Neuropsychologia 43 (2005) 227–237

Motor awareness without perceptual awareness�

Helen Johnson, Patrick Haggard∗

Institute of Cognitive Neuroscience and Department of Psychology, University College London,Alexandra House, 17 Queen Square, London WC1N 3AR, UK

Abstract

The control of action has traditionally been described as “automatic”. In particular, movement control may occur without consciousawareness, in contrast to normal visual perception. Studies on rapid visuomotor adjustment of reaching movements following a target shift haveplayed a large part in introducing such distinctions. We suggest that previous studies of the relation between motor performance and perceptualawareness have confounded two separate dissociations. These are: (a) the distinction between motoric and perceptual representations, and(b) an orthogonal distinction between conscious and unconscious processes. To articulate these differences more clearly, we propose a newmeasure ofmotorawareness, based on subjects’ ability to reproduce the spatial details of reaching movements they have just made. Herewe focus on the dissociation between motor awareness and perceptual awareness that may occur when subjects make rapid visuomotoradjustments to reaching movements following a target shift. In experiment 1, motor awareness was dissociated from perceptual awareness of at dependento nor motorp lications fort©

K

1

abfirsotqoc

sa

ean-g the

ach-an-pentss-

l.s to-eir

andd thent wasr to

sualtless,jectso theems

0d

arget shift during reaching movement. Participants’ reproduction of movement endpoints following visuomotor adjustment was inf whether they saw the target shift or not. Experiment 2 replicated this result, and further showed that neither motor awarenesserformance were disrupted by TMS over the parietal cortex. The neural mechanisms underlying motor awareness, and the imp

heories of consciousness, are discussed.2004 Elsevier Ltd. All rights reserved.

eywords:Human; Reaching movement; Unconscious processing; Consciousness; Transcranial magnetic stimulation

. Introduction

Studies of the neural mechanisms underlying coordinatedction have traditionally occupied a curious middle groundetween psychology and neuroscience. This has given theeld a very productive, interdisciplinary flavour, and allowedesearchers to raise strictly scientific questions, which at theame time seem of fundamental importance to the wider viewf the mind and human nature. This combination of scien-

ific knowledge and a willingness to tackle major intellectualuestions in a clear and practical way is very characteristicf the work of Marc Jeannerod, throughout his distinguishedareer.

In this paper, we particularly focus on the relation betweenkilled action and conscious experience. Marc Jeannerod wasmong the first to investigate this relation scientifically. The

� This study was carried out at the Institute of Cognitive Neuroscience.∗ Corresponding author. Tel.: +44 20 7679 1151, fax: +44 20 7813 2835.E-mail address:p.haggard@ucl.ac.uk (P. Haggard).

work was continued over several years by Prablanc, Jnerod and others, in an extensive series of studies usin“double-step paradigm” of visuomotor adjustments in reing movements (e.g.Pelisson, Prablanc, Goodale, & Jenerod, 1986; Prablanc & Martin, 1992). The double-steparadigm was initially studied in saccadic eye movem(Hallet & Lightstone, 1976) but we focus here on the clasic result in the case of reaching movements.Pelisson et a(1986)asked subjects to make rapid pointing movementwards a visual target, without vision of the hand. In thparadigm, subjects initially fixated a peripheral light,made a saccade to the target when it first appeared, anreached towards it (step 1). During the saccade, the targeunpredictably shifted a small distance either to the left othe right (step 2). Because of the known reduction in viperception during a saccade (Matin, 1974), subjects were noperceptually aware that the target had shifted. Nevertheinspection of the movement trajectories showed that subwere able to correct their ongoing trajectories according tdirection of the visual target shift. That is, the brain syst

028-3932/$ – see front matter © 2004 Elsevier Ltd. All rights reserved.oi:10.1016/j.neuropsychologia.2004.11.009

228 H. Johnson, P. Haggard / Neuropsychologia 43 (2005) 227–237

generating motor awareness clearly had access to informa-tion about the target location that did not enter into perceptualawareness.

Since this early work, several additional studies byMarc Jeannerod and others have replicated the basic result(Castiello, Paulignan, & Jeannerod, 1991; Day & Brown,2001; Day & Lyon, 2000; Desmurget et al., 1999; Paulignan,MacKenzie, Marteniuk, & Jeannerod, 1991; Prablanc & Mar-tin, 1992). This work became widely known and had strongimpact on our way of thinking about the mind and brain ingeneral. For example, the basic finding drove the develop-ment of the distinctions between the dorsal and ventral visualpathways (Goodale & Milner, 1992), and between the brainsystems for semantics and pragmatics (Jeannerod, 1994).

