesin the control of prehension

Dean R. Melmoth Æ Mithu Storoni Æ Georgina Todd ÆAlison L. Finlay Æ Simon Grant

Received: 16 June 2006 / Accepted: 20 June 2007 / Published online: 31 July 2007

� Springer-Verlag 2007

Abstract Binocular vision provides important advanta-

ges for controlling reach-to-grasp movements. We exam-

ined the possible source(s) of these advantages by

comparing prehension proficiency under four different

binocular viewing conditions, created by randomly placing

a neutral lens (control), an eight dioptre prism (Base In or

Base Out) or a low-power (2.00–3.75 dioptre) Plus lens

over the eye opposite the moving limb. The Base In versus

Base Out prisms were intended to selectively alter ver-

gence-specified distance (VSD) information, such that the

targets appeared beyond or closer than their actual physical

position, respectively. The Plus lens was individually

tailored to reduce each subject’s disparity sensitivity (to

400–800 arc s), while minimizing effects on distance

estimation. In pre-testing, subjects pointed (without visual

feedback) to mid-line targets at different distances, and

produced the systematic directional errors expected of

uncorrected movements programmed under each of the

perturbed conditions. For the prehension tasks, subjects

reached and precision grasped (with visual feedback

available) cylindrical objects (two sizes and three loca-

tions), either following a 3 s preview in which to plan their

actions or immediately after the object became visible.

Viewing condition markedly affected performance, but the

planning time allowed did not. Participants made the most

errors suggesting premature collision with the object

(shortest ‘braking’ times after peak deceleration; fastest

velocity and widest grip at initial contact) under Base In

prism viewing, consistent with over-reaching movements

programmed to transport the hand beyond the actual target

due to its ‘further’ VSD. Conversely, they produced the

longest terminal reaches and grip closure times, with

multiple corrections just before and after object contact,

under the Plus lens (reduced disparity) condition. Base Out

prism performance was intermediate between these two,

with significant increases in additional forward movements

during the transport end-phase, indicative of initial under-

reaching in response to the target’s ‘nearer’ VSD. Our

findings suggest dissociations between the role of vergence

and binocular disparity in natural prehension movements,

with vergence contributing mainly to reach planning and

high-grade disparity cues providing particular advantages

for grasp-point selection during grip programming and

application.

Keywords Reaching � Grasping � Pointing � Stereopsis �Visuomotor behaviour � Human

Introduction

The proficient control of prehension requires visual com-

putation of the precise three-dimensional (3D) properties of

the goal object within the scene, combined with vision of

the moving limb (see Jeannerod 1986). A major interest in

prehension research is to determine whether specific types

of visual information are required to mediate skilled per-

formance from among the range of potential 3D-cues

D. R. Melmoth � M. Storoni � G. Todd �A. L. Finlay � S. Grant (&)

Department of Optometry and Visual Science,

The Henry Wellcome Laboratories for Visual Sciences,

City University, Northampton Square,

London, EC1V 0HB, UK

e-mail: s.grant@city.ac.uk

Present Address:M. Storoni

Department of Ophthalmology,

Norfolk and Norwich University Hospital,

Colney Lane, Norwich, NR4 7UY, UK

123

Exp Brain Res (2007) 183:283–298

DOI 10.1007/s00221-007-1041-x

typically available in most everyday settings. Several

studies have contributed significantly to this enquiry by

comparing the competency of reach-to-grasp movements

performed by normal adults using binocular vision or with

one eye occluded (Servos et al. 1992; Jackson et al. 2002;

Watt and Bradshaw 2000, 2003; Loftus et al. 2004a;

Melmoth and Grant 2006) or in which an initial binocular

view was suddenly replaced by monocular vision during

movement execution (or vice versa) (Servos and Goodale

1994; Jackson et al. 1997; Bradshaw and Elliot 2003).

Most of these found that viewing condition made little

difference to initial hand transport—including velocity

scaling with object distance—implying that binocular cues

to target location are not needed for efficient reach pro-

gramming. In contrast, binocular vision has been repeat-

edly shown to improve the speed and accuracy of the

movement end-phase. These advantages have been attrib-

uted to the efficient use of binocular disparity cues in

computing the 3D-shape of the goal object and, particu-

larly, in providing on-line feedback about hand-target

depth during the terminal reach and the grasp.

These studies, however, suffer two main limitations.

First, although intuitively appealing, they provide no direct

evidence that disparity processing is the source of the

binocular advantages: they could result, instead, from

the presence of matching (i.e. concordant) information in

the two eyes (e.g. Jones and Lee 1981). Second, when

vision is restricted to one eye subjects usually open their

hands wider and further from the object at grip pre-shaping,

and then markedly slow their approach to it. These deficits

in performance also occur when no vision of the hand and/

or target is available after movement onset (Jakobson and

Goodale 1991; Berthier et al. 1996; Churchill et al. 2000;

Watt and Bradshaw 2000; Loftus et al. 2004a). Because of

this, it has been argued (e.g. Watt and Bradshaw 2000;

Loftus et al. 2004b) that they are not necessarily a specific

consequence of removing binocular information, but more

likely represent the default response of the prehension

system to all situations of degraded vision, of which

occluding one eye is just an example.

An alternative approach is to examine the effects of

more subtle perturbations of binocular cues on visuomotor

control. Among the few previous attempts to do this are

two important investigations of altering vergence-specified

distance (VSD) information, by the use of prisms, on

pointing accuracy (Tresilian et al. 1999) and prehension

kinematics (Mon-Williams and Dijkerman 1999). The

underlying principle is illustrated in Fig. 1. Placing a Base

In prism over one or both eyes forces a greater-than-normal

divergence angle in order to bi-foveate the target (Fig. 1b),

and so should cause the subject to over-estimate its phys-

ical distance: that is, to judge the target as being further

away than it actually is. Conversely, Base Out prisms force

more binocular convergence (Fig. 1c), which should cause

the subject to judge the target’s distance as being closer

than its real position.

Tresilian et al. (1999) demonstrated that a 5 dioptre (D)

prism oriented Base In or Base Out over one eye (cf.,

Fig. 1) resulted in systematic pointing-in-depth errors, in

which subjects over- or under-shot the target, consistent

with its altered VSD. These errors were observed under

open-loop conditions, but occurred even when multiple

(and conflicting) monocular distance cues were available

prior to movement onset, showing that VSD information

was incorporated into the programmed pointing behaviour.

Mon-Williams and Dijkerman (1999) placed a pair of 9Dprisms oriented Base In or Base Out over both their par-

ticipants’ eyes just before they reached and grasped dif-

ferent objects at three mid-line distances in a cue-rich

environment and with feedback available. When viewing

through the Base In prisms the peak acceleration (PA) and

peak velocity (PV) of the reach were significantly

increased and the time spent decelerating (after PV) was

significantly reduced, compared to the Base Out condition.

Since the PA and PV of the reach normally increase with

target distance, Mon-Williams and Dijkerman (1999)

explained these faster early kinematics when viewing Base

In as consistent with an initial over-aiming movement to

objects whose VSD made them appear further away than

they were. Similarly, they suggested that the slower early

reaches under Base Out conditions resulted from an initial

under-aiming for targets judged from their VSD as being

nearer, followed by a necessary extension of the movement

end-phase in order to physically grasp them. That is,

Fig. 1 Predicted effects on 3D-spatial localization of a target fixated

binocularly with b a Base In prism, c a Base Out prism and d a low-

power spherical Plus lens over the left eye, compared to a a Plano lens

of neutral refractive power. b The Base In prism requires extra

divergence to fixate the target, so that it should appear further to the

left and beyond its physical location (i.e. as ‘far’); c The Base Out

prism requires extra convergence to fixate the target, so that it should

appear further to the right and closer than its physical location (i.e. as

‘near’); d The Plus lens should slightly magnify the target, causing it

to appear closer to the left eye (i.e. to the left and nearer) than it

actually is

284 Exp Brain Res (2007) 183:283–298

123

contrary to inferences derived from several studies that

simply compared binocular with monocular performance

(e.g. Watt and Bradshaw 2000, 2003; Melmoth and Grant

2006), they concluded that vergence cues contribute

importantly to the programming of hand transport.

An important negative finding in the Mon-Williams and

Dijkerman (1999) study was that prism orientation had no

effect on their subjects’ peak grip aperture (PGA). Yet if

distances appeared further or nearer under the two condi-

tions, size-constancy mechanisms involving convergence

micropsia and divergence macropsia should also cause the

objects to be judged as, respectively, larger and smaller.

This, in turn, should have affected the PGA, since this is an

early (if indirect) programmed index of the estimated size

of the object to be grasped (e.g. Jeannerod 1986). Mon-

Williams and Dijkerman (1999) suggested that their failure

to observe this may have been due to their use of objects

that were of equal size in the depth-plane (i.e. opposition

axis) of the thumb and finger grasp points. Indeed, they

recommended that future studies vary this object attribute

when attempting to identify the binocular cues involved in

controlling the grip.

