Brain and Cognition 44, 124–143 (2000)

doi:10.1006/brcg.2000.1225, available online at http://www.idealibrary.com on

Compatibility between Observed and Executed

Finger Movements: Comparing Symbolic,
Spatial, and Imitative Cues

Marcel Brass,* Harold Bekkering,† Andreas Wohlschläger,† and

Wolfgang Prinz†
*Department of Neurology, Max Planck Institute of Cognitive Neuroscience, Leipzig,
Germany; and †Department of Cognition and Action, Max Planck Institute of
Psychological Research, Munich, Germany

Published online August 15, 2000

Intuitively, one can assume that imitating a movement is an easier task than re-
sponding to a symbolic stimulus like a verbal instruction. Support for this suggestion
can be found in neuropsychological research as well as in research on stimulus–
response compatibility. However controlled experimental evidence for this assump-
tion is still lacking. We used a stimulus–response compatibility paradigm to test
the assumption. In a series of experiments, it was tested whether observed finger
movements have a stronger influence on finger movement execution than a symbolic
or spatial cue. In the first experiment, we compared symbolic cues with observed
finger movements using an interference paradigm. Observing finger movements
strongly influenced movement execution, irrespective of whether the finger move-
ment was the relevant or the irrelevant stimulus dimension. In the second experi-
ment, effects of observed finger movements and spatial finger cues were compared.
The observed finger movement dominated the spatial finger cue. A reduction in the
similarity of observed and executed action in the third experiment led to a decrease
of the influence of observed finger movement, which demonstrates the crucial role
of the imitative relation of observed and executed action for the described effects.
The results are discussed in relation to recent models of stimulus–response compati-
bility. Neurocognitive support for the strong relationship between movement obser-
vation and movement execution is reported.  2000 Academic Press

I. INTRODUCTION
There are many possible ways to instruct an action. For example, if you
wanted to teach a child how to type the letter ‘‘t’’ on a computer keyboard,
you could provide a verbal command such as ‘‘press the sixth button of the

Address correspondence and reprint requests to Marcel Brass, Max Planck Institute of Cog-
nitive Neuroscience, Dept. Neurology, Stephanstr. 1A, D-4103 Leipzig, Germany. E-mail:
brass@cns.mpg.de. Fax: (49) 341 9940-221.
124
0278-2626/00 $35.00
Copyright  2000 by Academic Press
All rights of reproduction in any form reserved.
 IDEOMOTOR COMPATIBILITY 125

second row with the index finger of your left hand.’’ You could also some-
how mark the button and the finger to be used. A third possibility would be
to demonstrate the typing; that is, you could show the required hand position
and the finger pressing on the ‘‘t.’’ Intuitively, one would assume that the
third instruction would be most easily interpreted by the child, as it could
see what it is being asked to do.
Both, neuropsychological as well as cognitive research support the as-
sumption that there are different mechanisms mediating action imitation and
mediating responses to symbolic instruction. However controlled experimen-
tal evidence for such a distinction in normal subjects is still missing. In this
study the assumption of a privileged path from movement observation to
movement execution was tested using a stimulus–response compatibility
paradigm.

Imitation in Neuropsychological Research

From neuropsychological studies it is well known that apraxic patients
are selectively disturbed for different command modalities (e.g., Schnider,
Hanlon, Alexander, & Benson, 1997). On the one hand patients with ideomo-
tor apraxia typically make the most errors on pantomime to verbal command,
whereas imitation performance is less impaired or remains at least equally
impaired (De Renzi, Motti, & Nichelli, 1980; Heilman & Rothi, 1985; Alex-
ander et al., 1992). On the other hand, few cases with so-called conduction
apraxia have been described where the patient had problems imitating a ges-
ture while being less impaired on verbal command (Ochipa, Rothi, & Heil-
man, 1994; Mehler, 1987). These distinct error patterns led some authors
to suggest that two different routes mediate imitative actions and verbally
instructed actions (Rothi, Ochipa, & Heilman, 1991; Roy & Hall, 1992).
While in imitation the model provides some information that can be used for
specifying the response, executing the same action on verbal command requires
the internal generation of this information (Roy & Hall, 1992). The special
role of imitative action becomes also obvious in patients showing compulsive
imitative behavior (Lhermitte, Pillon, & Serdaru, 1986). These patients cannot
resist from imitating observed movements presented in front of them. These
results support the view that movement observation leads to an activation of
the corresponding response, which is typically inhibited in normal subjects.
But what are the cognitive mechanisms underlying imitative action?