Many studies using the double-step paradigm have trig-gered the target shift directly from the saccade, in order toprevent perceptual awareness of the target shift. Relativelyfew have measured perceptual awareness directly, and evenfewer have investigated what factors influence perceptualawareness. Therefore, the initial claim of dissociation be-tween motor performance and perceptual awareness was notexamined in detail. An important exception, however, was thestudy byCastiello et al. (1991). They used the basic double-step paradigm, as before. Unusually, they did not attemptto prevent vision of the target shift. Instead, they requireds s ine thed t shiftb condc sim-p finalc eacht onser eas av

urredw er-b wasp e tria( s).T onsea me-d con-s roma ses.T fora ptuala

haves o notc are-n hich if nced mo-t is ap ness,

or subjective state. Dissociating processes which differ inmore than one respect is problematic: which dimension ofdifference is responsible for the differences seen in experi-mental data? Moreover, processes can only be dissociated ifthey were previously accepted as unitary. As the number ofattributes along which the processes are accepted to differincreases, an approach based on dissociation becomes lessconvincing.

We therefore suggest that the scientific issues raised by thedouble-step paradigm can be broken down into investigationsof two separate dissociations. First, does motor performancedissociate from motor awareness? That is, what do peopleconsciously know about their own visuomotor adjustments?Second, does motor awareness dissociate from perceptualawareness? That is, can one be aware of a visuomotor ad-justment even when unaware of any target shift? We recentlygave an affirmative answer to the first question (Johnson, vanBeers, & Haggard, 2002). By asking subjects to reproducethe spatial path of a movement they had just made in thedouble-step paradigm, we found that subjects typically un-derestimated the latency and the gain of their own visuomotoradjustments. In an anti-point condition, however, both theseresults were reversed. This pattern of results suggested thatthe awareness subjects had of their visuomotor adjustmentdepended strongly on the role of intention in generating thea Weu ects’c y asa ” int

pro-p ptuala ay orm rial,a sac-c e full,l iatedt ov se thes y de-p if thes reachf ouldb ct be-c to ones iatedi eral,a witha r, thep otora atialw atureoa ciablea brains

ubjects to respond to the target shift in different wayach of three conditions. The first condition replicatedouble-step paradigm: subjects responded to the targey adjusting an ongoing reaching movement. In the seondition, subjects did not move their hand, but made ale vocal response on detecting the target shift. In theondition, subjects made both of these responses onrial. Castiello et al. argued that the simple verbal respequired perceptual awareness of the target shift, wherisuomotor adjustment did not.

The results showed that visuomotor adjustments occithin approximately 100 ms of the target shift, while val responses took over 300 ms. This latency differenceresent whether the responses were made on the samthird condition) or individually (first and second conditionhe authors interpreted the delay in the verbal resps reflecting the slow operation of the brain circuitsiating perceptual awareness. The circuits underlyingcious processing are either additional to, or distinct fnd slower than, circuits underlying visuomotor responhis result provides convergent behavioural evidencedissociation between motor performance and perce

wareness.While such studies have been very influential, and

hed new light on the brain’s visuomotor systems, they donclusively reveal the relations between action and awess. These studies appear to dissociate processes, w

act differ in at least two respects. That is, motor performaiffers from perceptual awareness in that the former is

or and the latter perceptual, but also in that the formererformance or behaviour, while the latter is an aware

l

n

djustment. Motor awareness followed motor intention.se the term “motor awareness” here to indicate subjonscious representation of their own action, effectiveln opposite to the traditional use of the term “automatic

he psychology of action.Here, we consider the second of the two dissociations

osed above: how is motor awareness related to percewareness? In the double-step reaching task, subjects may not be aware of the target shift on any particular tccording to how successfully the paradigm implementsadic suppression. At the same time, subjects may havimited, or no awareness that their movement has devo one side (Johnson et al., 2002). In principle, these twarieties of awareness should be interdependent becaupatial path of a movement towards the target physicallends on the spatial location of the target. For example,ubject sees the target jump to one side, and is trying toor that target, they should correct their movement, and she aware that they have done so. Conversely, if the subjeomes aware that their movement trajectory has veeredide, it seems logical to conclude that the trajectory devn pursuit of a target, which had shifted to that side. In genconsciously perceived target shift should be associatedwareness of any resulting motor adjustment. Howevesychological modules responsible for perceptual and mwareness may not follow the strict dependency of the sporld: indeed dissociative disorders are an established fef human consciousness (Farah & Feinberg, 2000). Motorwareness and perceptual awareness could be a dissottributes of the human mind, subserved by separatetructures.

H. Johnson, P. Haggard / Neuropsychologia 43 (2005) 227–237 229

Therefore, we performed two experiments to study therelationship between perceptual awareness and motor aware-ness in further detail. In experiment 1, we combined saccadicsuppression with a double-step reaching task. We asked sub-jects to reproduce each movement as a measure of motorawareness, and we considered how motor awareness var-ied between trials where subjects had reported perceptualawareness of the target shift and those where they had not.In experiment 2, we attempted to disrupt visuomotor ad-justments using TMS over the parietal cortex in a similarparadigm.

2. Experiment 1

2.1. Methods

The details of the method and apparatus were based on pre-vious studies (Desmurget et al., 1999; Pelisson et al., 1986).Fifteen right-handed subjects (10 female, mean age 26 years)participated with informed consent. Subjects sat at a table ina darkened room with their head on a chin rest. The right handwas occluded. Visual targets (7 mm diameter) were projectedonto a horizontal plane above the table, which the subjectviewed via a mirror, so that they appeared to be in front oft in thep woo argeti ed’iA efto ed at1 ndexfi pointa

peri-m rgeti 50t thec trials,i cts.