Two studies have since attempted this, using computer-

generated ‘virtual’ objects that were defined only by dis-

parity information (Hibbard and Bradshaw 2003) or in which

this information was removed by presenting separate,

matching images of the targets (‘bi-ocularly’) to each eye

(Bradshaw et al. 2004). Participants in both these experi-

ments opened their grips wider, prolonged their movements

and made more on-line corrections on disparity-altered and

concordant compared to normal binocular trials, as monoc-

ular observers tend to do. However, both studies were con-

ducted in relatively cue-sparse environments and with

neither the hand nor target visible during the movements. As

their authors acknowledge, the results thus only hint at a

possible advantage of binocular disparity for programming

more ‘natural’ grasps, and do not address whether its primary

function is to provide fast and accurate feedback for the on-

line control of the hand, as they and others have suggested

(e.g. Watt and Bradshaw 2000, 2003; Bradshaw and Elliot

2003; Loftus et al. 2004a; Melmoth and Grant 2006).

Another way of altering disparity information is to place

a positive (Plus) spherical lens of relatively low power over

one eye. This has two main optical consequences (Rabbetts

1998): it slightly defocuses all parts of the retinal image

and magnifies it by a factor of *1% per dioptre. The effect

of the defocus is to progressively attenuate the contrast of

low-to-high spatial frequencies in the affected eye and to

deprive binocular mechanisms of the same higher acuity

information. Previous work has shown that such reductions

in spatial resolution leave only ‘coarse’ disparities acces-

sible to the observer and cause greater decrements in bin-

ocular stereoacuity than when Plus lenses of equivalent

strengths are placed over both eyes (Levy and Glick 1974;

Wood 1983; Goodwin and Romano 1985). These findings

are consistent with evidence (Blakemore and Hague 1972;

Felton et al. 1972; Schor et al. 1984; Yang and Blake 1991)

that ‘fine’ disparity processing normally involves high

spatial-frequency filters in the human visual system, and

that these are particularly vulnerable to mismatched bin-

ocular input, as occurs with monocular image degradation.

We have taken advantage of this in the present study,

randomly inter-leaving a Plus lens with a neutral (control)

lens and with an 8D prism oriented either Base In or Base

Out over the same eye. Another advantage of this design is

that the other eye always received the same unrestricted

view, but with different types of conflicting 3D information

present in each experimental condition. A disadvantage is

that object locations should appear to shift laterally, as well

as in depth, under these conditions (see Fig. 1). In the case

of the prism, this is because its specific optical effect should

be to displace the image laterally, whereas for the Plus lens

the effect should occur because the magnified image size

causes objects to appear nearer to the affected eye. These

assumptions were confirmed in a preliminary pointing test

in which subjects exhibited the directional errors expected.

Since the hand takes a curved trajectory when approaching

the target in a natural reach (e.g. Jakobson and Goodale

1989; Melmoth and Grant 2006), we minimized this prob-

lem by always perturbing the eye opposite to the subjects’

preferred hand, so that the target lay further or closer along

the reach path in the Base In and Base Out prism conditions

of the prehension experiments. In these, we sought to re-

examine Mon-Williams and Dijkerman’s central findings

on the use of vergence in reach programming and to assess

the effect of reducing disparity sensitivity on performance.

As a further extension of the earlier work, subjects in two

separate experiments were either permitted a fairly lengthy

(3 s) preview of the task or started to move as soon as the

target became visible (cf., Mon-Williams and Dijkerman

1999). Our rationale for this was that allowing some sub-

jects to see the goal object for a period prior to movement

onset should improve the planning process and reduce their

reliance on subsequent visual feedback (see Glover 2003;

Melmoth and Grant 2006). Use of the two procedures were

thus attempts to accentuate possible dissociations between

vergence in prehension planning (Preview) and disparity in

its on-line control (No Preview).

Materials and methods

Participants

Subjects were screened to determine their visual status and

handedness prior to participation in the experiments, for

Exp Brain Res (2007) 183:283–298 285

123

which they were paid a small fee. All procedures were

approved by the City University Ethics Committee and met

the standards laid down in the 1964 Helsinki Declaration.

Subjects were young adults (aged 18–32 years), all of

whom had normal vision or were corrected-to-normal

through contact lens wear, with log Minimum Angle of

Resolution (logMAR) visual acuity (VA) in each eye of

� 0.0 (Snellen equivalent, 6/6 or better) and high-grade

binocular stereo vision. This latter was manifest by (1)

crossed and uncrossed stereoacuity thresholds of 40 arc s

(the lowest detectable) on the Wirt–Titmus stereogram test

(Stereooptical Co. Inc., Chicago, IL, USA) and (2) motor

fusion of � 40D Base Out (convergence) and � 16D Base

In (divergence) assessed using a variable prism bar

(Clement Clarke International, Cambridge, UK). These

binocular tests were conducted at a ‘near’ distance (40 cm)

similar to those used in the experiments. Subjects were

aware that aspects of their visuomotor behaviour were to be

studied under different viewing conditions, but were naıve

as to the specific purposes of this.

Binocular viewing conditions: Plano, prism

and Plus lenses

Performance on four different binocular viewing condi-

tions was compared in each experiment. To create these

conditions, subjects wore a set of optometric trial-frames

(The Norville Group Ltd., Gloucester, UK), with a dif-

ferent lens slotted into the frame in front of the eye

opposite their preferred hand prior to each movement. The

other frame always remained clear, to provide an unen-

cumbered view from the ipsi-manual eye. A Plano lens

(i.e. with no refractive power) which allowed normal

binocular viewing was used as the control condition. To

alter VSD information, an 8D Prism (also with no

refractive power) was used, with its base directed either in

or out, so requiring participants to either increase or

decrease their normal vergence angle, respectively, to bi-

foveate the target presented (Note: 1 prism D, by defini-

tion, is the angle with a tangent of 0.01 and displaces the

image laterally by 1 cm at a viewing distance of 1 m).

Disparity sensitivity was unaffected by either prism

orientation, as determined by re-testing of each subject’s

Wirt–Titmus stereo-thresholds, which remained at

40 arc s under both modified vergence conditions. To

alter disparity information, a Plus (spherical) lens was

employed with the specific power (between +2.00 and

+3.75 D) required to reduce each participant’s crossed

stereo-threshold to between 400 and 800 arc s. This was

achieved by identifying the lowest lens power that

degraded the subject’s near depth percept for the No. 2

Wirt–Titmus test circle (400 arc s), but not for the No. 1

test circle (at 800 arc s). There were several reasons why

we chose to elevate their stereo-threshold by this rela-

tively small amount. First, there is evidence that adult

subjects with similarly low-grade stereopsis (resulting

from childhood amblyopia) exhibit prehension deficits

(Grant et al. 2007). Second, the customized defocusing

power reduced the logMAR VA of the subjects’ affected

eye, but only to an average of 0.82 (range 0.72–0.96;

Snellen equivalents *6/30–6/60), so that spatial resolu-

tion was still between *5 and 10 min arc, representing

preservation of spatial frequencies below 3–6 cycles per

degree. Third, it had no effect on the motor fusion limits,

as assessed by re-testing with the prism bar, in some

(n = 10) of the participants, while reducing the Base Out

and/or Base In prism response—to respective maxima of

30D and 12D—in the others. These values are at the lower

end of the normal adult range for the limits of converge

and divergence, and exceed the challenge posed by the 8Dprism used in the experiments.

Viewing conditions were further controlled by having

the subjects wear a pair of liquid crystal PLATO goggles

(Translucent Technologies Inc., Toronto, ON, Canada)

over the Norville trial-frames. The goggles remained opa-

que while the different lenses were initially inserted into

the frame, but cleared simultaneously (opening times

<2 ms) to signal that the next trial had begun. Lenses were

removed after each completed movement. The viewing

condition and target employed were randomized from trial-

to-trial. These procedures prevented the subjects from

identifying the lens in place and reduced repetitive move-

ments to the same object.

Hand movement recordings

All experiments were conducted in a standard well-lit

laboratory environment, in which a variety of (monocular)

visuo-spatial cues were always present. Subjects sat at a

matt black table (60 cm wide · 72 cm deep) with their

eyes roughly level with its near edge and *40–50 cm

above its surface. A circular (30 mm diameter) ‘start but-

ton’ was secured to the table 12 cm from this edge and in

line with the subjects’ mid-sagittal axis. Small (7 mm

diameter) light-weight infra-red (IR) reflective markers

were attached to their preferred hand (and the target). The

instantaneous 3D-positions of these were recorded, at a

sampling rate of 60 Hz and with an accuracy of � 0.4 mm,

by three IR-emitting and detecting Motion Capture

Cameras (ProReflex system, Qualisys AB, Stockholm,

Sweden) wall-mounted *1.5 m above the workspace.

