Performing hand actions assists the visual discrimination of similar hand postures.

Abstract

Recent theoretical work has suggested that internal predictive signals are used for motor control and coordination. The predictive signal - proposed to be the output of a forward model - would be a sensory representation of action. Hence, these sensory representations could potentially influence other sensory processes. We report here how performance of hand actions assisted the visual discrimination of target hand postures presented at random times within an on-going series of hand images. Reaction times to discriminate the targets were significantly shorter when the displayed images were both sequential and congruent with the action being performed. Hence, the planning or execution of action appears to allow better prediction of a displayed series of congruent images. In further control experiments, we show that the motor-visual priming effect is unlikely to be due to differential attentional demands and it is specific to a first person perspective display; it is short lasting, being lost if a 500 ms delay is introduced between successive stimulus presentations. The data are interpreted as evidence supporting the hypothesis that forward models in the motor system provide action-specific sensory predictions that are available to cognitive processes.

Full text

Page 1

Neuropsychologia 44 (2006) 966–976
Performing hand actions assists the visual discrimination
of similar hand postures
R.C. Mialla,b,∗, J. Stanleya, S. Todhunterb, C. Levickb, S. Lindoa, J.D. Miallb,c
aBehavioural Brain Sciences, School of Psychology, University of Birmingham, Edgbaston, Birmingham B15 2TT, UK
bUniversity Laboratory of Physiology, Oxford OX1 3PT, UK
cCherwell School, Oxford OX2 7EE, UK
Available online 24 October 2005
Abstract
Recent theoretical work has suggested that internal predictive signals are used for motor control and coordination. The predictive signal –
proposed to be the output of a forward model – would be a sensory representation of action. Hence, these sensory representations could potentially
influence other sensory processes. We report here how performance of hand actions assisted the visual discrimination of target hand postures
presented at random times within an on-going series of hand images. Reaction times to discriminate the targets were significantly shorter when
the displayed images were both sequential and congruent with the action being performed. Hence, the planning or execution of action appears to
allow better prediction of a displayed series of congruent images. In further control experiments, we show that the motor–visual priming effect is
unlikely to be due to differential attentional demands and it is specific to a first person perspective display; it is short lasting, being lost if a 500ms
delay is introduced between successive stimulus presentations. The data are interpreted as evidence supporting the hypothesis that forward models
in the motor system provide action-specific sensory predictions that are available to cognitive processes.
© 2005 Elsevier Ltd. All rights reserved.
Keywords: Forward model; Prediction; Motor planning; Visual discrimination; Reaction time
1. Introduction
The concept that the nervous system internally models the
behaviourofthemotorsystemhasgainedincreasingprominence
over recent years (Jordan, 1995; Kawato, Furukawa, & Suzuki,
1987; Kawato, 1999). Within this concept, forward models cap-
ture the forward or causal relationship between actions and the
resultant change in the state of the motor system (Jordan &
Rumelhart, 1992). These models estimate the next sensory state
of the motor system based upon information about its current
state, its dynamics and the motor command being issued to it.
Forward models can support sensorimotor control in many
ways,includingsensoryconfirmation,internalfeedback,context
estimation and state estimation (Haruno, Wolpert, & Kawato,
2001; Jordan et al., 1992; Miall & Wolpert, 1996; Wolpert,
Ghahramani, & Jordan, 1995). There is also evidence that we
distinguish the sensory consequences of our own actions from
externally produced stimuli using a forward model (Wolpert
∗Corresponding author. Tel.: +44 121 414 2867; fax: +44 121 414 4897.
E-mail address: r.c.miall@bham.ac.uk (R.C. Miall).
et al., 1995; Wolpert & Ghahramani, 2000). Reafferent, self-
induced sensations can therefore be cancelled or attenuated
duringmovement,thushighlightingexafferentinformationcrit-
ical for control.
It has been proposed that the process used by the forward
model to predict the sensory consequences of one’s own move-
ments could also be available to cognitive processes (Decety &
Grezes, 1999; Frith, Blakemore, & Wolpert, 2000; Jeannerod &
Decety, 1995). Forward models providing sensory predictions
within the motor system could provide a visual signal, or simi-
lar representation, which would assist in vision-based cognitive
tasks. Psychophysical studies have suggested that internal rep-
resentation of action is indeed used to solve motor-related tasks.
When asked to judge the laterality of visually presented hands,
subjects mentally rotate their own hand into the stimulus orien-
tation for comparison (Gentilucci, Daprati, & Gangitano, 1998;
Parsons,1994;Parsonsetal.,1995).Similarly,whensubjectsare
askedtoestimatethefeasibilityofgraspingobjectsplacedatdif-
ferent orientations, there is a correspondence between response
time and the time taken to actually reach and grasp an object
placedatthesameorientation,suggestingthatthesubjectsmen-
tally move into the appropriate position in order to generate an
0028-3932/$ – see front matter © 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.neuropsychologia.2005.09.006