Each trial had three phases. In the first, reaching phase,the subject initially fixated the peripheral spot for 2–3 s. Thefixation spot disappeared and one of the three targets im-mediately appeared. Subjects looked at, and then reachedfor the illuminated target as quickly as possible. During thesaccade period (200–300 ms after target onset) the centraltarget might jump unpredictably by 4 cm to the left or right(Fig. 1). At the end of the movement, the target was extin-guished and subjects returned to the starting position. Thisinitiated the second, reproduction phase of each trial. A 2 mmwide guideline was shown across the entire table from left toright, at the level of the targets. Subjects were asked to re-peat the spatial detail of the movement path that they hadjust made. They were specifically instructed to reproducethe spatial features, without regard to movement speed orother temporal information. During the movement reproduc-tion phase, a guideline at the Y location of the targets wasprojected to indicate the appropriate movement amplitude,but no targets were visible. At the end of the reproducedmovement, the guideline disappeared and subjects again re-turned to the starting point. Third and finally, subjects wereasked to verbally report any target jump (saying ‘left’, ‘right’or ‘no’) seen during the reaching phase of the trial, as a mea-sure of perceptual awareness. The full sequence of events isshown inFig. 2.

3

3

ift on3

i-n rialsw ereo .8%o als)o m-pt own

e 7 mm

he hand. The central (unperturbed) target was alignedarasagittal plane 14 cm in front of the start position. Tther targets were at 8 cm to either side of the central t

n the frontoparallel plane. When the central target ‘jumpt moved 4 cm horizontally in either direction (seeFig. 1).n initial visual fixation point was shown 40 cm to the lf the central target. Reaching movements were record20 Hz with an electromagnetic tracker mounted on the ingernail. Movement start and end were defined as thet which the velocity of the finger fell below 5 cm/s.

Subjects performed 10 practice trials, then 240 exental trials in blocks of 20. In 60 trials, the central ta

lluminated; in 50 trials, the left target illuminated; and inrials, the right target illuminated. In a further 40 trials,entre target appeared and then jumped left, and in 40t jumped right. Trial order was randomised across subje

Fig. 1. Experimental set-up (note: targets wer

. Experiment 1

.1. Results

Subjects reported perceptual awareness of a target sh7% of jump trials.

Only trials in which the central target was initially illumated were analysed kinematically. Of these 6.34% of tere omitted due to equipment failure. A 2.11% of trials wmitted due to pre-emptive eye or hand movements. A 7f trials were omitted as false positives (in non-jump trir movement in the wrong direction (in jump trials). Exale traces for one participant are shown inFig. 3. The initial

rial and the subsequent trajectory reproduction are sh

in diameter, but are shown enlarged here for clarity).

230 H. Johnson, P. Haggard / Neuropsychologia 43 (2005) 227–237

Fig. 2. Events occurring in each trial.

for one trial on which the subject reported seeing the targetshift, and a second trial on which they did not. They compo-nent of the trajectories shows the smooth reaching movementseen in earlier studies. The movement duration was typicallyaround 600 ms. Thexcomponent shows a clear movement ad-justment towards the target, which occurs during the reach-ing movement, and without pause or reprogramming. Thisbegins around 150 ms after movement onset. Note that theadjustment is similar in trials with and without perceptualawareness of the target shift.

We initially hoped to compare the full spatial paths ofinitial and reproduced movements, in order to investigatewhether motor awareness of visuomotor adjustments. How-ever, inspection of the results revealed a strong rightward biasin the initial reaching movements. This was substantially re-duced in the reproduced movements. The difference in biastended to obscure the small lateral deviations, which mark theonset of the visuomotor adjustment. Therefore, we preferred

to study the final hand positions, by which point the visuo-motor adjustment was easily detectable despite the bias. Wediscuss this bias in more detail below.

Our main analysis therefore focussed on the final positionof the hand in thex direction. In the reaching phase of thetrial, this provides a measure of visuomotor adjustment fol-lowing a target jump. Trials in which the target jumped weredivided into those in which the subject detected the jump andthose in which they did not (unaware), and these were anal-ysed separately. Endpoints of all jump trials can be seen inTable 1.

3.2. Endpoints of initial reaching movements

We first analysed the initial movements to detect whethervisuomotor adjustments to the target jump occurred. Themean endpoint of each perturbed condition was comparedto the mean endpoint of the equivalent unperturbed condi-

Table 1Mean and S.D. across subjects of endpoint× deviation for initial and reproduced movements (mm)

Endpoint deviations Left (target =−40 mm) Centre (target = 0 mm) Right (target = 40 mm)

Aware Unaware n/a Aware Unaware

Initial movementMean (mm) −9 −11 25 57 63

R

N s are m

S.D. (mm) 29 30

eproduced movementMean (mm) −28 −32S.D. (mm) 32 29

ote: Negative numbers indication mm left of centre, positive number

30 36 34

14 55 5829 31 33

m right of centre.