Recording onset was synchronized with the specific ‘go’

signal used (e.g. clearing of the liquid crystal shutters) and

terminated 3 s later.

286 Exp Brain Res (2007) 183:283–298

123

Pre-testing: pointing in depth

Ten right-handed subjects (five males) aged 18–32 (mean

25 years) participated in preliminary tests aimed at deter-

mining the nature and extent of the spatial localization errors

induced by the prism and the Plus lens. Subjects sat with the

index finger of their preferred hand resting on the raised

centre of the start button and with an IR marker placed on the

tip of this finger-nail. An identical marker served as the

target. This was attached to the end of a transparent Perspex

rod lying on the table surface directly below a 5 mm thick

sheet of glass (30 · 30 cm) supported by plinths (12 mm

high) at each corner. Target presentations were at one of two

mid-line distances, 25 or 40 cm from the centre of the start

button. Subjects were instructed to reach out and place their

index fingertip ‘as accurately as possible’ directly over the

target and to hold their end-point finger position stationary

for *1 s before returning to the resting location.

Subjects pointed in depth under each viewing condition

in two open-loop (i.e. no visual feedback) Procedures with

different planning times available. In the Preview Proce-

dure, opening of the PLATO goggles was followed by a 3 s

interval in which the subjects could examine the scene,

with the cue to move signalled by the closing of the goggle

shutters at the end of this period. In the No Preview Pro-

cedure, the initial opening of the goggles was the cue to

move, but both shutters abruptly closed 500 ms later. This

viewing period was selected because it represents the

average movement onset for prehension in normal adults

(e.g. Servos et al. 1992; Melmoth and Grant 2006). We

informed our subjects of this and specifically advised them

not to try to complete their movement in the brief viewing

time allotted, as this would be impossible.

To ensure that the subjects properly followed our

instructions, they were given 4–6 practices of each proce-

dure using normal binocular vision to a mid-line target

distance (32.5 cm) not used during subsequent testing. They

then completed a block of 32 trials, with each combination

of parameters (4 lenses · 2 distances) repeated four times

in a random order. Equal numbers of participants began

with the Preview or with the No Preview trial block. Note

that the subjects’ pointing performance in both procedures

was based on target distance information available in the

initial view. This was done because we wanted to check that

the optical distortions affected distance judgements at the

planning stage, and this would have been obviated if they

had the opportunity to correct their movement using in-

flight visual feedback. At the end of the session, however,

participants completed several perturbation trials with

vision available throughout. It was clear from their (sur-

prised) reactions, especially to the larger displacements

induced by the prism, that they were previously unaware of

them (cf., Jakobson and Goodale 1989).

The overall pointing accuracy of each subject under

each viewing condition was determined from the spatial

offset between the final finger marker and target positions.

The difference between these positions was also calculated

for each trial in both the forward (x-axis) and lateral (y-

axis) directions, so that constant pointing errors indicative

of systematic target over- or under-shoot could be deter-

mined independently for each direction. End-point finger

positions that over-shot the target in either the forward or

lateral directions (i.e. that landed beyond or to the left of

the target in our right-handed subjects) were assigned

positive values, with under-shoots (i.e. nearer than or to the

right of the target) given negative values.

Experiments 1 and 2: prehension

The effects of selective alterations in binocular viewing on

reaching and grasping performance were also examined in

two experiments with different planning times available.

Eight right-handed subjects (five males) aged 18–29 (mean

21 years) participated in Experiment 1 in which opening

of the PLATO goggles was followed by a 3 s Preview,

and then by an auditory tone signalling that the movement

should begin. Ten different subjects (six males and six

right-handers) aged 20–32 (mean 25 years) participated in

Experiment 2 in which there was No Preview, with

opening of the PLATO goggles at the start of the trial also

representing the cue to move. Note that, unlike the

pointing test, subjects in both prehension tasks completed

their movements under closed-loop conditions that

enabled them to modify their performance using on-line

visual feedback (cf., Mon-Williams and Dijkerman 1999).

This was necessary because a key objective was to

determine whether in-flight control of the hand was

selectively disrupted by reducing disparity information in

the movement end-phase.

In both experiments, subjects sat at the table gripping

opposite poles of the circular start button between thumb

and index finger of their preferred hand. Three IR markers

were attached to this hand: one at the wrist (styloid process

of the radius), and one each at the opposing distal borders

of the thumb- and index finger-nails. Participants were

required to reach out and use a precision grip to pick up an

isolated target object on the table in one ‘swift, natural and

accurate’ movement, then place it to one side and return

their hand to the start position. Practice trials (typically six)

were run under normal binocular viewing to an object and

position not used again later, to confirm that the task was

performed according to the instructions given. Targets in

the experimental trials were cylindrical household objects

of ‘small’ (24 mm) or ‘large’ (48 mm) diameter and

bearing an IR marker on their upper surface. These were

Exp Brain Res (2007) 183:283–298 287

123

placed at one of three locations relative to the starting hand

position: either ‘near’ (25 cm) along the mid-line axis or at

‘far’ (40 cm) distances and 7.5 cm eccentric to the mid-

line in ipsilateral or contralateral hemi-space with respect

to the moving limb. Subjects in both experiments com-

pleted four blocks of the 24 possible randomized trial

combinations, with brief rests in between.

Recorded data were processed using purpose-written

routines in MATLAB software (MathWorks Ltd., Cam-

bridge, UK) which generated wrist velocity, spatial path

and grip aperture ‘profiles’ of each movement along with

a number of dependent kinematic measures. For these, the

moment of initial object contact (OC)—when the target

was first displaced by � 1 mm from its original posi-

tion—was used as a transitional landmark between the

reach and the grasp, with target displacement of � 10 mm

defined as the movement endpoint (ME) and usually

indicating that the object was just being lifted. These

determinations were established empirically, as discussed

previously (Melmoth and Grant 2006). Performance was

initially analysed from ten dependent measures obtained

from each trial:

1. Peak Acceleration (PA): the maximum acceleration

of the wrist marker during the reach.

2. Peak velocity (PV): the maximum velocity of the

wrist marker during the reach.

3. Velocity at Object Contact (VOC): the velocity of the

wrist marker at OC.

4. Time to Object Contact (TTC): the time from

Movement Onset (MO)—when the marker on the

wrist first exceeded a forward velocity of 50 mm/s—

to the point of OC (Note: the criterion for MO was

used to eliminate ‘false starts’ caused by small

spontaneous hand movements in the resting position).

5. Time to peak deceleration (ttPD): the time from MO

to the peak deceleration (PD) of the wrist marker

during the reach.

6. Terminal Reach Duration (TRD): the time in the low-

velocity or ‘braking’ phase, from PD to OC.

7. Peak Grip Aperture (PGA): the maximum aperture

between the thumb and finger markers during hand

pre-shaping [Note: to correct for between-subject

differences in hand size, this value was expressed as

the distance between the inner surfaces of the two

digits (as in Melmoth and Grant 2006) by first

determining the mean distance between the two digit

markers on the 30 mm start button for each partic-

ipant, and then subtracting this value (�30) from all

their grip aperture data].

8. Grip at Object Contact (GOC): the (corrected)

aperture between the thumb and finger markers at OC.

9. Grip Closure Time (GCT): the time from PGA to OC.

10. Grip Application Time (GAT): the time from OC to

the ME, when the object was usually just being

secured prior to lifting.

These measures were chosen because they should be

influenced by each experimental condition and in different

ways. The Base In prism should cause the objects to appear

further away—and, hence, also larger—than they were,

resulting in the programming of faster reaches (increased

PA and PV) and wider grips (increased PGA), since these

respective variables are known to increase linearly with

increasing target distance and size (e.g. Jeannerod 1986;

Kudoh et al. 1997). The planned over-reaching movements

should also result in an increased ttPD, but with corrections

likely to occur in the late deceleration phase (Mon-

Williams and Dijkerman 1999) marked by sudden braking

and hand closure (reduced TRD and GCT) and/or ‘harder’

object contacts (increased VOC and GOC), which might

reduce the overall reach duration (TTC). The Base Out

prism should result in opposite effects on the measures of

movement planning (reduced PA, PV, ttPD and PGA),

since subjects ought to judge the targets as being nearer—

and smaller—than they were, while requiring increases in

TTC, TRD and GCT to correct their initial under-reaching

movement. As outlined in the Introduction, it is generally

proposed that disparity cues underlie the advantages of

binocular vision in computing object size/shape for plan-

ning the grasp and in providing essential feedback about

target-hand relations for efficient on-line movement control

(e.g. Melmoth and Grant 2006). Reducing disparity infor-

mation with the Plus lens would thus be expected to

reproduce the deficits typically associated with monocular

vision, with uncertain grip programming (i.e. wider PGA)

accompanied by prolonged and/or inaccurate terminal

reaches (i.e. extended TTC, due to an increased TRD)

and grasps (increased GCT, GOC and GAT).