A Functional Mechanism Underlying Action Imitation

Different cognitive theories have emphasized the tight relationship be-
tween perception and action (e.g., Gibson, 1986; Prinz, 1990). A very promi-
nent approach emphasizing this point is the ecological theory (Gibson, 1986).
Central to this approach is the concept of affordance, which assumes that the
perception of the external world always implies information about potential
126 BRASS ET AL.

related actions. Recently, Tucker and Ellis (1998) used a stimulus–response

compatibility design to demonstrate that seen objects automatically potenti-
ate components of the actions they afford. They referred to a modified ver-
sion of the affordance concept introduced by Jeannerod (1994) to account
for their results. In addition a study by Craighero et al. (1996, 1998) demon-
strated that presentation of an object primes a response related to that object.
While the affordance concept seems to be a reasonable account for response
activation by visual presented objects, it is difficult to apply this concept to
imitation of actions that are not object related. For example, observing some-
one pointing in a specific direction typically does not afford a similar hand
gesture. Rather it provokes looking in the pointed direction.
An alternative view of imitative phenomena is provided by the theory
of ideomotor action (Greenwald, 1970a, 1970b). Greenwald argued that the
anticipatory representation of an action’s sensory feedback (a response im-
age) is used for action control. A key press, for example, is controlled by
a representation of its sensory consequence (the visual impression of the
movement or the tactile sensation of touching the key). When viewed from
this perspective, the mechanism underlying action imitation can be formu-
lated as follows: If a perceptual event is similar to the response image that
is used to control a response, then the perceived event should activate the
response image and, hence, influence the initiation of the response. This de-
scription of imitative action is rather consistent with the account presented
by Roy and Hall (1992), drawn from research on apraxia.
By introducing the concept of ideomotor compatibility, which Greenwald
defined as: ‘‘[The] dimension denoting the extent to which a stimulus corre-
sponds to sensory feedback from its required response’’ (Greenwald, 1970b,
p. 52), Greenwald extended this idea to compatibility phenomena as well. The
ideomotor theory predicts that the ease with which a stimulus can be trans-
formed into an action depends on the similarity between the observed event
and the executed action. In situations involving a verbal or a symbolic instruc-
tion, this similarity is minimal, whereas in situations involving an observed
action that matches the required action, this similarity is maximal. To test
this assumption we have used a stimulus–response compatibility paradigm.

Investigating Ideomotor Compatibility with a Stimulus–Response

Compatibility Paradigm
In cognitive psychology experimental studies on functional aspects of imi-
tation are rare. It is the field of stimulus–response compatibility (SRC) which
provides deeper insights into the question how stimulus properties influence
the selection and initiation of a response. SRC refers to the finding that hu-
man performance is not only affected by characteristics of the stimulus set
and the response set used in a task, but also by the combination of the sets
(e.g., Kornblum, Hasbroucq, & Osman, 1990; Hommel & Prinz, 1997). The
term SRC originally refers to the observation that a compatible mapping of
 IDEOMOTOR COMPATIBILITY 127