Page 2

R.C. Miall et al. / Neuropsychologia 44 (2006) 966–976
967
internalrepresentationofthearmthatthensupportsthedecision
process (Frak, Paulignan, & Jeannerod, 2001). However, these
experiments cannot distinguish whether the mental rehearsal of
the action is undertaken to generate a sensory representation, in
order to solve the cognitive task, or whether it is undertaken to
evokeamotorintentionorplan,whichisusedwithoutaspecific
sensory representation.
Recently, severalvisual
Bekkering, & Prinz, 2001; Brass, Bekkering, Wohlschlager,
& Prinz, 2000; Craighero, Bello, Fadiga, & Rizzolatti, 2002;
Craighero, Fadiga, Rizzolatti, & Umilta, 1999) have demon-
strated that visual images can prime the motor system and thus
lead to faster actions when the cue and the action are congruent
(visuo-motor priming). Craighero et al. (1999, 2002) showed
thatinitiationofapre-specifiedreachtograsphandactioncanbe
modulated by prior viewing of pictures of a hand that matched
or did not match the planned hand orientation. They argued
(Craighero et al., 2002) that the reduced response times were
consistent with motor–visual priming, such that the motor plan
led to more rapid visual processing of the cue to move. Unfortu-
nately, as they acknowledged (p. 498), they could not eliminate
the possibility that their results were in fact due to visuo-motor
priming, and that the congruence of the cue image resulted in
more rapid initiation of the planned action than did incongruent
cues.Weinterprettheirdataasaneffectofvisuo-motorpriming.
More recently, Hamilton, Wolpert, and Frith (2004) showed
that performing an action influenced judgement of observed
actions, but found that the effect was one of interference:
holding a heavy weight biased subjects to report that weights
they observed being lifted were lighter than they really were.
They suggest the forward model activated during the action
cannot simultaneously be used for the perceptual task, and
so the subjects’ reports are biased away from their performed
action. Other studies have tested the effects of action on per-
ception, under the theory of event coding (Hommel, Musseler,
Aschersleben, & Prinz, 2001), but without a specific, direct
relationshipbetweenvisualoutcomeofactionandtheperceived
image—for example, Musseler and co-workers (Musseler &
Hommel, 1997; Musseler, Steininger, & Wuhr, 2001) presented
an arrowhead whose orientation reflects which of two buttons
are struck, but the arrowhead has only an arbitrary relationship
to finger shape or position. Actions can also influence spatial
or temporal judgments about visual and somatosensory stimuli
(Ross, Morrone, Goldberg, & Burr, 2001; Smith, Rorden, &
Jackson, 2004; Yarrow, Haggard, Heal, Brown, & Rothwell,
2001). However, it is not clear that these phenomena are
based on specific sensory predictions of the outcome of the
planned action, rather than a more general remapping of the
spatio-temporal relationships of any sensory stimuli (Duhamel,
Colby, & Goldberg, 1992) around the time of action.
Hence,wedonotknowofanyreportsthathavedemonstrated
that motor preparation or performance can selectively facilitate
processes within the visual system, as might be expected from
a forward model prediction of the action. We report evidence of
this effect in a human visual discrimination task, in five related
experiments. Visual stimuli comprised of images of a hand pre-
sented in a slow sequence on a computer screen. Oddball target
cueingexperiments(Brass,
imageswithinthesequenceshowedadifferenthandposture,and
the subjects vocally responded to these oddball images. During
the task, subjects made slow hand actions that were congruent
or incongruent with the displayed image series. We hypothe-
sise that prediction of the observed sequence of images would
allow more rapid discrimination of the oddball targets. Hence,
we expect that if the observed image sequence is predictable,
discrimination will be faster than if it is random. If the predic-
tion of the observed sequence is facilitated by performance of
a congruent action, then we expect yet faster discrimination. In
contrast, if Hamilton et al. (2004) are correct, then the opposite
effect should be seen: performance of congruent actions should
impede discrimination.
2. Experiment 1: Congruency and prediction
2.1. Methods
Twenty-eight subjects (9 female, 19 male) participated in the
first experiment, after giving informed consent; the experiments
wereapprovedbytheCentralOxfordshireResearchEthicsCom-
mittee. To motivate them, each subject received £5 payment;
subjects with mean reaction times that were in the top third of
the distribution received an additional £5 reward. The mean age
ofthesubjectgroupwas19.4years,rangingfrom18to22years.
All were right handed, had normal or corrected to normal vision
and were na¨ ıve to the purpose of the experiment.
The experiment took place in a sound attenuated, dimly lit
room. Participants sat at a table in front of a 44cm computer
monitor with a viewing distance of 60cm; they wore a micro-
phone headset and held their left hand either above their left leg,
orresteditonthetable(Fig.1A),intheperipheryoftheirvision.
Thetaskconsistedofdiscriminatingatargetimagerandomly
presented within a sequence of static images (Fig. 2), vocally
responding ‘ta’ as fast as possible into a microphone that was
connected to a voice-activated switch. All images were static
viewsofahumanlefthand,renderedas450×450pixelbitmaps
withthePoserPro4animationpackage(CuriousLabsInc.),and
displayed at a rate of one image per second using Presentation
(NeurobehavioralSystems,USA).Twomainsequencesshowed
the hand in 15 positions ranging from a closed fist to full exten-
sion of the fingers (the ‘fist’ sequence, Fig. 2), or from the open
handinpronationtosupination(the‘flip’sequence).Eachimage
was presented to the left of a fixation pointer, in a panel 8cm
square, occupying approximately 2–10 degrees left of fixation.
Each trial consisted of 200 image presentations from the
main sequence, either selected randomly from the 15 images, or
cyclingbackwardsandforwardsthroughthesequence,selecting
everyfourthimageinorder.Thus,thedisplayedseriesofimages
was either unpredictable or showed a repeated hand action at a
rate of two complete cycles every 15s. Each image was pre-
sented for 1s, without gaps; we chose to use slow presentation
of static images so that the random sequence could be used.
Two oddball target images showed a hand with two extended
fingers (‘peace’, used for the fist series, Fig. 2) or with opposed
indexfingerandthumb(the‘OK’signal,usedfortheflipseries).
These images were chosen to have some similarity with the

Page 3

968
R.C. Miall et al. / Neuropsychologia 44 (2006) 966–976
Fig. 1. An experimental setup: (A) subject is shown wearing the microphone with her left hand in the starting position. (B) Image and fixation point (right hand small
black “lollipop” icon) are shown at the top, as seen on the computer monitor; hence, the image of the hand is presented to the left visual hemifield; the participant’s
left hand is shown in the foreground (bottom). The oscillating motion of fixation pointer was used to cue and synchronize the hand movements of the subject. (C)
Examples of the realistic and box-rendered images used in Experiment 5.
Fig. 2. Sequences of stimuli in Experiment 1: (A) images from the ‘fist’ series
were presented each second. On appearance of the randomly presented ‘peace’
target image (bottom right), participants were to say ‘ta’ into a microphone. The
target image was randomly inserted into the main sequence. (B) Images from
the ‘flip’ series and the ‘okay’ target image (right).
hand postures of the main series, so that subjects could not dis-
criminate the targets on the basis of the overall image size or
orientation, for example. The target image was inserted at infre-
quent random points within the main series of presentations,
replacing the images at those points. Oddball targets were never
presented within the first 10s (10 stimulus presentations) of a
trial; the inter-target interval was a minimum of 5s; and other-
wise their positions in the sequence were chosen with a random
number generator with a probability of 0.1 for each sequence
position. On average, targets were presented 8.5 times in the
series of 200 images.
Reaction times (RT) to the target were measured from the
subject’s vocal responses. This avoided the inter-manual inter-
ference that had been detected in pilot experiments in which
subjects reacted by clicking a computer mouse button. A voice-
activated switch was used to generate a TTL pulse that was
detected by the Presentation program. Prior to the start of
the experiment, each subject practiced responding sufficiently
loudly to activate the switch.
While discriminating these images, subjects also performed
either the fist or flip action with their left hand. The images dis-
playedonscreenwerethereforefromactionsthatwerethesame
as those being performed (congruent condition, both displayed
images and performed action were flip, for example) or differ-
ent (incongruent condition, performing flip but observing fist,
or vice versa). Note that we use the term congruent to indicate
the match between image set and action: in the random condi-
tion there is no temporal match between the performed action
and observed images. A small lollipop-shaped fixation pointer
was displayed 4cm to the right of the stimulus panel (Fig. 1B),
oscillating left and right in time with the presentation of the
images (i.e. at a rate of two cycles every 15s) and thus acting as
a visual metronome for the hand movement the subject was to
perform. Subjects were instructed to time their hand movement