The profiles of each movement were also inspected for

the different types of error expected of the perturbed con-

ditions. These were: (1) ‘Collisions’ due to planned target

over-reaching (predicted for the Base In prism), in which

there was little evidence of a low-velocity phase just before

object contact (see Fig. 5b) nor of digit closure after PGA

(see Fig. 6b); or in which the object was actually knocked

over; (2) ‘Short-falls’ due to planned target under-reaching

(predicted for the Base Out prism), in which an extra for-

ward movement in the spatial hand path was required in

order to establish object contact (not shown, but see Mel-

moth and Grant 2006, Fig. 2c); (3) ‘Under-aims’ resulting

from under-estimation of target location or uncertainty

about hand-target depth (predicted for the Base Out prism

and Plus lens, respectively), in which there were changes in

hand velocity (Fig. 5c) or in the grip aperture (Fig. 6c) just

prior to object contact; and (4) ‘Mis-grasps’ resulting from

288 Exp Brain Res (2007) 183:283–298

123

inaccuracies in initial contact (predicted for the Plus lens),

in which the hand velocity (Fig. 5d) or grip aperture

(Fig. 6d) was adjusted after object contact or in which

contact was prolonged prior to lifting (� 150 ms, see

Melmoth and Grant 2006).

The mean kinematics and number of errors were cal-

culated across all trial types under each viewing condition

for each subject and analysed by Huynh-Feldt adjusted

repeated-measures ANOVA (SPSS UK Ltd., Woking, UK),

with ‘Procedure’ (Preview and No Preview) as a between-

subjects factor. Significant main effects of viewing

condition were examined by planned Least Significant

Difference (LSD) pair-wise comparisons to identify the

source(s) of the effect. Some selected kinematic measures

were subsequently entered into more complete (4 lens · 3

location · 2 size) ANOVAs to examine possible interac-

tions between viewing condition and the objects’ proper-

ties, and were probed for biases in performance via an

analysis of quartiles. For this latter analysis, the values

obtained for the given measure for each subject were rank

ordered, and the relative proportions of the top and the

bottom quartiles (i.e. 25%) of the data set represented by

each viewing condition were determined. Significant over-

or under-representation of each lens within these quartiles

was assessed separately, as above.

The Fisher LSD procedure was chosen for the pair-wise

comparisons at it is generally the most sensitive post hoc

test for grouped contrasts following rejection of the null

hypothesis and, hence, most appropriate for examining the

relatively subtle effects that our experimental manipula-

tions had on performance. The procedure calculates the

error mean square of the ANOVA table and so corrects for

other between-condition errors, but with less adjustment

with regard to the errors associated with the specific pair of

contrasts under scrutiny. In particular, it avoided some

anomalous outcomes that occurred when we used a much

stricter (Bonferroni) adjustment, in which no significant

sources were identified to account for a main effect of

viewing condition suggested by the preceding ANOVA. In

our post hoc analyses, an alpha level of <0.05 was taken as

significant. Since we report the outcomes of *80 pair-wise

comparisons, however, a few Type-1 errors would be

expected at this probability level. Because of this, we have

exercised caution in interpreting findings with alpha scores

between 0.05 and 0.01, none of which were revealed as

significantly different when applying the Bonferroni test

post hoc. There were, however, 15 such outcomes, many

more than would be expected from chance alone.

Results

Pointing in depth

Pointing performance in the control condition for both dis-

tances in both the No Preview and Preview Procedures was

quite accurate, with an overall mean absolute error of

1.3 cm—equivalent to a Weber fraction of <3% of the

absolute viewing distances—although different subjects

tended to either over- or under-shoot the target, mainly in the

forward direction. Thus to directly evaluate the effects of

each binocular perturbation, performance in these was

determined relative to their own baseline (Plano lens)

pointing accuracy. Table 1 show the mean (±the average

standard error) normalized directional pointing responses

for each experimental lens. Overall, the Base In prism

caused the subjects to point beyond the target in both the

forward and lateral directions (positive errors) relative to

control viewing, the Base Out prism caused consistent

under-shooting in both directions (negative errors), and the

Plus lens resulted in the subjects pointing nearer in the for-

ward direction, but further laterally. These constant errors

are in the directions—although smaller in magnitude—

predicted from simple optical geometry (see Fig. 1). The

mean normalized errors of each subject in each direction

were submitted to separate (3 lens · 2 distance) · 2 Pro-

cedure repeated measures ANOVA. These revealed signif-

icant main effects of viewing condition for both directions

(F(2,36) > 18.0 and P < 0.001), with no effect of distance or

Procedure in either direction (P > 0.6, for both sets of

comparisons). Post hoc LSD tests showed that the differ-

ences between each altered view were reliable (P < 0.001),

except for the similar extent of forward under-shooting in

the Base Out versus Plus lens conditions (P = 0.25).

Figure 2 summarizes the data obtained for each altered

binocular view collapsed across target location and

Table 1 Mean (±average SE) normalized pointing errors by altered viewing condition

Procedure Target distance Base In versus Plano Base Out versus Plano Plus versus Plano

Forward (mm) Lateral (mm) Forward (mm) Lateral (mm) Forward (mm) Lateral (mm)

No Preview Near (25 cm) +6.4 (2.3) +11.5 (1.3) �6.9 (2.3) �6.2 (1.4) �3.8 (1.8) +2.2 (1.8)

Far (40 cm) +5.0 (2.0) +10.8 (2.1) �4.3 (2.1) �15.2 (2.4) �3.3 (3.9) +3.4 (1.7)

Preview Near (25 cm) +5.9 (1.7) +7.4 (2.9) �5.3 (2.2) �8.4 (2.1) �2.5 (1.8) +2.2 (1.6)

Far (40 cm) +3.4 (3.2) +11.8 (2.4) �4.5 (3.4) �6.7 (1.1) �2.4 (4.1) +2.8 (1.4)

Exp Brain Res (2007) 183:283–298 289

123

Procedure. There are two key points to note. First, we can

estimate the weighting attached by the subjects to the

altered binocular information, from the ratio between

the observed normalized pointing responses and the size of

the errors expected, where a value of 1 (or 100%) repre-

sents total reliance on this information (see Tresilian et al.

1999). For example, based on an average eye-height above

the table of *45 cm (see Methods), the absolute viewing

distances for the near and far targets would be *50 and

*60 cm, respectively. From simple optics, the 8D prism

should—by definition—laterally displace the images by

*40 and *48 mm at these two distances, so the observed

mean lateral pointing errors of *10 mm (Table 1) imply

relative weightings of between 20 and 25% for these

altered conditions. Similarly, a Plus lens of around +3.00 D

was typically used to reduce each subject’s stereoacuity,

which would have magnified the image by *3% compared

to the other eye, equivalent to an inter-ocular vertical size

ratio (VSR) of 1.03. Rogers and Bradshaw (1993) have

calculated the relationships between such ‘vertical’ dis-

parities, viewing distance (D) and object eccentricity (e), as

follows:

where IPD, represents the inter-pupillary distance.

Assuming that this was 6.5 cm, as in most ‘standard’

adults, a VSR of 1.03 is generated by targets with eccen-

tricities of *12.5� and 15� (or *110 and 160 mm from

the mid-line), respectively, at the two distances used. Since

our subjects pointed only 2.7 mm, on average, lateral to the

real mid-line targets, this implies that vertical disparity

cues received a weighting of <2.5% in the Plus lens con-

dition. In sum, these estimates suggest that the pointing

responses were a compromise between the conflicting cues

provided to each eye in the three experimental conditions,

but with the prism providing a greater distance cue-conflict

than the defocusing lens, as was intended. The second point

is that as the mean absolute pointing errors (Fig. 2) across

subjects were significantly smaller (F(2,78) = 9.8 and

P < 0.001) with the Plus lens compared to both Base In and

Base Out prism conditions (P = 0.001, for both compari-

sons), it should have the least effect on reaching behaviour.

Prehension

Viewing condition had major effects on both prehension

kinematics and movement errors, but the planning time

allowed did not. ANOVA with ‘Procedure’ as a between-

subjects factor revealed no significant main effects on any

parameter (with P > 0.5 in most cases), and only a few

reliable ‘Procedural’ interactions. This was not because

subjects given No Preview autonomously extended their

planning time, since average movement onsets after the

different ‘go’ signals used were similar in the two experi-

ments (Preview = 478 ms ± 98 SD; No Preview = 491

ms ± 82 SD, F(1,70) = 0.4 and P = 0.52). Clearly our

assumption that previewing the task would markedly affect

subsequent performance was mistaken, especially as it did

not affect within-subject pointing behaviour either. For this

reason, we present the combined data from all 18 subjects

in the two prehension experiments together, with the sig-

nificant interactions considered later.