stimulus and response position leads to faster responses than an incompatible

mapping (Fitts & Deininger, 1954, Fitts & Seeger, 1953). Most recent models
of SRC focus on the similarity between stimulus and response to explain
SRC effects (Kornblum, Hasbroucq, & Osman, 1990; Hommel, Müsseler,
Aschersleben, & Prinz, 1999). Thus it is surprising that only very few com-
patibility studies have investigated the question how movement observation
affects movement execution (Kerzel & Bekkering, in press; Stürmer, Ascher-
sleben, & Prinz, in press; Brass, Bekkering, & Prinz, in press). However, these
studies did not compare compatibility effects for different types of stimuli.
The present study was designed to address this issue by directly comparing
ideomotor-compatible and more symbolic stimuli, using an interference par-
adigm similar to Kornblum and Lee (1995). In this paradigm, three different
conditions are contrasted: baseline, congruent, and incongruent conditions.
A baseline condition established the standard reaction time (RT) for a single,
relevant, imperative stimulus dimension. In the Kornblum and Lee study,
for example, the letter ‘‘J’’ appeared as the imperative stimulus between the
two hands (e.g., the ‘‘J’’ indicated a key press of the right middle finger).
In the congruent condition, the imperative stimulus (e.g., the letter ‘‘J’’) was
shown in a congruent relation to a response-irrelevant stimulus dimension
(e.g., the ‘‘J’’ appeared on the fingernail of the right middle finger). In the
incongruent condition, however, the imperative stimulus appeared at a differ-
ent, incongruent location (e.g., ‘‘J’’ appeared on the left index finger). The
introduction of a baseline condition enabled the evaluation of the irrelevant
stimulus information’s function in the congruent trials (i.e., whether the irrel-
evant stimulus information facilitated response initiation). It also enabled the
evaluation of the irrelevant dimension’s function in the incongruent trials
(i.e., whether the irrelevant stimulus dimension interferes with response initi-
ation). The exchange of the relevant and the irrelevant stimulus dimensions
makes it possible to directly compare the influence of both stimulus dimen-
sions.
Three experiments were conducted using this paradigm in order to com-
pare ideomotor-compatible with symbolic stimuli. In Experiment 1, highly
ideomotor-compatible videotaped finger movements (the lifting of the index
or the middle finger) were contrasted with symbolic cues (the number ‘‘1’’
or ‘‘2’’). Experiment 2 contrasted the videotaped finger movements with
spatial finger cues (an ‘‘x’’ on the fingernail of the index or middle finger).
In Experiment 3, the same stimuli as in Experiment 2 were presented; but
responses (tapping with the index or middle finger) were no longer
ideomotor-compatible to the observed finger movement (lifting of the index
or middle finger).

EXPERIMENT 1
Two types of stimulus dimensions were presented in the first experiment:
finger movements (lifting movement of index or middle finger) and symbolic
128 BRASS ET AL.

cues (the digits 1 and 2). Participants had to execute a finger movement in
response to either the finger movements or the symbolic cues (1 required an
index-finger movement, whereas 2 required a middle finger movement). As
in a Stroop and Simon task, a relevant dimension was compared with an
irrelevant dimension. Expressed in terms of the ideomotor theory, switching
from the symbolic-cues to the finger movements as the relevant dimension
should lead to a RT advantage. In addition, in the symbolic cue condition,
the irrelevant congruent movement is expected to exert a facilitative influ-
ence. That is, observing a congruent finger movement should activate the
congruent response and, hence, should facilitate response initiation. Observ-
ing an incongruent finger movement should activate the incorrect response
and, therefore, should lead to an interference effect. On the other hand, the
finger movement condition was expected to be unaffected by the congruency
of the symbolic cues.
Method
Participants
Eight students of the University of Munich participated in a 30-min-long session. All were
right-handed, had normal or corrected-to-normal vision, and were naive with respect to the
purpose of the study. They were paid for their participation.

Apparatus and Material

A Compaq Pentium 200-MHz computer controlled stimulus presentation and data acquisi-
tion. Short, digitized video sequences of an animated hand were presented on a 17-in. screen
with a frame frequency of 60 Hz. The video sequences consisted of five frames. The animated
hand had a visual angle of approximately 6.5° vertically and 10° horizontally. Participants
were seated approximately 80 cm from the screen. The animated hand was positioned as to
appear as a mirror image of the participant’s hand, which rested on a table. Movements were
recorded by an Optotrak system (Norton Digital). Six cameras were installed to sample the
raw position data of active infrared markers (with a sampling rate of 250 Hz) located on the
fingernails of the subject’s index and middle fingers. The system allows for a reconstruction
of a movement’s trajectory in three-dimensional space. Movement onsets were computed
based on these trajectories and analyzed off-line, using MATLAB 5.0 routines. For detection
of the movement onset, a relative velocity threshold was used (10% of peak velocity). Move-
ment onset was defined as the moment in which the velocity exceeded this threshold and after
which a minimum displacement (10% of movement amplitude) took place within a maximum
amount of time (100 ms).