H. Johnson, P. Haggard / Neuropsychologia 43 (2005) 227–237 231

Fig. 3. Typical perturbation (“jump”) trials for one participant showing the trajectories for their initial and subsequently reproduced (“copied”) movements.The upper row shows a trial after which the subject reported seeing the target shift. In the trial in the lower row, the subject reported no change in target position.

tion. Family wise Bonferroni corrections were performed,and so the significance level for the followingt-tests is 0.025.All movement endpoints differed significantly from the un-perturbed trials regardless of whether the subject was aware(left: t14 = 11.43, P> 0.001, right: t14 =−8.07, P> 0.001)or unaware (left:t14 = 16.75,P> 0.001, right:t14 =−21.41,P> 0.001) of the target shift. The effect of awareness on ad-justment amplitude was then considered. No significant dif-ference was found between the endpoints of aware and un-aware trials for jumps to either the left (t14 = 1.54,P= 0.15)or the right (t14 =−0.58,P= 0.57). That is, subjects werejust as efficient in adjusting their movements following tar-get jumps regardless of whether they were aware of the jumpor not: motor performance was independent of perceptualawareness. The average endpoints to targets that jumped tothe left were 10 mm to the left of centre. Average endpointsto targets that jumped to the right were 60 mm to the right ofcentre. Thus, there was an overall rightward bias (see above),

but also clear evidence of a normal visuomotor adjustmentpattern in the initial movements.

3.3. Endpoints of reproduced movements

We then investigated whether subjects were aware of theirvisuomotor adjustments by repeating the same comparisonsfor the reproduced movements. The reproduced endpoints ofeach perturbed condition were compared to the reproducedendpoints in the unperturbed condition. Following familywise Bonferroni corrections, the significance level for thefollowing t-tests is again 0.025. All reproduced endpointson perturbed trials differed significantly from those on un-perturbed trials regardless of whether the subject was aware(left: t14 = 10.06,P> 0.001, right:t14 =−11.36,P> 0.001)or unaware (left:t14 = 17.26,P> 0.001, right:t14 =−15.78,P> 0.001) of the target shift. The effect of awareness onthe extent of the reproduced adjustments was then consid-

232 H. Johnson, P. Haggard / Neuropsychologia 43 (2005) 227–237

ered. No significant difference was found between the repro-duced endpoints of aware and unaware trials for jumps toeither the left (t14 =−1.20,P= 0.25) or the right (t14 = 1.28,P= 0.22).

The relationship between initial endpoints and reproducedmovements in jump trials gives a measure of motor awareness(seeTable 1). Inspection of the table shows that the subjectsaccurately reproduced the direction of their visuomotor ad-justments in the immediately preceding movement.

To investigate the relation between perceptual awarenessand motor awareness, a 2× 2× 2 ANOVA was conducted onthe perturbation trial data (seeFig. 4). The factors were direc-tion of target jump (left, right), perceptual awareness (aware,unaware) and phase of trial (initial movement, reproducedmovement).

There was a significant effect of side (F1,14= 265.96,P> 0.001). On average, endpoints to leftward target shiftswere 20 mm left of centre, and 59 mm right of centre forrightward shifts. There was no significant effect of aware-ness (F1,14= 0.35,P= 0.56) There was a significant effect oftrial phase (F1,14= 22.49,P> 0.001). This occurred becausethe rightward bias previously mentioned in the initial move-ments was reduced in the reproduced movements. A signif-icant interaction between side and movement (F1,14= 24.1,P> 0.001) indicated that this reduction was most marked forl tionbnt d trialp tor wase ss oft

4

4

4the

t ean

endpoint of the central trials. That is, subjects performedfunctional visuomotor adjustments following the target shift.However, no significant difference was found between theendpoints for those trials in which the subject was aware ofthe target shift and those in which they were not. This con-firms previous findings that efficient adjustments can be madeto an ongoing movement in response to a target shift regard-less of whether the subject is aware of that target shift or not(Day & Lyon, 2000; Desmurget et al., 1999; Paulignan et al.,1991; Pelisson et al., 1986). The above studies have generallystudied manual responses to targets that shift during saccadicsuppression of conscious experience. Therefore, the dissoci-ation between manual performance and visual awareness thatthey report correlates with a difference between effectors: thehand and arm in one case, and the oculomotor system in theother. However, the dissociation between performance andawareness can be found even in a single effector.Gaveau et al.(2003)recently reported that subjects adjusted saccade kine-matics for small changes in visual target position of whichthey were perceptually unaware.

Returning to the present experiment, initial movements inall conditions were subject to a rightward bias of approxi-mately 2 cm. Such biases have also seen in other similar ex-periments (Desmurget et al., 1999). We believe it may reflectbiomechanical factors (Carey et al., 1996; Carey & Otto deH ndp uringt ed tor

4Sub-

j just-m inghe imbt ent.T ouldr tialm trulyl thep s thep . Ourm ac-c ift orn nessc ite re-p e-l iationi mo-t ciousk n ofr ew, ins terald con-c tions

eftward target jumps. There was no significant interacetween trial phase and awareness (F1,14= 2.92,P= 0.11),or awareness and side (F1,14= 1.91,P= 0.19). Critically,

here was no interaction between side, awareness anhase (F1,14= 0.015,P= 0.91). That is, subjects’ abilityeproduce the spatial detail of visuomotor adjustmentsquivalent in trials with and without perceptual awarene

he target jump.