Effects of viewing condition

Table 2 shows that the altered viewing conditions affected

most of the mean dependent prehension measures—the PA

of the reach and its initial duration up to PD being the only

exceptions—and planned comparisons (column 7) revealed

several predicted perturbation effects. First, our subjects

produced slightly faster (by 20 mm/s) peak reaching

velocities under Base In versus Base Out prism viewing

(P = 0.046), with Base Out reaches also appearing slower

(by 21 mm/s) than in the control condition (P = 0.027), so

Fig. 2 Observed effects on overall mean, normalized, 3D-pointing

errors. Filled symbols represent the actual target location; opensymbols show the average end-point finger positions. a The Base In

prism resulted in systematic over-shooting in both the forward and

lateral directions; b The Base Out prism resulted in systematic under-

shooting in both the forward and lateral directions; and c The Plus

lens resulting slight under-shoots in the forward direction and over-

shoots in the lateral axis

VSR ¼ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

D2/cos2eþ (D� tan e� IPD)þ (IPD2/4)� ��

D2/cos2e� (D� tan e� IPD)þ (IOD2/4)� �

q

;

290 Exp Brain Res (2007) 183:283–298

123

reproducing two earlier results of Mon-Williams and Dijk-

erman (1999). More strikingly, the Base In prism was

associated with significantly higher terminal reach veloci-

ties at OC (mean VOC = 97 mm/s) compared to all other

conditions (all P < 0.01) and with the widest grip aperture at

contact (mean GOC = 46 mm), a reliable increase com-

pared to both Plano and Plus lens viewing (both P < 0.01).

These effects are consistent with the subjects programming

a relatively fast reach based on the ‘further’ VSD of the goal

objects through the Base In prism, followed by a tendency to

collide with them because they were nearer than anticipated.

Second, subjects produced the longest overall reach dura-

tions (mean TTC = 834 ms) when viewing through the

defocussing lens, because they selectively extended its end-

phase (TRD) by *20–40 ms compared to the other condi-

tions, and they also spent the longest time (by *20–30 ms)

closing their grip just before contact in the Plus lens con-

dition (mean GCT = 247 ms). These deficits are consistent

with an uncertain final approach to the objects due to loss of

relative hand-target depth information. They were particu-

larly marked compared to control and Base In viewing

(P < 0.01, for all comparisons), but less so versus Base Out

performance (P = 0.02–0.047).

Indeed, a prolonged TRD and GCT were also predicted

effects of the Base Out prism, because subjects were

expected to ‘under-reach’ the actual target location based

on its ‘closer’ VSD. However, the small increases (of *5–

20 ms) in both parameters in this condition compared to

control and Base In viewing did not approach significance

(all P > 0.15). Main effects on the remaining grasping

parameters (i.e. other than the GCT) were due to consis-

tently better performance under normal binocular view-

ing—with reduced grip sizes at hand pre-shaping (PGA)

and object contact (GOC) followed by shorter manipula-

tion times (GAT)—compared to most or all of the per-

turbed conditions.

Performance was also generally less variable in the

control condition (see standard deviations, Table 2).

Quartiles analyses were conducted on selected measures,

including the PGA and GAT, to determine whether there

were perturbation-related biases in the kinematics that

were obscured within the overall mean data. However,

these added nothing new. Further analyses also showed

that parameters of the reach and grasp generally increased

with increasing target distance and size, respectively, but

that there were no significant interactions between these

object properties and viewing condition. This finding is

exemplified by the TRD (Fig. 3) which increased similarly

with object distance in all four conditions, while being

consistently prolonged when disparity sensitivity was

reduced by defocus. The exception was the terminal VOC,

which was unaffected by target location and always

greatest for the Base In prism (Fig. 4).Ta

ble

2E

ffec

tso

fv

iew

ing

con

dit

ion

on

reac

han

dg

rasp

kin

emat

ics

(mea

n±

SD

)

Dep

end

ent

mea

sure

Pla

no

Bas

eIn

Bas

eO

ut

Plu

sF

(3,4

8)

stat

isti

csL

SD

:p

airw

ise

com

par

iso

ns

Rea

chp

aram

eter

s

Pea

kac

cele

rati

on

(cm

/s2)

40

(6)

40

(7)

40

(7)

39

(8)

0.6

,P

=0

.6(n

s)

Pea

kv

elo

city

(mm

/s)

79

5(1

07

)7

94

(10

7)

77

4(9

3)

78

2(1

13

)4

.1,

P=

0.0

12

Pla

no

>B

ase

Ou

tP

=0

.02

7an

d>

Plu

sP

=0

.02

4;

Bas

eIn

>B

ase

Ou

tP

=0

.04

6

Vel

oci

tyat

ob

ject

con

tact

(mm

/s)

78

(17

)9

7(2

3)

88

(21

)7

8(2

1)

7.6

,P

=0

.00

6B

ase

In>

Pla

no

,B

ase

Ou

tan

dP

lus,

all

P<

0.0

1;

Bas

eO

ut

>P

lus

P<

0.0

01

Tim

eto

ob

ject

con

tact

(ms)

79

4(8

0)

81

2(8

8)

81

5(9

2)

83

4(9

1)

6.4

,P

=0

.00

1P

lus

>P

lan

oan

dB

ase

InP

<0

.01

and

>B

ase

Ou

tP

=0

.02

Tim

eto

pea

kd

ecel

erat

ion

(ms)

50

5(6

7)

51

0(7

5)

50

3(7

3)

50

2(6

4)

0.6

,P

=0

.6(n

s)

Ter

min

alre

ach

du

rati

on

(ms)

28

8(5

4)

30

1(6

8)

31

1(8

5)

33

2(6

9)

5.4

,P

=0

.00

3P

lus

>P

lan

oan

dB

ase

InP

<0

.01

and

Bas

eO

ut

P=

0.0

47

Gra

spp

aram

eter

s

Pea

kg

rip

aper

ture

(mm

)7

1(7

)7

4(7

)7

3(7

)7

3(7

)5

.2,

P=

0.0

03

Pla

no

<B

ase

In,

Bas

eO

ut

and

Plu

s,al

lP

<0

.01

Gri

pap

ertu

reat

ob

ject

con

tact

(mm

)4

2(3

)4

6(4

)4

5(4

)4

3(2

)5

.6,

P=

0.0

05

Pla

no

<B

ase

InP

=0

.00

1an

d<

Bas

eO

ut

P=

0.0

24

;P

lus

<B

ase

In

P=

0.0

15

Gri

pcl

osu

reti

me

(ms)

21

9(4

9)

22

3(5

7)

22

9(7

3)

24

7(6

2)

5.1

,P

=0

.00

7P

lus

>P

lan

oan

dB

ase

InP

<0

.01

and

>B

ase

Ou

tP

=0

.02

Gri

pap

pli

cati

on

tim

e(m

s)1

40

(25

)1

54

(38

)1

57

(35

)1

52

(35

)3

.6,

P=

0.0

26

Pla

no

<B

ase

Ou

tan

dP

lus

P<

0.0

1an

d<

Bas

eIn

P=

0.0

29

Exp Brain Res (2007) 183:283–298 291

123

Examples of collisions, under-aiming and mis-grasping

movements made on individual Base In, Base Out and Plus

lens trials are shown in Figs. 5 and 6b–d, respectively, with

normal velocity and grip aperture profiles executed under

Plano lens viewing illustrated (Figs. 5a, 6a) for compari-

son. ANOVA demonstrated that the overall occurrence of

the different types of error was significantly affected by

viewing condition (see Fig. 7). The Base In prism resulted

in significantly more (x2-4) collisions (F(3,48) = 8.1 and

P = 0.001) compared to all other conditions (P < 0.01 for

the Plano and Plus lens; P = 0.024 versus Base Out), a

finding already suggested by the kinematics. However,

both the Base Out prism and the Plus lens were associated

with an increased occurrence of short-falls in reaching

distance (F(3,48) = 6.6 and P = 0.001), in under-aims dur-

ing the terminal reach/grip closure phase (F(3,48) = 5.8 and

P = 0.006) and in errors in grip application between object

contact and lifting (F(3,48) = 6.5 and P = 0.001). Post hoc

tests revealed that these increases were generally signifi-

cant in relation to both the Plano lens and Base In prism

(all P < 0.01, except for Plus lens ‘short-falls’ versus Plano

and Base In, P = 0.015 and 0.07, respectively), but not

with respect to each other (all P > 0.25). Thus while the

kinematics showed that reducing disparity information

selectively prolonged the TRD and GCT, both this and the

Base Out condition resulted in more on-line corrections

just before and during the grasp.