Design and Procedure

The experiment was divided into two blocks: one block with finger movements as the rele-
vant dimension (finger movement condition) and one block with symbolic cues as the relevant
dimension (symbolic cue condition). The finger-movement condition required that the partici-
pants execute an index-finger movement in response to an observed index-finger movement
and a middle-finger movement in response to an observed middle-finger movement. Three
stimulus events were randomly (with a maximum of three stimulus-event repetitions) presented
 IDEOMOTOR COMPATIBILITY 129

baseline

congruent incongruent

FIG. 1. One of the two possible stimulus events that appeared in baseline, congruent, and
incongruent trials (only the last frame of the video sequence is displayed). The baseline is
different in the symbolic-cue condition (right picture) and the finger-movement condition (left
picture). Stimuli for congruent and incongruent trials were identical for both conditions.

in this block (see Fig. 1). In the baseline trials, only a finger movement was presented. In the
congruent trials, a congruent number—‘‘1’’ in the case of an index-finger movement and ‘‘2’’
in the case of a middle-finger movement—was displayed between the index and the middle
fingers. In the incongruent trials, an incongruent number—‘‘1’’ in the case of a middle-finger
movement and ‘‘2’’ in the case of an index-finger movement—was displayed. In the symbolic
cue condition (see Fig. 1), participants were instructed to execute an index-finger movement
in response to the presentation of the number 1 or a middle-finger movement in response to
the presentation of the number 2. In the baseline trials, only the numbers were presented. In
the congruent trials, a congruent finger movement accompanied the number; in incongruent
trials, an incongruent finger movement accompanied the presented number.
The video sequences began with a frame showing the hand in a resting position. This initial
frame, which was identical for all stimuli, remained visible for about 560 ms. Then, three
frames (each lasting approximately 34 ms) presented the finger movement, the numbers, or
both. Finally, a fifth frame was presented for approximately 500 ms. The overall displacement
of the presented finger movement was about 2°. Between the trials, the screen turned blue
for 1840 ms; hence, the total trial length was 3 s. Both blocks contained 120 trials (40 baseline
trials, 40 congruent trials, and 40 incongruent trials). Block order was counterbalanced across
participants. Participants were instructed to execute their movement as quickly as possible.
Before each block, 10 test trials were presented. Test trials were excluded from further analysis.
130 BRASS ET AL.

Data Analysis
Errors. Incorrect movements were defined to be errors when the amplitude was found to
be larger than 1/3 of the instructed finger movement. The mean error rate was 2.8%.
Response Times. Errors were excluded from the RT analysis. An ANOVA was computed
with two within-subject factors, using type of instruction (finger movement condition vs
symbolic-cue condition) as the first factor and congruency (baseline, congruent, and incongru-
ent) as the second factor.

Results
Error analysis revealed no significant results. A robust main effect (91
ms) was, however, found for the type of instruction, F(1, 7) ⫽ 92.9, p ⬍
.001. Participants generally reacted more quickly to the finger movement
then when they reacted to a symbolic cue (273 ms versus 364 ms). Another
main effect was found, this time regarding congruency: F(2, 14) ⫽ 27.3,
p ⬍ .001. In other words, participants’ reactions in congruent trials (304 ms)
were faster, t(7) ⫽ 3.5, p ⬍ .05, than in baseline trials (318 ms). Participants’
reactions were, in turn, faster in the baseline trials, t(7) ⫽ 6.1, p ⬍ .001,
than in the incongruent trials (333 ms). The most interesting effect was found
in the two-way interaction between type of instruction and congruency, F(2,
14) ⫽ 19.9, p ⬍ .001 (see Fig. 2).
Congruency affected reaction times in the symbolic-cue condition (con-

420

symbolic cue condition

400

380
movement onset (ms)

360

340

320

300
finger movement condition

280

260
congruent baseline incongruent

congruency

FIG. 2. Movement onset (RT) as a function of congruency (congruent, baseline, and in-
congruent) and type of instruction (finger-movement condition and symbolic cue condition).
In the symbolic-cue condition the observed finger movement was the irrelevant dimension,
while in the finger-movement condition the symbolic cue was the irrelevant dimension.
 IDEOMOTOR COMPATIBILITY 131

gruent: 336 ms, baseline: 363 ms, incongruent: 392 ms) but not in the finger-
movement condition (congruent: 272 ms, baseline: 274 ms, incongruent: 274
ms). If subjects responded to the symbolic cue, the irrelevant-congruent fin-
ger movement led to a significant facilitation effect of 27 ms, t(7) ⫽ 3.93,
p ⬍ .01. In addition, the irrelevant-incongruent finger movement led to a
significant interference effect of 29 ms, t(7) ⫽ 4.83, p ⬍ .012.