. Experiment 1

.1. Discussion

.1.1. End points of initial movementsThe mean endpoint of the movements during which

arget jumped were significantly different from the m

Fig. 4. Mean endpoints in perturbation (“jump”) trials (mm).

art, 2001), coupled with the lack of visual feedback of haosition. We speculate that the visual guide presented d

he reproduction phase of the movement may have helpeduce this bias.

.1.2. Reproduced movements and motor awarenessThe most important results concern motor awareness.

ects reproduced the spatial details of the visuomotor adent on the immediately preceding trial. A similar findad been reported previously (Johnson et al., 2002). How-ver, in that earlier study, subjects had full vision of the lhroughout both the initial and the reproduced movemherefore, the reproduced movements in that study ceflect a visual memory of limb positions during the iniovement, as well as, or instead of a representation

inked to movement. Moreover, that study focussed onerceived latency of visuomotor adjustments, wherearesent analysis of endpoints focuses on perceptual gaineasures of motor awareness did not differ significantly

ording to whether subjects were aware of the target shot. This demonstrates the novel finding that motor awarean dissociate from perceptual awareness. That is, desporting that the target didnot shift, subjects were neverth

ess aware that their movement contained a lateral devn the appropriate direction. They are able to report theiror awareness of their movement despite having no consnowledge of the target event that drove it. This patteresults seems quite paradoxical. The subjects clearly knome sense, that their initial movements contained a laeviation. It is tempting to speculate on how subjects reile this awareness of their own motor adjustment devia

H. Johnson, P. Haggard / Neuropsychologia 43 (2005) 227–237 233

with manifest unawareness of any reason why the movementrequired adjusting. Subjects did not seem at all troubled bythis incompatibility, and did not seem to find the motor ad-justments phenomenologically remarkable in the way thatnon-intentional movements evoked by transcranial magneticstimulation, or by Kohnstamm’s manoeuvre often are.

5. Experiment 2

5.1. Introduction

Our initial aim in this experiment was to try to dissociatemotor performance from motor awareness. It is widely ac-knowledged that much, even most, motor function does notrequire conscious thought (Haggard, 2001). The term “auto-matic” is often used in such cases. Therefore, motor perfor-mance is often dissociated from motor awareness in everydaylife. However, to understandwhymotor performance some-times does enter awareness and sometimes does not, an inter-ventive experimental approach is required, in which the ex-perimenter can manipulate whether a given feature of move-ment either does or does not enter awareness. One previousattempt at this approach was the work ofHaggard and Magno(1999). They used the well-established delay in reaction timesc imu-l torp lyj thodo l,1 wareo est-i ancea

-m ishedb ver,s rietalce ess.W isuo-

motor adjustments in our first experiment might not arise inthe parietal structures that generate the adjustment itself, butin a separate cortical pathway. This reasoning suggested theintriguing possibility that parietal TMS might suppress thevisuomotor adjustment itself, while leaving the subject withthe (erroneous) conscious awareness that they had adjustedtheir movement. Experiment 2 sets out to investigate this pos-sibility.

6. Experiment 2

6.1. Method

The method for experiment 2 was identical to that of ex-periment 1, which in turn was partly based on the study ofDesmurget et al. (1999). Subjects again made reaching move-ments to targets that were unpredictably perturbed. In thisexperiment, however, TMS was delivered on 50% of trialsselected at random. On those trials, TMS was delivered at thestart of movement when the hand velocity exceeded 5 cm/s.A focal TMS coil was held by the experimenter over theleft PPC. The exact position was localised according to themethod ofDesmurget et al. (1999); 4.5 cm caudal and 0.5 cmmedial from the hand motor area. Output was set at 120%o ea-s ed ata tivitya on tot xper-i

e 26y

7

7

ift on4 7%

TM d repr

P Centr

TMSa

n/ab

I22

32

R18

34

N s are m

aused by appropriately timed transcranial magnetic station (TMS) over the motor cortex to delay subjects’ moerformance (Day et al., 1989). Their subjects additional

udged the perceived time of their reactions, using the mef Libet and colleagues (Libet, Gleason, Wright, & Pear983). The results showed that subjects were largely unaf the performance delay produced by cortical TMS, sugg

ng at least a partial dissociation between motor performnd motor awareness.

More recently,Desmurget et al. (1999)reported that visuootor adjustments in the double-step task could be aboly TMS over the posterior parietal cortex (PPC). Moreoeveral authors have additionally proposed that the paortex acts as an “auto-pilot” for aimed movement (Pisellat al., 2000) and does not contribute to conscious awarene therefore reasoned that the motor awareness of v

able 2ean and S.D. across subjects of endpoint× deviation (mm), for initial an

osition of target Left (target =−40 mm)

TMSa No TMSa

Awareb Unawareb Awareb Unawareb

nitial movementMean (mm) −12 −18 −15 −17S.D. (mm) 46 39 44 44

eproduced movementMean (mm) −30 −24 −30 −28S.D. (mm) 36 38 29 36

ote: Negative numbers indication mm left of centre, positive numbera Presence or absence of TMS.b Subjective awareness of target jump.

f the subject’s motor threshold. Horizontal EOG was mured with surface electrodes. Target jumps were triggerrandom 25–100 ms interval after the onset of EOG acssociated with the large saccade from peripheral fixati

he central target. The methods were otherwise as for ement 1.