Effects of procedure

Planning time affected performance in only two meaning-

ful ways. Participants who moved just after the target was

presented (No Preview) opened their hand with a wider

peak grip (PGA, Procedure · Size interaction, F(1,16) = 6.2

and P = 0.021) across all viewing conditions compared to

those allowed to Preview the task, particularly (by 3–

4 mm) when grasping the smaller object (overall means,

65 mm (±7.0 SD) and 61 mm (±5.9 SD), respectively,

P = 0.008). This was not due to between-group differences

in hand or digit size, since grip apertures were corrected for

this (see Methods). They then generally made fewer cor-

rections during the final approach to the target (Under-

aims, Procedure · View interaction, F(1,3) = 4.8 and

P = 0.013), most notably in the Plus lens condition

(P = 0.01). This effect was opposite to that expected, since

we anticipated that reducing the time allowed to plan the

movement would require these subjects to place more

emphasis on feedback control. In fact, programming an

extra safety margin into their grip at pre-shaping would

have been a sensible precaution, given the short planning

time, and this may have contributed to the relative reduc-

tion in their terminal reaching errors. But why this reduc-

tion was most evident with the Plus lens is hard to explain,

especially as an increased PGA does not notably reduce the

occurrence of these types of error under monocular con-

ditions (e.g. Melmoth and Grant 2006). Perhaps this odd

result is one that occurred by chance.

Discussion

We compared the effects of perturbing vergence or bin-

ocular disparity cues, by optically altering the input to one

eye, on goal-directed reach-to-grasp movements. Pre-

liminary checks established that the two prism orientations

and the Plus lens resulted in different constant errors on an

open-loop pointing test that were in the direction—

although smaller in amplitude—predicted by the optical

distortion. This suggests that distance judgements were

differentially affected by each perturbation at the

Fig. 3 Mean terminal reach duration (from peak deceleration to

object contact) as a function of target location, under each of the four

binocular viewing conditions. C control (Plano) lens, BI Base In

prism, BO Base Out prism P Plus lens. Bars represent the standard

errors

Fig. 4 Mean velocity at object contact as a function of target

location, under each of the four binocular viewing conditions.

Conventions, as in Fig. 3

292 Exp Brain Res (2007) 183:283–298

123

movement programming stage, but with less weighting

attached to the altered binocular view than to alternative

cues to distance that were available pictorially in the other

eye (e.g. height-in-scene, perspective and familiar size) or

extra-retinally (e.g. vertical gaze angle). Despite this, pre-

hension performance was markedly affected by the altered

binocular cues, particularly on parameters related to the

movement end-phases.

Subjects appeared to achieve higher peak reaching

velocities under Base In versus Base Out prism orienta-

tions, with Base In viewing clearly resulting in more hard

collisions with the objects compared to all other conditions.

Such behaviour suggests that Base In prism reaches were

programmed to transport the hand beyond the actual target

location due to its far VSD, and that this ‘further’ move-

ment plan was not fully corrected by in-flight visual

feedback indicating imminent hand-target impact, at least

on a substantial number of trials. The Base Out prism, on

the other hand, caused an increase in extra forward and

under-aiming movements (Fig. 7) following its slightly

lower PV compared to control and Base In viewing,

suggesting that a shorter and slower reach had been pro-

grammed based on the targets’ near VSD. These findings

support the original conclusion of Mon-Williams and Di-

jkerman (1999) that vergence cues to distance are selec-

tively utilized for reach planning. Unlike this earlier study,

however, we observed no predicted effects of prism ori-

entation on the PA (or ttPD) of hand transport. We surmise

that these measures of reach programming were less sen-

sitive than PV to the subtle distance perturbations induced

by the prism in our experiments.

In contrast, grip programming was more uniformly

affected by the three binocular perturbations, with the

maximum aperture at pre-shaping increasing similarly

compared to control viewing. This outcome was supported

by quartiles analysis, indicating that there was not even a

hidden bias in binocularly altered behaviour. Our finding

that the PGA was no larger under Base In compared to

Base Out prism viewing—as would be predicted from

VSD-size constancy—is in agreement with Mon-Williams

and Dijkerman (1999), but was obtained with objects dif-

fering in size in the grasp plane. It thus appears that

Fig. 5 Velocity profiles of movements executed by one subject to the

same target, following a 3 s preview, under each of the four binocular

viewing conditions. The auditory ‘go’ signal occurred at time 0 ms;

the open circles show the point of peak deceleration (PD) in the

reach; the filled circles indicate the moment of initial object contact

(OC), defined as the recording frame in which the marker on the

target was first displaced, in three-dimensions, by � 1 mm. a A

normal movement, with initial OC occurring shortly after PD and just

before the point of minimum terminal reach velocity, followed by a

brief re-acceleration as the target was quickly grasped and lifted. b A

truncated movement in which OC occurred just before the PD and

shows no subsequent low velocity phase, suggesting a premature

collision with the target; c A pre-contact ‘movement unit’ consisting

of an extra acceleration/deceleration (‘peak’) during the terminal

reach before OC, indicative of target under-shooting; d A post-contact

‘movement unit’ consisting of a continuous low-velocity plateau after

OC, suggesting a prolonged period of uncertain grip application prior

to lifting the target

Exp Brain Res (2007) 183:283–298 293

123

conflicting binocular cues can be largely ignored in favour

of more ‘reliable’ monocular information when planning

the grasp, yet simultaneously influence the programming of

hand transport. This accords with evidence that different

object properties—like distance, size and shape—are pro-

cessed relatively independently for visual perception

(Landy et al. 1995; Brenner and van Damme 1999; Bing-

ham 2005), and that distance and size information feed

different channels in the prehension system (Jeannerod

1986; Kudoh et al. 1997; Mon-Williams and Tresilian

1999). In fact, Bradshaw et al. (2004) provide compelling

evidence for dissociation between distance-size cue-

weighting in prehension planning. Their subjects fixated a

mirror (at 37.6 cm) on which they viewed stereo images of

a workspace containing objects of three different virtual

Fig. 6 Grip aperture profiles of movements executed by another

subject to the same target, following a 3 s preview, under each of the

four binocular viewing conditions. Other conventions are as in Fig. 5.

a A normal grasping movement, with the hand opened to a peak grip

aperture close to the object, then closing rapidly to establish initial

object contact (OC) before quickly securing and lifting it; b A very

wide grip at OC with no subsequent closure of thumb and finger,

suggesting a premature collision with the target; c A pre-contact

‘movement unit’ consisting of an extra opening/closing (‘peak’) of

the grip prior to OC, indicating an attempted grasp of a ‘phantom’

target; d A post-contact ‘movement unit’ consisting of an extra

opening/closing (‘peak’) of the grip after initial OC, indicating that

the grasp had to be re-applied in order to secure the target. Note:

starting grip apertures are shown as zero at time 0 ms in these profiles,

simply for the sake of clarity. Note also that the very early ‘peak’

occurring just (*100–200 ms) after movement onset in each trace

results from initial release of the start button and is distinct from the

maximum aperture occurring at pre-shaping later in the movement.

Although its significance remains unclear, it is a product of grip

programming known to be most associated with situations in which

the starting digit positions are wider than the object-to-be-grasped

(Saling et al. 1996; Meulenbroek et al. 2001). Indeed, it was a

common occurrence among our subjects when the smaller object was

the target, but also occurred idiosyncratically for the larger target

(with a diameter slightly greater than the start button), as in the

profiles shown

Fig. 7 Mean number of collisions, reaching short-falls, pre-contact

under-aims and post-contact mis-grasps made under each of the four

binocular viewing conditions. Asterisks denote significant differences

(P < 0.01) compared to Plano lens viewing. Other conventions, as in

Fig. 3

294 Exp Brain Res (2007) 183:283–298

123

distances and sizes, using binocular, monocular or match-

ing bi-ocular vision. In the bi-ocular condition, participants

reached with the same PV regardless of target distance,

suggesting that the vergence angle adopted (which was

fixed in relation to the mirror) completely dominated

(*100% weighting) the conflicting monocular pictorial

cues to distance also present in the workspace layout. Yet

they simultaneously used monocular cues to scale their

PGA appropriately to the targets’ size.

Our major new findings concern the effects of the de-

focusing Plus lens. Compared to normal binocular vision,

this condition led to a (possible) reduction in peak reaching

velocity with a significantly larger PGA, followed by more

prolonged and error-prone terminal reaches and grasps.

That is, reducing disparity cues caused reliable deficits in

both grip planning and movement execution, and in similar

ways to the use of one eye alone (e.g. Watt and Bradshaw

2000; Bradshaw and Elliot 2003; Loftus et al. 2004a;

Melmoth and Grant 2006) and to programmed (i.e. open-

loop) binocular performance in artificial disparity-altered

settings (Hibbard and Bradshaw 2003; Bradshaw et al.