Page 4

R.C. Miall et al. / Neuropsychologia 44 (2006) 966–976
969
using this metronome, rather than by attempting to match their
hand movement to the displayed action.
At the start of each trial, subjects were visually instructed of
the sequence order to be displayed (random or sequential), the
action to perform (flip or fist), the target image to be detected
(peace or OK), and whether the display and action sequences
were congruent or incongruent. Verbal instructions were also
given for initial trials within the experiment, to ensure that each
subject was aware of the experimental conditions. Each con-
dition was presented using both main image sequences (i.e. the
sequenceandtargetcombinations:flipandOK,orfistandpeace).
Each subject performed 18 trials each lasting 200s, com-
prised of two practice conditions and eight test conditions
repeated twice in succession. The order of tests was balanced
across subjects to reduce any influence of learning or fatigue.
Between trials, participants were permitted a small break (usu-
ally less than a minute) to rest their hands and eyes.
Twotypesoferrorsweredetectedanddiscardedfromfurther
analysis: late or missed responses (considered to be RTs longer
than 1000ms), and anticipation errors (defined as RTs shorter
than350ms).Falsepositiveerrors–respondingtoanon-target–
were not logged by the program, but were very infrequent. Each
subject’sreactiontimeswerethenaveragedacrossallremaining
responseswithineachcondition.Becauseoftherandompresen-
tation of targets and occasional missed responses, between 15
and 22 responses were recorded per condition (typically 17).
Subject mean responses were then averaged across the group,
and differences tested with repeated measures ANOVA using
SPSS. A 2×2 factorial within-subject design was used with
the factors: congruency between the performed hand action and
the displayed sequence (congruent versus incongruent) and the
order of the displayed sequence (sequential versus random).
2.2. Results
We first tested whether predictability of the main sequence
of images improved discrimination performance and in addi-
tion whether congruency between the hand actions that were
performed and the images seen on screen affected discrimina-
tion.Subjectsthereforeviewedeitherrandom(unpredictable)or
sequential series of hand images while performing the slow fist
or flip hand actions, timed by the oscillating motion of a small
fixation pointer (Fig. 1).
There were no anticipation errors. Late or missed responses
to the targets (RTs>1000ms) accounted for 0.42% of all trials
(5.1% of target trials). However, two subjects had error rates of
20 and 26% for target trials, falling outside the 99% confidence
intervalofthegroupmean,andwerediscardedfromfurtheranal-
ysis because of poor compliance with the task. Of the remaining
26subjects,themeanerrorratefortargettrialswas3.5%andthe
maximumwas6%.Astheseerrorrateswerelow,noattemptwas
made to analyse their occurrence with respect to task condition.
Mean reaction times for the four experimental condi-
tions, averaged across the two image series are shown in
Fig. 3. There was a highly significant effect of sequence order
(F(1,25)=13.889;p=0.001)withparticipantsrespondingfaster
to targets within a predictive main sequence than to targets
Fig. 3. Experiment 1: average reaction times (±1S.E.M.) for discrimination of
the target images during performance of congruent (filled circles, solid line) or
incongruentactions(hollowcircle,dashedline).Theorderofthemaindisplayed
image was sequential or random.
within a randomly presented main sequence (17.2ms differ-
ence, S.E.=4.9). However, most of this difference was driven
by the large reduction in reaction times in the congruent move-
ment condition. RTs were faster overall in the two congruent
conditions than in the two incongruent conditions (mean dif-
ference: 7.7ms, S.E.=2.3; F(1,25)=10.997; p=0.003), and the
interaction between congruency and sequence was highly sig-
nificant (F(1,25)=13.889; p<0.001). That is, the shortest reac-
tion times were for the sequential–congruent condition, 23.1ms
faster than the sequential-incongruent condition (S.E. of dif-
ference=4.8ms). Post hoc paired sample t-tests confirmed that
responses in the sequential–congruent condition were signifi-
cantlyfasterthaninallthreeotherconditions(26.9msfasterthan
the mean of other three conditions, S.E.=4.9ms; statistically
significantforbothimageseries:d.f.=25;t>3.95;p< =0.001).
The1.8msdifference(S.E.=6.7ms)betweenthesequentialand
random series in the incongruent condition was not significant.
Unexpectedly, the image series also had an effect on reaction
times, with faster mean response times in the ‘fist’ sequence.
Re-analysisofthedatausinga2×2×2factorialdesign(includ-
ing the two image series as the third factor), showed that the
main effect of image series was significant (F(1,25)=16.414;
p<0.001), and that interactions between series and congru-
ency and between series and sequence and congruency were
also significant (F(1,25)=5.44, 6.927; p=0.028, 0.014, respec-
tively). However, crucially, the interaction between congruency
and sequence remained highly significant (F(1,25)=10.997;
p=0.003). Fig. 4 shows the data separated by image series; in
both sets, the RTs in the sequential condition are lowest for the
congruent condition, and in the random condition there is no
difference between congruent and incongruent tasks.
2.3. Discussion
For this first experiment, we had hypothesised that predic-
tion of the main sequence of displayed images would aid in
the discrimination of the target images. Cued by this predic-
tion, detection of the target images should be facilitated, as

Page 5

970
R.C. Miall et al. / Neuropsychologia 44 (2006) 966–976
Fig. 4. Experiment 1: average reaction times (ms, ±S.E.M.) for detection of
targetstimuli,duringsequentialandrandomvisualpresentationsofhandstimuli.
Data are divided by displayed hand stimulus (fist or flip series) and whether the
participant’sperformedhandmovementwascongruent(filledcircles,solidline)
or incongruent (hollow circles, dotted line) with this series.
the oddball targets do not match this anticipated visual repre-
sentation. Reaction times were therefore expected to be faster
when images were presented in a sequence rather than at ran-
dom, and this main effect was found. This effect was small,
however, and non-significant in the incongruent conditions. We
further hypothesised that when a voluntary action is performed
that is congruent with the displayed images, the correspond-
ing prediction produced by a forward model should also assist
in visual discrimination and further reduce reaction times. This
second hypothesis was also supported by finding a significant
interactionbetweenthecongruencyandimagesequencefactors.
The conjunction of these two predictive conditions – sequential
image display and congruent action – produced reaction times
significantly shorter than in the other three conditions, with a
mean difference of 27ms.
3. Experiment 2: Attention
Observing human actions while performing different actions
has been shown to affect performance (Kilner, Paulignan, &
Blakemore, 2003); it is easier to perform actions when they are
congruentwiththeobservedaction.Thesameistrueforcongru-
encebetweenactionsandvisualfeedback(Poulton,1974).Thus,
one possible explanation for the reaction time savings seen dur-
ing performance of congruent actions compared to incongruent
actions is that there was an attentional difference between the
tasks. To address this, we ran an experiment in which subjects
were challenged to discriminate a change in the shape of the fix-
ation pointer. If performance of congruent actions allows more
attention to be paid to the discrimination task, then we would
expect reaction time savings to be seen in this task as well.
3.1. Methods
Sixteen subjects participated in this experiment (5 male, 11
female); none had taken part in the previous experiment; age
range was 18–30 years (mean 21.6 years).
All subjects performed two of the conditions presented in
Experiment 1, namely the discrimination of the target hand
imageswithinthesequential“fist”and“flip”imageseries,while
performing congruent and incongruent hand actions. They were
alsotestedinamodifieddiscriminationtask(the“fixationtask”)
inwhichtheyrespondedtothesuddenchangeofthefixationcue
from a rotating pointer (Fig. 1) to a cross, while again perform-
ing hand actions congruent or incongruent with the displayed
image sequences.
The task, apparatus, stimuli and the experimental situation
were the same as Experiment 1 with the exception that the fix-
ation cue changed to a small cross, in the fixation target task.
As in Experiment 1, the presentation of all images (whether the
hand sequence, the hand target or the fixation cross) lasted for
1s.
3.2. Results
For this new group of 16 subjects, there were no anticipation
errors; late or missing responses (RT>1000ms) accounted for
0.17% of all trials and 2.44% of target trials. One of the 16 sub-
jects performed poorly compared to the rest of the group, with
RT standard deviations>130ms for six out of eight conditions,
and was excluded from analysis. As expected from Experi-
ment 1, responses were faster in the congruent condition than in
the incongruent conditions (Fig. 5: 13.7ms, S.E.=7.1ms; one-
tailedt-test,t=1.916,p=0.038);however,theplannedANOVA
did not reach statistical significance. The difference between
reaction times when discriminating change of the fixation cue
wasnegligible(2.4ms,S.E.=5.9ms;two-tailedt-test,t=0.406,
p=0.691).
3.3. Discussion
This experiment confirmed the reaction time savings for dis-
criminationofhandimagescongruentwiththeperformedaction,
Fig. 5. Experiment 2: average reaction times (±1S.E.M.) for discrimination of
the target images during performance of congruent (filled circles, solid line) or
incongruent actions (hollow circle, dashed line). The task was to discriminate
target hands within the displayed sequence of hand images (“hand task”) or to
discriminate change in the fixation pointer from a lollipop shape (Fig. 1) to a
cross (“fixation task”).