Five new right-handed subjects (4 female, mean agears) participated in the experiment.

. Experiment 2

.1. Results

Subjects reported perceptual awareness of a target sh7% of jump trials in which TMS was delivered, and 4

oduced movements with and without TMS

e (target = 0 mm) Right (target = 40 mm)

No TMSa TMSa No TMSa

n/ab Awareb Unawareb Awareb Unawareb

21 64 62 62 6235 52 50 47 54

19 72 67 69 6738 45 46 39 52

m right of centre.

234 H. Johnson, P. Haggard / Neuropsychologia 43 (2005) 227–237

Fig. 5. Mean endpoints in perturbation (“jump”) trials (mm).

of jump trials without TMS. A 2.7% of trials were elimi-nated from the kinematic analysis because of pre-emptiveeye or hand movements, 10.6% because of equipment fail-ure, and 8.2% because of movements in the wrong directionwith respect to the target shift. The endpoints for initial andreproduced perturbed movements are shown inTable 2andFig. 5.

We first analysed the no TMS trials to assess whether thepattern of results found in experiment 1 was replicated inexperiment 2. To establish whether visuomotor adjustmentshad been made, the endpoints in each target jump conditionwere compared with the endpoints of the unperturbed centretarget trials.

Family wise Bonferroni corrections were performed, andthe significance level for all of the followingt-tests is 0.025.

All of the non-TMS jump trials differed significantly fromthe unperturbed trials regardless of whether the subject wasaware (left:t4 = 10.85,P< 0.001, right:t4 =−8.69,P= 0.001)or unaware (left: t4 = 8.30, P= 0.001, right: t4 =−7.19,P= 0.002) of the target shift. This confirms that visuomo-tor adjustments were being made regardless of perceptualawareness.

To establish whether these adjustments entered into mo-tor awareness, as they had in experiment 1, the reproducedendpoints of jump trials were compared with the reproducede cedm un-p ware( -ao

ffecto ls inw aredw nif-

icant difference was found for leftward (t4 = 0.39,P= 0.71)or rightward (t4 = 0.45,P= 0.68) adjustments. The pattern ofresults seen in experiment 1 was therefore replicated in thenon-TMS trials of experiment 2.

7.1.1. No effect of TMS on visuomotor adjustmentA further series of comparisons focussed on effects of

TMS on visuomotor adjustment. The endpoints of the TMSjump trials were compared with the endpoints of the unper-turbed TMS centre point trials. Family wise Bonferroni cor-rections were again performed, and the significance level forall of the followingt-tests is 0.025.

All of the TMS jump trials differed significantly from theunperturbed TMS trials regardless of whether the subject wasaware (left:t4 = 7.88,P= 0.001, right:t4 =−6.08,P= 0.004)or unaware (left: t4 = 8.90, P= 0.001, right: t4 =−7.51,P= 0.002) of the target shift. Therefore, visuomotor adjust-ments were clearly present in trials where TMS was appliedto the PPC.

Although TMS did not abolish visuomotor adjustmentsit might nevertheless have subtly changed their kinematicpattern. Therefore, the initial movement endpoints of jumptrials in which TMS was administered were compared withthe endpoints of no TMS jump trials. No significant differ-ences were found regardless of whether subjects were aware( -at

ve inj intsf ointd

re-d af-f main-i f theT un-tP

8

8

, wer tionb een ine ucedm erep rgetj ove-m nesse 1.

siono by

ndpoints from the centre point condition. In the reproduovements, jump trials again differed significantly fromerturbed trials regardless of whether the subject was aleft: t4 = 6.28,P= 0.003, right:t4 =−10.32,P< 0.001) or unware (left:t4 = 7.84,P= 0.001, right:t4 =−6.56,P= 0.003)f the target shift.

Finally, to confirm that perceptual awareness had no en motor awareness reproduced endpoints of jump triahich subjects were aware of the target shift were compith those on trials in which they were unaware. No sig

left: t4 =−1.87,P= 0.14, right:t4 =−0.38,P= 0.73) or unware (left:t4 = 0.24,P= 0.82, right:t4 = 0.19,P= 0.86) ofhe target shift.

We also speculated that TMS could have been effectiust a small subset of trials. However, plotting the endpoor individual trials showed essentially overlapping endpistributions for TMS and non-TMS trials (Fig. 6).

Finally, although it would have been contrary to our pictions, it could still have been possible that TMS had

ected motor awareness, despite motor performance reng unaffected. Therefore, the reproduced endpoints oMS jump trials were compared with their non-TMS co

erparts. No significant differences were found (allt4 < 2, all> 0.20).