2004). In support of a specific role for disparity in planning

the grasp, our own experience of Plus lens viewing is that it

generated more obvious perceptual uncertainty than did the

prism about the precise 3D-shape and width of the goal

object, and thus in deciding in advance of the movement

where best to position the grip on its surface. It is probable

that our subjects were also aware that information about the

objects’ intrinsic dimensions was unreliable in this condi-

tion, and so modified their PGA to incorporate a greater

margin of error.

Loss of disparity information resulted in selective

extensions of the reaching end-phase and grip closure

times. It has been argued that binocular disparity improves

the on-control of prehension in two distinct ways—via a

‘nulling’ process which signals that the gap between hand

and target is narrowing, and by guiding the closing digits to

the pre-selected grip points on the object that appear

optimal for stable grasping (e.g. Jackson et al. 1997;

Bradshaw et al. 2000, 2004; Watt and Bradshaw 2000,

2003; Bingham et al. 2001; Melmoth and Grant 2006). The

nulling process only requires, in principle, that the observer

can determine that crossed (near) disparities between hand

and target and that uncrossed (far) disparities between

target and hand are reducing concurrently. As such, it could

be mediated by relatively coarse detectors that compute

changes in the sign of disparity and/or stereo motion

towards/away from the observer—such as the near and far

disparity-tuned and directionally opponent cells of the

visual cortex (Zeki 1974; Poggio and Talbot 1981;

Maunsell and Van Essen 1983; Poggio et al. 1988). The

output of these cells is believed to contribute to qualitative

depth estimation, even for unfused binocular images, and

so should have been operating in the defocused condition

in which coarse disparities (>400 arc s) were discernible.

Monitoring the progress of each digit towards their optimal

grasp points, however, is computationally more demanding

and to do this efficiently probably requires access to more

precise depth information, such as provided by the output

of tuned excitatory cells which are tightly selective for fine

disparities close to the fixation plane and for directionally

non-opponent stereo motion roughly parallel to it (e.g.

Poggio and Talbot 1981). This is because the thumb- and

finger-tips move independently to their relatively lateral

destinations on cylindrical objects during precision grip

closure, with the thumb, in particular, probably guided by

visual fixation of its landing site (Wing and Fraser 1983;

Paulignan et al. 1997; Smeets and Brenner 1999; Galea

et al. 2001; Melmoth and Grant 2005).

We suggest that the defocus-induced loss of fine dis-

parity sensitivity compromised this aspect of grip execu-

tion, with subjects having to depend on less reliable depth

cues to control this movement phase. Since grip closure

occurs at the end of the ‘terminal reach’, this would also

explain why the GCT and TRD were similarly prolonged in

this condition (Table 2). But an alternate possibility is that

the problem of grasp-point selection arose solely at the

planning stage as a result of uncertainty about where to

place the initial grip. If so, subjects might be expected to

employ other precautions, such as programming their ini-

tial reach to land well short of the target (to avoid colliding

with it), which could then account for the increased time

subsequently spent in the final approach. Importantly, this

did not seem to be their strategy, since evidence of it (i.e.

‘short-falls’, Fig. 7) only occurred on *5% (1.3/24) of all

Plus lens trials. Yet Base Out prism viewing resulted in a

somewhat higher incidence of short-falls, but had smaller

effects on the TRD and GCT (Table 2). One way of

probing this issue further would be to compare within-

subject performance when fine-grade disparity cues are

available for planning the grip but not during its closure

and in the converse situation. In fact, analogous studies that

alternated participants’ viewing between binocular and

monocular vision for prehension planning versus execution

(Servos and Goodale 1994; Jackson et al. 1997) have

already shown that movement durations just prior to object

contact are prolonged in the absence of simultaneous bin-

ocular feedback despite the presence of disparity infor-

mation at the programming stage, but are reduced when

binocular disparity cues are only available in the immediate

pre-contact period (see also Bradshaw and Elliot 2003).

Thus the available evidence supports a role for disparity in

both grip planning and on-line control.

Problems with grasp-point selection that occur with

monocular vision are also associated with increased grip

application times and mis-grasping errors (Melmoth and

Exp Brain Res (2007) 183:283–298 295

123

Grant 2006), so we expected that these aspects of grasping

would also be selectively affected by the Plus lens. Both

parameters were significantly increased compared to nor-

mal viewing, and occurred even though OC was estab-

lished with the same terminal velocity and grip size (VOC

and GOC) in the two conditions (Table 2). This further

suggests that, despite these appearances of ‘normality’, the

original digit positions were unsuited to stable grasping in

the reduced disparity condition. But the Base Out prism

also resulted in increases in the GAT and mis-grasping

errors. It is notable that this prism orientation was associ-

ated with generally ‘harder’ OCs (e.g. increased VOC and

GOC) versus the Plano or Plus lens. It is, therefore, pos-

sible that the need to re-programme the movement close to

the target with Base Out prism viewing, having unex-

pectedly under-shot it, lead to subsequent mistakes at and

immediately after contact.

Vergence and disparity are normally linked in ‘real

world’ settings, as mimicked in our prehension experi-

ments, so it is important to consider the extent to which we

were able to dissociate the two. In screening the subjects,

we found that wearing the prism in either orientation did

not affect their horizontal disparity sensitivity (i.e. ste-

reoacuity thresholds). Altering the vergence signal should,

however, influence disparity scaling. This is because the

angle of disparity generated by two points separated by

the same physical depth in 3D-space is largely a function of

the square of the fixation distance. As a consequence,

identical points on an objects’ surface should appear to be

elongated in depth under divergent conditions (far fixation,

Base In) and as contracted in depth with convergence (near

fixation, Base Out). Mon-Williams et al. (2000) have pro-

vided empirical support for such distortions in 3D-shape

perception: their subjects reported that the same object

(a wire pyramid) appeared both further away and as

extended in depth when wearing Base In prisms compared

to normal vision. Our subjects initially contacted the object

with a wider grip when viewing through the Base In (mean

45.9 mm) versus the Base Out (mean 44.6 mm) prism.

While this prism-orientation effect did not approach sig-

nificance (P > 0.25), the increased GOC with excess

divergence was reliably greater than for the Plano and

(possibly) the Plus lenses (Table 2). We interpret this

larger grip at contact as being yet another consequence of

the higher incidence of premature collisions with Base In

viewing. But it is possible that this also caused the objects

to appear stretched in depth, and so the subjects attempted

initial contact with more widely separated grip positions to

match.

Further ambiguities relate to the Plus lens, since its

effect in magnifying the retinal image alters vertical dis-

parities, as well as attenuating horizontal disparity sensi-

tivity via image blur. One effect of introducing such an

inter-ocular size difference is that objects should appear

closer to the affected eye. Pointing responses confirmed

this, although they indicated that effect received a very

low weighting. However, vertical disparity processing

may also provide information about fixation distance and

so, like vergence, be used to calibrate horizontal disparity

cues to object size and depth (Bishop 1989; Rogers and

Bradshaw 1993). It is thus conceivable that during Plus

lens prehension, subjects fixated a ‘virtual’ target located

just in front of the real object (see Fig. 2) whose apparent

3D-properties were re-scaled accordingly (i.e. slightly

narrower width and shallower depth). This is unlikely to

have occurred, since the programmed PGA would be

expected to be similarly reduced in width, not increased,

as we observed. Future studies comparing the effects on

binocular prehension of diffusing foils (which reduce

stereoacuity, without changing image size) and anisei-

konic lenses (which produce a size change, with little

effect on stereoacuity) may help to definitively resolve

this issue.

In summary, our data suggest that vergence contributes

mainly to reach planning (cf., Mon-Williams and Dijker-

man 1999; Bradshaw et al. 2004), whereas disparity cues

are required for proficient control of the grasp. This con-

clusion accords with evidence of neurological dissociations

between regions of the posterior parietal cortex in medi-

ating these different aspects of prehension. Specifically,

reach-related areas of the superior parietal lobule contain

cells whose activity is strongly influenced by extra-retinal

signals linked to gaze position and eye movements, while

neurons located in grasp-related areas along the intra-

parietal sulcus show specializations for binocular disparity,

3D-shape and -surface orientation (reviewed in Caminiti

et al. 1998; Sakata et al. 1999; Culham and Kanwisher

2001). In this context, the ability to process fine disparities

seems to make an indispensable contribution to grip for-

mation, since adults with long-term stereoacuity losses

exhibit similar grasping deficits (Grant et al. 2007) to those

found here following its temporary reduction.

Acknowledgements Funded by The Wellcome Trust (Grant

066282). We thank Prof. Michael Morgan for comments on the work.

References

Berthier NE, Clifton RK, Gullapalli V, McCall DD, Robin DJ (1996)

Visual information and object size in the control of reaching.