Page 6

R.C. Miall et al. / Neuropsychologia 44 (2006) 966–976
971
while finding no reaction time differences for discrimination of
the pointer. It allows us to rule out differences in attentional
resources between the congruent and incongruent conditions as
the cause of the reaction time savings observed.
4. Experiment 3: Time course
We used the presentation of a slow series of static images,
initially chosen so that comparisons could be made between
thesequenceandrandomconditionsofExperiment1.However,
this then raises a question of when the predictions of each forth-
coming image are generated and/or used: are the predictions
continuously generated, while the subjects perform the slow,
continuous hand movements, or are they generated or used only
at the time of each visual presentation? Previous experiments
on visuo-motor priming have shown that the effect of prim-
ing stimuli is short-lived, and have identified a time-window of
300–700ms over which the onset of a congruent cue can prime
responding (Vogt, Taylor, & Hopkins, 2003). We make a pre-
liminary address to this question by modifying the display, so
thateachimagewaspresentedfor500ms,separatedbyaneutral
background for 500ms.
4.1. Methods
Twelve (seven male, five female) of the original participants
in Experiment 1 participated in the third experiment, after an
interval of 5 months. Age range, 20–22 years (mean=20.8).
A simple comparison was made between two conditions:
congruent versus incongruent actions, in both cases for the
sequential picture display only. Unlike Experiment 1, we did
notusetherandomconditioninthisexperiment.Asbefore,both
‘fist’ and ‘flip’ image series were presented in each condition,
creating a 2×2 factorial design. Again, the order of test presen-
tations was manipulated between participants to minimize any
order effects.
The task, apparatus, stimuli and the experimental situation
werethesameasExperiment1withtheexceptionthatthevisual
images shown on the screen were reduced in duration from
1000 to 500ms, and were separated by a neutral grey screen
for 500ms. Hence, a short delay was introduced between pre-
sentations of each visual stimulus.
4.2. Results
Priortoanalysingthisexperiment,andinordertoconfirmthat
this subset of 12 subjects was typical of the original group of
subjects tested in Experiment 1, we correlated the performance
ofthetwosubgroups(n=12versustheremainingn=14)across
all eight conditions tested in Experiment 1 (two series, four task
conditions). The correlation was highly significant (r2=0.699,
p=0.009).Themeanreactiontimesavingseenforthissubgroup
(24.3ms,S.E.=7.3ms)inthecongruentsequentialconditionof
Experiment 1 versus the other three conditions was comparable
to that found in the whole group (27ms, S.E.=4.9ms). Hence,
thesubgroupretestedinExperiment3wastypicaloftheoriginal
group.
They showed no anticipation errors or late responses
(RT>1000ms); missing responses accounted for 0.1% of all
trials, 1.3% of target trials. As in Experiment 1, a significant
effect of image series was found (F(1,11)=5.361, p=0.041),
with shorter mean reaction times in the ‘fist’ condition. How-
ever, the effect of congruency was not significant—the mean
reaction times were only 3.7ms faster in the congruent con-
dition (S.E.=4.6). Nor was there any significant interaction
between the image sequence and the effect of congruency
(F(1,11)<0.492, p>0.247).
4.3. Discussion
These results indicate that, with the introduction of a 500ms
interval between successive stimulus presentations, the reaction
time savings seen in the first experiment were lost. While there
were again response differences between the two series (with a
small response advantage for the congruent condition over the
incongruent condition in the ‘flip’ series, and the reverse for
the ‘fist’ series), in neither case was the difference statistically
significant.
In our experiment, we assume (but cannot yet demonstrate)
that the forward model prediction of action is continuous, as the
actions are themselves continuous. But our results show that the
processthatweassumeunderliestheintegrationofacontinuous
internal forward model prediction of action with the periodic
process of visual discrimination appears to be short-lived and
does not span the 500ms interval. Vogt et al. (2003) reported a
similareffect,withatime-windowforeffectiveprimingofabout
300–700ms from prime cue onset to response onset. In other
words, if the priming stimulus was present for longer durations,
itwasineffective,stronglysuggestingthatitistheonsetmoment
that is critical. Hence, in our task, the effect of the internal,
action-related predictions on visual discrimination may also be
time-locked to the visual stimulus onset—in our case each 1s
updateofthestaticimagesofthehand.Clarifyingthisinteraction
will require additional experiments.
5. Experiment 4: Perspective
Visuo-motor priming appears to depend on perspective
(Craighero et al., 2002; Vogt et al., 2003), with differences
reported for movements cued by images of a hand shown in the
first person perspective against cues in the third person perspec-
tive. A third person viewpoint advantage might reflect imitative
experience,orexperienceofimagesseeninmirrors(Craigheroet
al., 2002); in contrast the first person effect observed for images
of hands may reflect the action-relevance of the cue, allowing
direct matching of the cue image with hand posture (Vogt et al.,
2003). The objective of this experiment was thus to determine
whetherthemotor–visualprimingeffectsshowninExperiments
1 and 2 are dependent on the perspective of the image being dis-
played concurrently on the screen. Hence, we repeated the two
mainconditionsofcongruentandincongruentactionsperformed
during a sequential image presentation (as tested in Experiment
1 and 2) in first versus third person perspective conditions, in
another group of subjects.