. Experiment 2

.1. Discussion

Three features of experiment 2 are noteworthy. Firsteplicated in a further group of subjects the dissociaetween motor awareness and perceptual awareness sxperiment 1. Again, we observed that subjects reprodotor adjustments following target jumps of which they werceptually unaware. The use of EOG to trigger the ta

ump in the present experiment is a methodological imprent which makes the lack of any perceptual awareffects more convincing, if anything, than in experiment

Second, we entirely failed to obtain the suppresf visuomotor adjustments by parietal TMS reported

H. Johnson, P. Haggard / Neuropsychologia 43 (2005) 227–237 235

Fig. 6. Endpoints of perturbed and unperturbed trials for each subject, with and without TMS (mm). The target was located at (0, 240).

Desmurget et al. (1999)despite using comparable methods.We speculate on two possible reasons for this. Although weused their procedure for localising the TMS site, Desmurgetet al. subsequently confirmed the location of their stimula-tion using MRI. They also noted that one of their subjectswho showed a slightly different pattern of results had beenstimulated more medially than the others. We did not con-firm our stimulation site with MRI, so anatomical variationsor incorrect localisation might have resulted in inappropriatestimulation sites for some of our subjects. On the other hand,one of their subjects who did not show the inhibitory effectof TMS, did appear to have been stimulated appropriately.Therefore, there may be substantial individual differences inresponsiveness to parietal TMS. Finally, the TMS coil andstimulator in their study may have been more powerful thanours. In particular their stimulation may have been effectiveat deeper sites within the intraparietal sulcus than ours.

Third, we found no effects of parietal TMS on motorawareness, measured as the discrepancy between initial andreproduced visuomotor adjustments. This is, of course, a nullresult, and we had not predicted any such effects. Moreover,one might argue that the absence of any TMS effect on motorperformance would make effects on motor awareness un-likely. Nevertheless, we note that our failure to alter motorawareness is at least consistent with the view that the PPCprovides an “auto-pilot” for on-line movement control, (Greaet al., 2002; Pisella et al., 2000) independent of conscious ex-perience.

8.2. General discussion and conclusions

Previous studies have suggested that the motor system canrespond to information of which we are perceptually unaware(Pelisson et al., 1986) Our own work had further shown that

236 H. Johnson, P. Haggard / Neuropsychologia 43 (2005) 227–237

people’s awareness of their own motor output may involvean attenuated or an exaggerated perception of their actualmovement, essentially depending on the extent to which themovement was intended or not (Johnson et al., 2002). In thepresent study, we have investigated the link between these twoapproaches, by investigating whether perceptual awareness ofa target shift modulates motor awareness of the visuomotoradjustment to that shift during reaching movements.

Our results show that awareness of adjustment to ongoingmovements is independent of perceptual awareness of any tar-get displacement necessitating such adjustment. Therefore,we have shown that motor awareness and perceptual aware-ness can be dissociated, at least in the double-step task. Thisdissociation appears relatively complete: the magnitudes oflateral deviations in reproduced movements were numeri-cally close to those in the corresponding initial movements,and not just attenuated versions of the original movements. Toconclude, we comment on some psychologically importantfeatures of this dissociation.

Our result suggests a paradox or incompatibility at theheart of subject’s experience of their own action. In some tri-als, our subjects reported a stationary target on the one hand,yet a clearly deviated trajectory on the other. At face value,resolution of these incompatible situations would seem possi-ble only if subjects generate an alternative explanation, othert tra-j ablya cu-t mentt ith-f thisi tiona e ofm t theirl rrors( kt cep-t otora ferena erentn somer cro-c1

con-c ed byJ ilityt bodym andsO scioua n thee r. Web s andm nder-s rateda h the

brain processes and the conscious experience of being anagent.

Acknowledgements

HJ was supported by a BBSRC studentship. Further sup-port was provided from a BBSRC Project grant to PH, a Lev-erhulme Research Fellowship awarded to PH, and a RoyalSociety Equipment grant.We are grateful to Louise Whiteleyfor assistance with preparing this work for publication.

References

Carey, D. P., Hargreaves, E. L., & Goodale, M. A. (1996). Reachingto ipsilateral or contralateral targets: within-hemisphere visuomotorprocessing cannot explain hemispatial differences in motor control.Experimental Brain Research, 112, 496–504.

Carey, D. P., & Otto de Hart, E. G. (2001). Hemispatial differences invisually guided aiming are neither hemispatial nor visual.Neuropsy-chologia, 39, 885–894.

Castiello, U., Paulignan, Y., & Jeannerod, M. (1991). Temporal dissocia-tion of motor responses and subjective awareness. A study in normalsubjects.Brain, 114, 2639–2655.

Day, B. L., & Brown, P. (2001). Evidence for subcortical involvement inthe visual control of human reaching.Brain, 124, 1832–1840.

D ticl

D hout,utionn in

D nder,x ine

F and-nitive

G ., &ove-

G for

G fton,upts

H -

H with

H ts to

J motor.