J Mot Behav 28:187–197

Blakemore C, Hague B (1972) Evidence for disparity detecting

neurons in the human visual system. J Physiol (Lond) 225:437–

455

Bingham GP (2005) Calibration of distance and size does not

calibrate shape information: comparison of dynamic monocular

and static and dynamic binocular vision. Ecol Psychol 17:55–74

296 Exp Brain Res (2007) 183:283–298

123

Bingham GP, Bradley A, Bailey M, Vinner R (2001) Accommoda-

tion, occlusion, and disparity matching are used to guide

reaching: a comparison of actual versus virtual environments.

J Exp Psychol Hum Percept Perform 27:1314–1334

Bishop PO (1989) Vertical disparity, egocentric distance and

stereoscopic depth constancy: a new interpretation. Proc R Soc

London Biol 237:445–469

Bradshaw MF, Elliot KM (2003) The role of binocular information in

the ‘on-line’ control of prehension. Spat Vis 16:295–309

Bradshaw MF, Parton AD, Glennerster A (2000) The task-dependent

use of binocular disparity and motion parallax information.

Vision Res 40:3725–3734

Bradshaw MF, Elliot KM, Watt SJ, Hibbard PB, Davies IR, Simpson

PJ (2004) Binocular cues and the control of prehension. Spat Vis

17:95–110

Brenner E, van Damme WJM (1999) Perceived distance, shape and

size. Vision Res 39:975–986

Caminiti R, Ferraina S, Mayer AB (1998) Visuomotor transforma-

tions: early cortical mechanisms of reaching. Curr Opin Neuro-

biol 8:753–761

Churchill A, Hopkins B, Ronnqvist L, Vogt S (2000) Vision of the

hand and environmental context in human prehension. Exp Brain

Res 134:81–89

Culham JC, Kanwisher NG (2001) Neuroimaging of cognitive

functions in human parietal cortex. Curr Opin Neurobiol

11:157–163

Felton TB, Richards W, Smith RA Jr (1972) Disparity processing of

spatial frequencies in man. J Physiol (Lond) 225:349–362

Galea MP, Castiello U, Dalwood N (2001) Thumb invariance during

prehension movement: effects of object orientation. Neuroreport

12:2185–2187

Glover S (2003) Optic ataxia as a deficit specific to the on-line control

of actions. Neurosci Behav Rev 27:447–456

Goodwin RT, Romano PE (1985) Stereoacuity degradation by

experimental and real monocular and binocular amblyopia.

Invest Ophthalmol Vis Sci 26:917–923

Grant S, Melmoth DR, Morgan MJ, Finlay AL (2007) Prehension

deficits in amblyopia. Invest Ophthalmol Vis Sci 48:1139–1148

Hibbard PB, Bradshaw MF (2003) Reaching for virtual objects:

binocular disparity and the control of prehension. Exp Brain Res

148:196–201

Jackson SR, Jones CA, Newport R, Pritchard C (1997) A kinematic

analysis of goal-directed prehension movements executed under

binocular, monocular, and memory-guided viewing conditions.

Vis Cog 4:113–142

Jackson SR, Newport R, Shaw A (2002) Monocular vision leads to

dissociation between grip force and grip aperture scaling during

reach-to-grasp movements. Curr Biol 12:237–240

Jakobson LS, Goodale MA (1989) Trajectories of reaches to

prismatically-displaced targets: evidence for ‘automatic’ visuo-

motor recalibration. Exp Brain Res 78:575–587

Jakobson LS, Goodale MA (1991) Factors affecting higher-order

movement planning: a kinematic analysis of human prehension.

Exp Brain Res 86:199–208

Jeannerod M (1986) Mechanisms of visuomotor coordination: a study

in normal and brain-damaged subjects. Neuropsychologica

24:41–78

Jones RK, Lee DN (1981) Why two eyes are better than one: the two

views of binocular vision. J Exp Psychol Human Percept

Perform 7:30–40

Kudoh N, Hattori M, Numata N, Maruyama K (1997) An analysis of

spatiotemporal variability during prehension movements: effects

of object size and distance. Exp Brain Res 117:457–464

Landy MS, Maloney LT, Johnston EB, Young M (1995) Measure-

ment and modeling of depth cue combination: in defense of

weak fusion. Vision Res 35:389–412

Levy NS, Glick EB (1974) Stereoscopic perception and Snellen

acuity. Am J Ophthalmol 78:722–724

Loftus A, Servos P, Goodale MA, Mendarozqueta N, Mon-Williams

M (2004a) When two eyes are better than one in prehension:

monocular viewing and end-point variance. Exp Brain Res

158:317–327

Loftus A, Murphy S, McKenna I, Mon-Williams M (2004b) Reduced

fields of view are neither necessary nor sufficient for distance

underestimation but reduce precision and may cause calibration

problems. Exp Brain Res 158:328–335

Maunsell JHR, Van Essen DC (1983) Functional properties of

neurons in middle temporal visual area of the macaque monkey.

II Binocular interactions and sensitivity to binocular disparity.

J Neurophysiol 49:1148–1167

Melmoth DR, Grant S (2005) Vision of the thumb as the guide to

prehension. Perception 34:Supplement 243

Melmoth DR, Grant S (2006) Advantages of binocular vision for

the control of reaching and grasping. Exp Brain Res 171:371–

388

Meulenbroek RGJ, Rosenbaum DA, Jansen C, Vaughan J, Vogt S

(2001) Multijoint grasping movements. Simulated and observed

effects of object location, object size, and initial aperture. Exp

Brain Res 138:219–234

Mon-Williams M, Dijkerman HC (1999) The use of vergence

information in the programming of prehension. Exp Brain Res

128:578–582

Mon-Williams M, Tresilian JR (1999) The size-distance paradox is a

cognitive phenomenon. Exp Brain Res 126:578–582

Mon-Williams M, Tresilian JR, Roberts A (2000) Vergence provides

veridical depth perception from horizontal retinal image dispar-

ities. Exp Brain Res 133:407–413

Paulignan Y, Frak VG, Toni I, Jeannerod M (1997) Influence of

object position and size on human prehension movements. Exp

Brain Res 114:226–234

Poggio GF, Talbot WH (1981) Mechanisms of static and dynamic

stereopsis in foveal cortex of the rhesus monkey. J Physiol

(Lond) 315:469–492

Poggio GF, Gonzalez F, Krause F (1988) Stereoscopic mechanisms in

monkey visual cortex: binocular correlation and disparity

selectivity. J Neurosci 8:4531–4550

Rabbetts RB (1998) Bennett & Rabbetts’ clinical visual optics, 3rd

edn. Butterworth Heinemann, London

Rogers B, Bradshaw MF (1993) Vertical disparities, differential

perspective and binocular stereopsis. Nature 361:253–255

Sakata H, Taira M, Kusunoki M, Murata A, Tsutsui K, Tanaka Y,

Shein WN, Miyashita Y (1999) Neural representation of three-

dimensional features of manipulation objects with stereopsis.

Exp Brain Res 128:160–169

Saling M, Mescheriakov S, Molokanova E, Stelmach GE, Berger M

(1996) Grip reorganization during wrist transport: the influence

of an altered aperture. Exp Brain Res 108:493–500

Schor CM, Wood IC, Ogawa J (1984) Spatial tuning of static and

dynamic local stereopsis. Vision Res 24:573–578

Servos P, Goodale MA (1994) Binocular vision and the on-line

control of human prehension. Exp Brain Res 98:119–127

Servos P, Goodale MA, Jakobson LS (1992) The role of binocular

vision in prehension: a kinematic analysis. Vision Res 32:1513–

1521

Smeets JBJ, Brenner E (1999) A new view on grasping. Motor

Control 3:237–271

Tresilian JR, Mon-Williams M, Kelly BM (1999) Increasing confi-

dence in vergence as a distance cue. Proc R Soc Lond Biol Sci

266:39–44

Watt SJ, Bradshaw MF (2000) Binocular cues are important in

controlling the grasp but not the reach in natural prehension

movements. Neuropsychologica 38:1473–1481

Exp Brain Res (2007) 183:283–298 297

123

Watt SJ, Bradshaw MF (2003) The visual control of reaching and

grasping: binocular disparity and motion parallax. J Exp Psychol

Hum Percept Perform 29:404–415

Wing AM, Fraser C (1983) The contribution of the thumb to reaching

movements. Q J Exp Psychol 35A:297–309

Wood ICJ (1983) Stereopsis with spatially-degraded images. Ophthal

Physiol Opt 3:337–340

Yang Y, Blake R (1991) Spatial frequency tuning of human

stereopsis. Vision Res 31:1177–1183

Zeki SM (1974) Cells responding to changing image size and

disparity in the cortex of the rhesus monkey. J Physiol (Lond)

236:549–573

298 Exp Brain Res (2007) 183:283–298

123