Page 7

972
R.C. Miall et al. / Neuropsychologia 44 (2006) 966–976
5.1. Methods
Eighteen subjects participated (10 male, 8 female); none had
taken part in the previous experiments; age range was 20–22
years.
A 2×2 factorial within-subject design was used with one
factor being the congruency between subjects’ own actions and
the main sequence displayed, and the other factor being the
perspective view of the images (first versus third). The main
sequence was presented only in sequential mode, as used in
Experiment 2, showing a predictably cyclic action. As before,
the main sequence presented was one of two hand actions (‘fist’
or ‘flip’). Again, the order of test presentations was manipulated
between participants to minimize any order effects.
The task, apparatus, stimuli and the experimental situation
were the same as Experiments 1 and 2 with the exception that
the visual images used in the third person view were top-bottom
inverted from the first person view, as if one’s own hand was
viewed in a mirror.
5.2. Results
While 5 of the 18 subjects performed relatively poorly com-
pared to the previous groups, missing 10–14% of target tri-
als, none were identified as outliers. There were no anticipa-
tion errors; overall, late or missing responses (RT>1000ms)
accounted for 0.4% of all trials and 5.3% of target trials.
The results for Experiment 4 are displayed in Fig. 6. For this
group, the main effect of congruence was not significant and the
main effect of the view was just outside significance (p=0.074)
withatrendtowardfasterresponsesinthefirstpersonview(8ms
mean difference, S.E.=3.8ms). However, there was a signifi-
cant interaction between the perspective view and congruency
(F(1,17)=9.935, p=0.006). As expected from Experiments 1
and 2, responses were faster in the first person congruent condi-
tionthaninthefirstpersonincongruentconditions(9.4msmean
difference, S.E.=4.6ms; one-tailed t-test, t=2.03, p=0.029).
Fig. 6. Experiment 4: average reaction times (±1S.E.M.) for discrimination of
the target images during performance of congruent (filled circles, solid line) or
incongruentactions(hollowcircle,dashedline).Theperspectiveofthedisplayed
images was either upright (“first person”, see Figs. 1 and 2) or inverted (“third
person”).
Thedifferencecausedbythefirstversusthirdpersonperspec-
tive view within the congruent condition was highly significant
(13.5ms, S.E.=4.55ms, two-tailed t-test, t=2.957, p=0.009),
whilethedifferencebetweenthefirstandthirdpersonincongru-
ent conditions was negligible (<1ms, S.E.=4.04ms, t=0.3).
As in the other experiments, there was a difference between
the two image series used (fist versus flip; 2×2×2 ANOVA
F(1,17)=8.149, p=0.011); however, the interaction of series,
view and congruency was not significant.
5.3. Discussion
Experiment 4 indicates that the previously observed RT
advantage for congruent movements is lost when the images are
putintothethirdpersonperspective.FromExperiments1and2,
weexpectedandfoundasignificantdifferencebetweenfirstper-
son congruent and incongruent conditions; that this difference
was smaller than in Experiment 1 (9.4ms versus 23.1ms) may
reflectthesmallersubjectgroup(n=18versus26),ofwhichsev-
eralperformedthetaskrelativelypoorly.Itisthereforeimportant
thatdespitethesefactors,thedifferencebetweenfirstpersonand
third person view was highly significant in the congruent con-
dition, while it was insignificant in the incongruent condition.
Experiment 4 result is also consistent with Experiment 2: the
congruency effect was seen only when discriminating the first
personhandimages,andnottheunrelatedchangeinfixationcue
shape. This suggests the forward model prediction is advanta-
geous only for discrimination of first person perspective images
of the congruent hand action.
6. Experiment 5: Realistic biological rendering
Kilner et al. (2003) suggested that the conflict caused by
visuo-motorincongruenceaffectedmovementperformanceonly
when the observed actor was a biological (human) agent, as no
conflict was seen when the copied actor was a robot arm. In the
same vein, imaging studies suggest separate processing of bio-
logicalandnon-biologicalrenderedactionimages(Peranietal.,
2001), and realistic and cartoon action sequences (Han, Jiang,
Humphreys, Zhou, & Cai, 2005). However, in Kilner’s study
the kinematics differed between biological and non-biological
agents, hence it is not clear whether the important congruency
difference was due to the actor’s kinematics or their biological
nature. Many studies of biological motion show that the kine-
matics are important, as biological motion is readily attributed
tomovinglights(Johansson,1973)aslongastheyobeythenor-
mal kinematic patterns (Ahlstrom, Blake, & Ahlstrom, 1997;
Grossman et al., 2000). Recent work by Aymoz and Viviani
(2004) has suggested that observation of biological agents per-
forming actions allows anticipation within visual processing
streams, and this effect was absent for non-biological agents.
Again, their conditions changed both the agency and the kine-
matic parameters of the observed actions, and moreover, their
task did not involve action performance, so involvement of the
motor planning system, while likely, was not obligatory.
Werepeatedthetwomainconditionsofcongruentandincon-
gruentactionsperformedduringasequentialimagepresentation

Page 8

R.C. Miall et al. / Neuropsychologia 44 (2006) 966–976
973
(as tested in Experiments 1, 2 and 4) with the observed images
rendered as photo-realistic biologically rendered hands, or as
non-biological boxes. The kinematics of the slowly displayed
sequences were however identical.
6.1. Methods
Ten subjects participated in the final experiment (six male,
four female); none had taken part in previous experiments; age
range was 21–60 years (median 21).
A2×2factorialwithin-subjectdesignwasusedwithonefac-
tor being the congruency between subjects’ own actions and the
main sequence displayed, and the other factor being the graphi-
calrenderingoftheimages(photo-realisticversusbox-rendered;
Fig. 1C). The main sequence was presented only in sequential
mode, as used in Experiment 2, showing a predictably cyclic
action. As before, the main sequence presented was one of two
hand actions (‘fist’ or ‘flip’). Again, the order of test presen-
tations was manipulated between participants to minimize any
order effects.
The task, apparatus, stimuli and the experimental situation
were the same as Experiments 1, 2 and 4 with the exception that
all images (main sequence and targets) used in the unrealistic
condition were rendered as a set of cuboids (using the Poser
“box-rendered” option; Fig. 1C).
6.2. Results
All subjects performed well; there were nine anticipation
errors:0.03%ofalltrialsandonly0.69%oftargettrials.Overall,
late or missing responses (RT>1000ms) accounted for 0.48%
of all trials and 4.6% of target trials.
The results for Experiment 5 are given in Fig. 7. For
this group, the main effect of congruence was significant
(ANOVA F(1,9)=10.114, p=0.011), with faster responses in
the congruent conditions, while the effect of photo-realistic
rendering was not significant (F(1,9)=0.3, p>0.5). There was
no significant interaction between the factors (F(1,9)=0.56,
p>0.4).
Fig. 7. Experiment 5: average reaction times (±1S.E.M.) for discrimination of
the target images during performance of congruent (filled circles, solid line) or
incongruent actions (hollow circle, dashed line). The rendering of the displayed
images was either photo-realistic or unrealistic (“box-rendered”, see Fig. 1C).
6.3. Discussion
This final experiment indicates that the visual representation
ofthebiologicalagentperformingtheobservedhandmovements
does not significantly affect the discrimination task, although
there was a weak trend toward a greater effect for the photo-
realistically rendered condition (Fig. 7). It replicates Experi-
ments 1, 2 and 4 by showing reaction time advantages when
thereiscongruencebetweentheobservedimagesandtheactions
being performed.
7. General discussion
We aimed to investigate whether the reaction time (RT) to
discriminate target images among a sequence of visual stim-
uli could be reduced by performing actions congruent with the
visual images of hands presented on the screen. We argue that if
this effect was seen, it would suggest that enhanced prediction
of the main sequence of images allowed more rapid discrimina-
tionoftheoddballtargets.RTswereindeedfasterinExperiment
1 when the main images were presented in sequence, allowing
the participants to predict the visual image series, but this effect
wassmall.RTswerealsofasterwhentheimageswerecongruent
with the hand action being performed. But, crucially, reaction
times were significantly shorter in the congruent and sequential
condition than in the other three, suggesting that performance
of the congruent action was the most important factor in pre-
dicting the image series, over and above the effect of an ordered
visualsequence.Thisadvantageforthecongruentconditionwas
alsoreplicatedinExperiments2,4and5,usingdifferentsubject
groups for each experiment.
Our second experiment indicates that the RT savings were
not due to reduced attentional load in the congruent condition,
as might be expected if controlling the hand action were easier
in that condition. There was no difference in subjects’ ability to
discriminate a change of the fixation cue, suggesting that equal
attention was available to the discrimination task, when per-
forming congruent and incongruent actions. Two caveats need
to be mentioned here regarding the fixation pointer discrimina-
tion task. First, the target stimuli are presented in a different
spatial location to the hand stimuli (see Fig. 1B); second, the
nature of the target stimulus (a pointer rather than the hand) dif-
fersconsiderablyfromthehandstimuli.Hence,onemightargue
that demands of this fixation task may be too far removed from
the original to provide a sensitive test of attentional load. How-
ever, all dual task experiments incorporate, by intention, wide
differences between the two tasks, and the attentional demand
of one task are seen in the second. Hence, we believe that if
attentional differences were important, they would show some
effect on the fixation task as well as on the hand discrimination
task.
The use of third person perspective stimuli (Experiment 4)
provides additional evidence against differences in attentional
loadcausingthereactiontimedifferenceswehaveobserved.The
third person discrimination task has identical task demands to
themain,firstpersontask.Furthermore,thestimuliareidentical
in all aspects to the first person task, except for perspective.