J et

J reness

han a displacement of the target, of why they made aectory deviation. For example, subjects could conceivttribute the deviation of their initial movement to an exe

ion error that accidentally arose in the course of a moveo an apparently stationary target. They would then faully reproduce this error. Two reasons lead us to rejectdea. First, our subjects did not show the sort of frustrand expletives typically associated with the experiencovement error. Second, subjects presumably knew tha

ateral deviations were larger than typical execution evan Beers, Haggard, & Wolpert, 2004). Instead, we thinhe mind does not even attempt to resolve lack of perual awareness with positive motor awareness. Rather, mwareness and perceptual awareness might be quite difnd independent processes in the brain, which have diffeural substrates and different natural time courses. Inespects, this position recalls the concept of multiple “mionsciousnesses” within the visual system (Zeki & Bartels,999).

We end by noting a possible relation between ourept of motor awareness and the concept of agency useannerod (2002). The sense of agency derives from an abo compare motor commands with reafference from theovements and external events caused by the commur data suggests that subjects can have a clear conwareness of their own movement patterns even whexternal stimulus event causing the movement is uncleaelieve the dissociation between perceptual awarenesotor awareness may be an important clue towards u

tanding how the brain distinguishes between self-genend external events, and towards understanding bot

t

.s

ay, B. L., & Lyon, I. N. (2000). Voluntary modification of automaarm movements evoked by motion of a visual target.ExperimentaBrain Research, 130, 159–168.

ay, B. L., Rothwell, J. C., Thompson, P. D., Maertens de NoordA., Nakashima, K., Shannon, K., et al. (1989). Delay in the execof voluntary movement by electrical or magnetic brain stimulatiointact man.Brain, 112, 649–663.

esmurget, M., Epstein, C. M., Turner, R. S., Prablanc, C., AlexaG. E., & Grafton, T. (1999). Role of the posterior parietal corteupdating reaching movements to a visual target.Nature Neuroscienc,2, 563–567.

arah, M. J., & Feinberg, T. E. (2000). Disorders of perceptionawareness. In M. J. Farah & T. E. Feinberg (Eds.),Patient-based Approaches to Cognitive Neuroscience. Issues in Clinical and CogNeuropsychology. Cambridge, MA: MIT Press.

aveau, V., Martin, O., Prablanc, C., Pelisson, D., Urquizar, CDesmurget, M. (2003). On-line modification of saccadic eye mments by retinal signals.Neuroreport, 14(6), 875–878.

oodale, M. A., & Milner, A. D. (1992). Separate visual pathwaysperception and action.Trends in Neuroscience, 15, 20–25.

rea, H., Pisella, L., Rossetti, Y., Desmurget, M., Tilikete, C., GraS., et al. (2002). A lesion of the posterior parietal cortex disron-line adjustments during aiming movements.Neuropsychologia, 40,2471–2480.

aggard, P. (2001). The psychology of action.British Journal of Psychology, 92, 113–128.

aggard, P., & Magno, E. (1999). Localising awareness of actiontranscranial magnetic stimulation.Experimental Brain Research, 127,102–107.

allet, P. E., & Lightstone, A. D. (1976). Saccadic eye movemenflashed targets.Vision Research, 16, 107–114.

eannerod, M. (1994). The representing brain: Neural correlates ofintention and imagery.Behavioral and Brain Sciences, 17, 187–245

eannerod, M. (2002).La nature de l’esprit: Sciences CognitivesCerveau. Paris: Odile Jacob.

ohnson, H., van Beers, R. J., & Haggard, P. (2002). Action and awain pointing tasks.Experimental Brain Research, 146, 451–459.

H. Johnson, P. Haggard / Neuropsychologia 43 (2005) 227–237 237

Libet, B., Gleason, C. A., Wright, E. W., & Pearl, D. K. (1983). Timeof conscious intention to act in relation to onset of cerebral activities(readiness potential): The unconscious initiation of a freely voluntaryact.Brain, 106, 623–642.

Matin, E. (1974). Saccadic suppression: Review and an analysis.Psycho-logical Bulletin, 81, 899–917.

Paulignan, Y., MacKenzie, C., Marteniuk, R., & Jeannerod, M. (1991).Selective perturbation of visual input during prehension movements.Experimental Brain Research, 83, 502–512.

Pelisson, D., Prablanc, C., Goodale, M. A., & Jeannerod, M. (1986).Visual control of reaching movements without vision of the limb.Experimental Brain Research, 62, 303–311.

Pisella, L., Grea, H., Tilikete, C., Vighetto, A., Desmurget, M., Rode, G.,et al. (2000). An automatic pilot for the hand in human posterior pari-etal cortex: Toward reinterpreting optic ataxia.Nature Neuroscience,3, 729–736.

Prablanc, C., & Martin, O. (1992). Automatic control during reaching atundetected two-dimensional target displacements.Journal of Neuro-physiology, 67, 455–469.

van Beers, R. J., Haggard, P., & Wolpert, D. M. (2004). The role ofexecution noise in movement variability.Journal of Neurophysiology,91, 1050–1063.

Zeki, S., & Bartels, A. (1999). Toward a theory of visual consciousness.Consciousness and Cognition, 8, 225–259.