Page 9

974
R.C. Miall et al. / Neuropsychologia 44 (2006) 966–976
Given that humans have a preference for specular imitation, we
suggest that performing congruent movements in this condition
should be of equivalent difficulty, or even easier than in the first
person perspective task. Hence, attentional load should be equal
or possibly even lower. The lack of any priming while viewing
thethirdpersonperspectivestimulisuggeststhatthecongruency
advantageobservedinExperiment1isnotsimplyduetoreduced
attentional load in the congruent condition, thus freeing greater
attentional resources for the discrimination task.
However, our third experiment showed that, with the intro-
ductionofa500msintervalbetweensuccessivestimuluspresen-
tations, the reaction time savings were lost. Hence, the priming
effect of action-related predictions assumed to underlie these
performance differences appears to be short-lived and does not
span the 500ms inter-stimulus interval. This result is not easily
integrated with the idea of a forward model based on the on-
going action. Since the executed hand actions were continuous,
despite the intermittent visual display, we would expect that the
forward model predictions would also be continuous. Instead, it
appearsthatthelinkbetweenpredictionsbasedonmotorexecu-
tion and the visual discrimination processes is short-lived. One
possibility is that visual stimuli engage the motor–visual pre-
dictive system, so that the output of the forward model is used
by visual areas in a process time-locked to visual events. In this
case, the presentation of the blank screen for 500ms may act
to disengage the system. This suggestion of a visual priming
or engagement of a continuous motor prediction may be consis-
tentwiththe300–700mswindowforprimingeffectsreportedby
Vogtetal.(2003);theyfoundthattheprimingprocesswastime-
locked to onset of the visual priming stimulus. It suggests that
thevisualpredictionsfromtheforwardmodelareonlyavailable
in the context of visually relevant tasks.
However, it does not seem possible to explain the totality of
our results in terms of visuo-motor priming. The major result
seen in Experiments 1, 2, 4 and 5 – that congruency between
observedandexecutedactionsfacilitatesvisualdiscrimination–
canonlybeexplainedbyvisuo-motoreffectsifoneassumesthat
these visual–motor effects alter the neural resources available to
the discrimination task. This possibility has been discounted by
the results of Experiments 2 and 4, in which no evidence was
found for facilitated performance in the fixation task or the third
person perspective conditions.
Anotherimportantquestioniswhywefindafacilitatingeffect
ofactiononvisualdiscriminationwhereasHamiltonetal.(2004)
found a contrastive effect. Again, it may be because of the short
time scale of the integration between action and perception. In
their experiments, subjects lifted a weight some 1–2s before
observing a video clip of a similar action, and kept the weight
elevated until after the clip was finished. Hence, judgment of
the video clips was quite separate in time from the performed
lifting action. Additional experiments will be needed to better
understand these temporal relationships.
In sum, we interpret the results of these experiments as evi-
dence for the idea of motor–visual priming. They suggest that
the internal representation of an action during its motor execu-
tion can influence the visual system that would analyse visual
representations of the same action. The fourth experiment fur-
ther suggests that the observer viewpoint is important, and that
the predictive, internal representation of action aids discrimi-
nation of images seen in the first person perspective. It confers
advantageneitherindiscriminationofimagesthatareinthethird
person view, nor in the discrimination of changes in the fixation
cue.ThesimilarityofthesenullresultsinExperiment2(fixation
condition) and Experiment 4 (third person view condition) pro-
videsfurthersupportagainstaroleofdifferentialattentionalload
causing the RT savings. If congruence between observed and
executed actions allowed greater attention to the discrimination
task,wewouldalsoexpectthattohaveoccurredinthethirdper-
son condition, as the advantage of imitated actions can be found
for both first and third person perspectives (Brass et al., 2001).
These results are therefore consistent with the ideomotor
principle (Greenwald, 1970) or common event coding theory
(Hommel et al., 2001). According to the event coding theory
actions are coded in terms of the perceivable effects they gener-
ate. Since perception and anticipated action effects share the
same code, this approach predicts that perceived events can
prime or induce compatible actions. Indeed, Kilner et al. (2003)
have recently shown that observing another human making
incongruent movement has a significant interference effect on
theexecutionofmovement.Butthecommoncodingtheoryalso
predicts that action – even if just intended – could modulate per-
ceptual processing. These results are consistent with the report
by Aymoz and Viviani (2004). However, our results may go
further than predicted by the common event coding theory: that
theorywouldnotpredictthatthereactiontimesavingsshouldbe
specific to first person viewpoint, as the third person viewpoint
stimulialsoshareacommoncodewiththeplannedhandactions.
One unexpected result we found was the significant differ-
ence in response speeds between the two image types (fist and
flip). It may simply be that the oddball targets were easier to
discriminate in one series than the other, perhaps because some
unavoidable differences between the target and main images
aided their detection (Fig. 2). In an additional control experi-
ment,eightsubjectswereinstructedtoreacttothetargetimages,
without performing any hand actions. All four combinations of
target (OK or peace) and images series (fist or flip) were tested.
Thissubjectgroupshowedthesamepatternofresponsesasseen
in the main experiments—responses were significantly faster
fortargetsshownduringthefistsequencethantheflipsequence,
regardlessofthetargetimage.Therewasalsoasignificantinter-
action between target image and the series, further suggesting a
differenceinmatchingdifficulty.However,thesedifferencesare
secondary to the main effect seen and the analyses undertaken
that included image series as an additional factor showed the
effects of interest (congruency versus sequence, or congruency
versus viewpoint) remained statistically significant.
Some recent findings have provided a neural mechanism that
would allow a direct matching between the visual description
of an action and its execution. Recordings of single cell activity
in macque monkeys have shown that a subset of neurons in
the ventral premotor cortex – area F5 – (Di Pellegrino, Fadiga,
Fogassi, Gallese, & Rizzolatti, 1992; Gallese, Fadiga, Fogassi,
& Rizzolatti, 1996) and the anterior part of the inferior parietal
cortex–PForBrodmann’sarea7b–(Gallese,Fogassi,Fadiga,&

Page 10

R.C. Miall et al. / Neuropsychologia 44 (2006) 966–976
975
Rizzolatti,2002;Rizzolatti,Fogassi,&Gallese,2001)discharge
during execution of movement and when it observes another
individual performing the same or similar action. Mirror-like
neurons have also been reported in the superior temporal sulcus
(STS).Neuronsinthisregionrespondtoface,eyeandhandgoal
directed actions (Perrett et al., 1989). However, they discharge
onlyduringobservationofanactionandnotduringitsexecution.
It therefore seems that there are three areas in the primate
brain that contain neurons capable of forming a cortical action
observation system. Neurons similar to those discovered in
monkeys may also exist in humans. For example, transcra-
nial magnetic stimulation of the human motor cortex during
observation of hand movements selectively increases motor
evoked potentials in muscles normally used to perform the
observed hand action (Fadiga, Fogassi, Pavesi, & Rizzolatti,
1995;Gangitano,Mottaghy,&Pascual-Leone,2004).Themotor
cortex is therefore selectively potentiated during action obser-
vation. Results from other electrophysiological studies (Hari
et al., 1998; Strafella & Paus, 2000) support this conclusion,
althoughsomespinalchangesmayalsobeevoked(Vargasetal.,
2004).Furthermore,neuroimagingstudiesusingfMRI(Buccino
et al., 2001; Iacoboni et al., 1999; Iacoboni et al., 2001), or PET
(Grafton, Arbib, Fadiga, & Rizzolatti, 1996; Rizzolatti et al.,
1996) and event-related MEG (Nishitani & Hari, 2000) have
localised the neurons responsible for these effects in humans to
the ventral premotor and parietal cortices.
So how would action observation and forward modelling
interact? Carr, Iacoboni, Dubeau, Mazziotta, and Lenzi (2003)
propose that neurons in the STS, PF and F5 form an action
representation circuit. During action observation these connec-
tions form an inverse model, they claim, converting the visual
representation of actions in the STS into a motor plan in F5.
For imitation these connections would then act in reverse as a
forward model, converting an efferent copy of the motor plan
back into a predicted visual representation for comparison with
exemplar visual images (a sensory outcome of action). This
mechanismcouldprovidethebasisforunderstandingtheresults
of our experiments. When subjects prepare and execute their
movements they would activate neurons located in area F5, and
through the “forward” connections activate neurons located in
areas PF and STS. Thus, preparation of movement may evoke
an action plan in motor terms (involving F5 and PF mirror neu-
ron activity) and also in visual terms (invoking PF and STS
activity). These motor-evoked visual representations in STS or
perhapsotherextrastriateareaswouldthenfacilitatediscrimina-
tionoftargetimages,whichdonotcorrespondwiththepredicted
sequence (motor–visual priming).
Finally, given that the mirror neuron system has been related
to imitating and understanding the actions of others, it was
unclearwhetheronewouldexpectthepredictiverepresentations
tobecongruentwithfirstorthirdpersonpresentation(Craighero
etal.,2002;Vogtetal.,2003).Mostofthedataonmirrorneurons
has been collected in allocentric third person conditions. How-
ever, it has been proposed that when one perceives one’s own
actions, there is a closer match between the predicted and actual
outcomesoftheactionthanwhenweperceivetheactionsofoth-
ers(Knoblich,Seigerschmidt,Flach,&Prinz,2002).Therefore,
it could be expected that in motor–visual priming the first per-
sonperspectivewouldshowstrongereffectsthanthethirdperson
perspective, and this was indeed the case. Effects of congruent
image presentation were only found during the presentation of
stimuli in the first person perspective.
8. Conclusion
We have shown that action execution can assist the discrimi-
nation of visual images, when the image sequence is congruent
with the executed action, and when the images are shown in
the first person perspective. The link between these predictions
and visual processing appears to be short lasting. We interpret
our results as evidence supporting the hypothesis that forward
models in the motor system provide action-specific sensory pre-
dictions that are available to cognitive processes independent of
motor control.
Acknowledgements
ThisworkwassupportedbygrantsfromtheJamesSMcDon-
nell Foundation and the Wellcome Trust. Paul Wainman con-
ducted preliminary experiments that helped in the development
of these studies.
References
Ahlstrom, V., Blake, R., & Ahlstrom, U. (1997). Perception of biological
motion. Perception, 26, 1539–1548.
Aymoz, C., & Viviani, P. (2004). Perceptual asynchronies for biological and
non-biological visual events. Vision Research, 44, 1547–1563.
Brass, M., Bekkering, H., & Prinz, W. (2001). Movement observation affects
movement execution in a simple response task. Acta Psychology (Ams-
terdam), 106, 3–22.
Brass, M., Bekkering, H., Wohlschlager, A., & Prinz, W. (2000). Com-
patibility between observed and executed finger movements: Comparing
symbolic, spatial, and imitative cues. Brain and Cognition, 44, 124–143.
Buccino, G., Binkofski, F., Fink, G. R., Fadiga, L., Fogassi, L., Gallese, V., et
al. (2001). Action observation activates premotor and parietal areas in a
somatotopic manner: An fMRI study. European Journal of Neuroscience,
13, 400–404.
Carr, L., Iacoboni, M., Dubeau, M. C., Mazziotta, J. C., & Lenzi, G. L.
(2003). Neural mechanisms of empathy in humans: A relay from neu-
ral systems for imitation to limbic areas. Proceedings of the National
Academy of Sciences of the United States of America, 100, 5497–5502.
Craighero, L., Bello, A., Fadiga, L., & Rizzolatti, G. (2002). Hand action
preparation influences the responses to hand pictures. Neuropsychologia,
40, 492–502.
Craighero, L., Fadiga, L., Rizzolatti, G., & Umilta, C. (1999). Action for
perception: A motor–visual attentional effect. Journal of Experimental
Psychology: Human Perception and Performance, 25, 1673–1692.
Decety, J., & Grezes, J. (1999). Neural mechanisms subserving the perception
of human actions. Trends in Cognitive Science, 3, 172–178.
Di Pellegrino, G., Fadiga, L., Fogassi, L., Gallese, V., & Rizzolatti, G. (1992).
Understanding motor events: A neurophysiological study. Experimental
Brain Research, 91, 176–180.
Duhamel, J. R., Colby, C. L., & Goldberg, M. E. (1992). The updating
of the representation of visual space in parietal cortex by intended eye
movements. Science, 255, 90–92.
Fadiga, L., Fogassi, L., Pavesi, G., & Rizzolatti, G. (1995). Motor facilita-
tion during action observation: A magnetic stimulation study. Journal of
Neurophysiology, 73, 2608–2611.

Page 11

976
R.C. Miall et al. / Neuropsychologia 44 (2006) 966–976
Frak, V., Paulignan, Y., & Jeannerod, M. (2001). Orientation of the opposition
axis in mentally simulated grasping. Experimental Brain Research, 136,
120–127.
Frith, C. D., Blakemore, S. J., & Wolpert, D. M. (2000). Abnormalities in
the awareness and control of action. Philosophical Transactions of the
Royal Society of London. Series B: Biological Sciences, 355, 1771–1788.
Gallese, V., Fadiga, L., Fogassi, L., & Rizzolatti, G. (1996). Action recogni-
tion in the premotor cortex. Brain, 119(Pt 2), 593–609.
Gallese, V., Fogassi, L., Fadiga, L., & Rizzolatti, G. (2002). Action represen-
tation and the inferior parietal lobule. In W. Prinz & B. Hommel (Eds.),
Attention and performance: vol. XIX (pp. 247–266). Oxford University
Press.
Gangitano, M., Mottaghy, F. M., & Pascual-Leone, A. (2004). Modulation
of premotor mirror neuron activity during observation of unpredictable
grasping movements. European Journal of Neuroscience, 20, 2193–2202.
Gentilucci, M., Daprati, E., & Gangitano, M. (1998). Right-handers and left-
handers have different representations of their own hand. Brain Research.
Cognitive Brain Research, 6, 185–192.
Grafton, S. T., Arbib, M. A., Fadiga, L., & Rizzolatti, G. (1996). Localization
of grasp representations in humans by positron emission tomography. 2.
Observation compared with imagination. Experimental Brain Research,
112, 103–111.
Greenwald, A. G. (1970). Sensory feedback mechanisms in performance con-
trol: With special reference to the ideo-motor mechanism. Psychological
Reviews, 77, 73–99.
Grossman, E., Donnelly, M., Price, R., Pickens, D., Morgan, V., Neighbor,
G., et al. (2000). Brain areas involved in perception of biological motion.
Journal of Cognitive Neuroscience, 12, 711–720.
Hamilton, A., Wolpert, D., & Frith, U. (2004). Your own action influences
how you perceive another person’s action. Current Biology, 14, 493–498.
Han, S., Jiang, Y., Humphreys, G. W., Zhou, T., & Cai, P. (2005). Distinct
neural substrates for the perception of real and virtual visual worlds.
Neuroimage, 24, 928–935.
Hari, R., Forss, N., Avikainen, S., Kirveskari, E., Salenius, S., & Rizzolatti,
G. (1998). Activation of human primary motor cortex during action obser-
vation: A neuromagnetic study. Proceedings of the National Academy of
Sciences of the United States of America, 95, 15061–15065.
Haruno, M., Wolpert, D. M., & Kawato, M. (2001). Mosaic model for sen-
sorimotor learning and control. Neural Computation, 13, 2201–2220.
Hommel, B., Musseler, J., Aschersleben, G., & Prinz, W. (2001). The Theory
of Event Coding (TEC): A framework for perception and action planning.
Behaviour and Brain Science, 24, 849–878.
Iacoboni, M., Koski, L. M., Brass, M., Bekkering, H., Woods, R. P., Dubeau,
M. C., et al. (2001). Reafferent copies of imitated actions in the right
superior temporal cortex. Proceedings of the National Academy of Sci-
ences of the United States of America, 98, 13995–13999.
Iacoboni, M., Woods, R. P., Brass, M., Bekkering, H., Mazziotta, J. C., &
Rizzolatti, G. (1999). Cortical mechanisms of human imitation. Science,
286, 2526–2528.
Jeannerod, M., & Decety, J. (1995). Mental motor imagery: A window into
the representational stages of action. Current Opinion in Neurobiology,
5, 727–732.
Johansson, G. (1973). Visual perception of biological motion and a model
for its analysis. Perception and Psychophysics, 14, 201–211.
Jordan, M. I. (1995). Computational motor control. In M. Gazzaniga (Ed.),
The cognitive neurosciences (pp. 597–609). Cambridge, Mass: MIT Press.
Jordan, M. I., & Rumelhart, D. E. (1992). Forward models: Supervised learn-
ing with a distal teacher. Cognitive Science, 16, 307–354.
Kawato, M. (1999). Internal models for motor control and trajectory planning.
Current Opinions in Neurobiology, 9, 718–727.
Kawato, M., Furukawa, K., & Suzuki, R. (1987). A hierarchical neural net-
work model for control and learning of voluntary movement. Biological
Cybernetics, 57, 169–186.
Kilner, J. M., Paulignan, Y., & Blakemore, S. J. (2003). An interference
effect of observed biological movement on action. Current Biology, 13,
522–525.
Knoblich, G., Seigerschmidt, E., Flach, R., & Prinz, W. (2002). Authorship
effects in the prediction of handwriting strokes: Evidence for action sim-
ulation during action perception. The Quarterly Journal of Experimental
Psychology A, 55, 1027–1046.
Miall, R. C., & Wolpert, D. M. (1996). Forward models for physiological
motor control. Neural Networks, 9, 1265–1279.
Musseler, J., & Hommel, B. (1997). Blindness to response-compatible stimuli.
Journal of Experimental Psychology. Human Perception and Performance,
23, 861–872.
Musseler, J., Steininger, S., & Wuhr, P. (2001). Can actions affect perceptual
processing. The Quarterly Journal of Experimental Psychology A, 54,
137–154.
Nishitani, N., & Hari, R. (2000). Temporal dynamics of cortical representation
for action. Proceedings of the National Academy of Sciences of the United
States of America, 97, 913–918.
Parsons, L. M. (1994). Temporal and kinematic properties of motor behavior
reflected in mentally simulated action. Journal of Experimental Psychol-
ogy. Human Perception and Performance, 20, 709–730.
Parsons, L. M., Fox, P. T., Hunter Downs, J., Glass, T., Hirsch, T. B., Martin,
C. C., et al. (1995). Use of implicit motor imagery for visual shape
discrimination as revealed by PET. Nature, 375, 54–58.
Perani, D., Fazio, F., Borghese, N. A., Tettamanti, M., Ferrari, S., Decety,
J., et al. (2001). Different brain correlates for watching real and virtual
hand actions. Neuroimage, 14, 749–758.
Perrett, D. I., Harries, M. H., Bevan, R., Thomas, S., Benson, P. J., Mistlin,
A. J., et al. (1989). Frameworks of analysis for the neural representation
of animate objects and actions. Journal of Experimental Biology, 146,
87–113.
Poulton, E. C. (1974). Tracking skill and manual control. London, NY: Aca-
demic Press.
Rizzolatti, G., Fadiga, L., Matelli, M., Bettinardi, V., Paulesu, E., Perani,
D., et al. (1996). Localization of grasp representations in humans by
PET: 1. Obsevation versus execution. Experimental Brain Research, 111,
246–252.
Rizzolatti, G., Fogassi, L., & Gallese, V. (2001). Neurophysiological mech-
anisms underlying the understanding and imitation of action. Nature
Reviews. Neuroscience, 2, 661–670.
Ross, J., Morrone, M. C., Goldberg, M. E., & Burr, D. C. (2001). Changes
in visual perception at the time of saccades. Trends in Neurosciences, 24,
113–121.
Smith, D. T., Rorden, C., & Jackson, S. R. (2004). Exogenous orienting of
attention depends upon the ability to execute eye movements. Current
Biology, 14, 792–795.
Strafella, A. P., & Paus, T. (2000). Modulation of cortical excitability during
action observation: A transcranial magnetic stimulation study. Neurore-
port, 11, 2289–2292.
Vargas, C. D., Olivier, E., Craighero, L., Fadiga, L., Duhamel, J. R., & Sirigu,
A. (2004). The influence of hand posture on corticospinal excitability dur-
ing motor imagery: A transcranial magnetic stimulation study. Cerebral
Cortex, 14, 1200–1206.
Vogt, S., Taylor, P., & Hopkins, B. (2003). Visuomotor priming by pic-
tures of hand postures: Perspective matters. Neuropsychologia, 41, 941–
951.
Wolpert, D. M., & Ghahramani, Z. (2000). Computational principles of move-
ment neuroscience. Nature Neuroscience, (Suppl. 3), 1212–1217.
Wolpert, D. M., Ghahramani, Z., & Jordan, M. I. (1995). An internal model
for sensorimotor control. Science, 269, 1880–1882.
Yarrow, K., Haggard, P., Heal, R., Brown, P., & Rothwell, J. C. (2001).
Illusory perceptions of space and time preserve cross-saccadic perceptual
continuity. Nature, 414, 302–305.