Discussion
The main effect found for type of instruction indicated that responses were
more quickly initiated in the finger-movement condition than in the sym-
bolic-cue condition. In addition, responses to the symbolic cues were facili-
tated when the irrelevant finger movement was congruent. An interference
effect was found when the irrelevant finger movement was incongruent. Both
the facilitation and the interference effects provided evidence for a strong
and automatic influence of observed finger movements on movement execu-
tion. On the other hand, the presentation of the irrelevant symbolic cue did
not influence the response in the finger-movement condition.
Although these results are quite consistent with our hypothesis, similar
results already have been obtained for a comparison between symbolic cues
similar to the ones used in Experiment 1 and spatial finger cues (e.g., Korn-
blum & Lee, 1995). However, the ideomotor theory predicts that the ob-
served finger movement should be more ideomotor-compatible to the re-
sponse than a spatial finger cue. We tested this assumption by comparing
the finger-movement condition with a spatial finger-cue condition.

EXPERIMENT 2
In Experiment 2 the symbolic cues were replaced by spatial finger cues
(see Fig. 3). The finger was cued by an ‘‘x’’ drawn on the fingernail of
either the index or the middle finger. Again, both stimulus dimensions were
contrasted in an interference paradigm in which each dimension was relevant
in one block and irrelevant in the other block. The ideomotor theory would
predict that the finger movement be more similar to the sensory feedback
of its required response and, hence, more ideomotor-compatible to the re-
sponse than the spatial finger cue. Thus, the presentation of finger movement
should affect response initiation more than the presentation of a spatial finger
cue. As a consequence, faster responses are expected in the case that the
finger movement is the relevant dimension than in the case that the spatial
finger cue is the relevant dimension. In addition, the theory predicts that
the irrelevant finger movement should affect performance in the finger-cue
condition. In other words, congruent-irrelevant finger movements should fa-
cilitate response initiation, whereas incongruent-irrelevant finger movements
should slow down response initiation in the spatial finger cue condition.
132 BRASS ET AL.

baseline

congruent incongruent

FIG. 3. One of two possible stimulus events that appeared in baseline, congruent, and
incongruent trials (only the last frame of the video sequence is displayed). The baseline in
the finger-movement condition was identical to Experiment 1 (left picture), while in the finger-
cue condition it is a cross on the fingernail (right picture). Congruent and incongruent trials
were again identical in both conditions.

Method
Participants
Eight students of the University of Munich participated in a 30-min session. All participants
were right-handed, had normal or corrected to normal vision, and were naive with respect to
the purpose of the study. They were paid for their participation.

Apparatus and Material

Stimulus presentation and measurement procedure was identical to that in Experiment 1.
The finger cue consisted of a black ‘‘x’’ which was drawn on a fingernail of either the index
or the middle finger. In the baseline trials of the spatial finger-cue condition, the cross appeared
on a fingernail, but the hand appeared motionless. In the congruent trials, the cross appeared
on the moving finger, in the incongruent trials the cross appeared on the nonmoving finger.

Design and Procedure

The experiment was again divided into two blocks. In one block (spatial finger-cue condi-
tion), participants were required to lift the finger equivalent to the one marked by the cross.
 IDEOMOTOR COMPATIBILITY 133

In the other block (finger-movement condition) participants had to respond to the observed
finger movement, as they did in Experiment 1. The temporal order of the digitized video
sequences was identical to Experiment 1. In the second frame (the first frame in which move-
ment was displayed), the cross appeared. As in Experiment 1, 10 test trials were performed
before each block. Test trials were excluded from further analysis.

Data Analysis
Errors. The mean error rate was 2.18 %.
Response times. Errors were excluded from further RT analysis. An ANOVA was computed
using the factors type of instruction (finger-movement condition vs spatial finger-cue condi-
tion) and congruency (baseline, congruent, and incongruent).

Results
Error analysis revealed only a significant main effect for congruency, F(2,
14) ⫽ 8.20, p ⬍ .01, which was in the same direction as the RT effect,
indicating no speed-accuracy trade-off. A main effect was also found for the
type of instruction, F(1, 7) ⫽ 118.9, p ⬍ .001. When participants responded
to the finger movement they were faster than with the spatial finger-cue con-
dition (277 ms versus 318 ms). In addition a main effect was found for con-
gruency, F(2, 14) ⫽ 46.1, p ⬍ .001. In other words, participants responded
more quickly in congruent trials (276 ms), t(7) ⫽ 2.6, p ⬍ .05 than in baseline
trials (283 ms). Participants’ reactions in baseline trials were, in turn, faster,
t(7) ⫽ 7.1, p ⬍ .001, than in incongruent trials (334 ms). Again, the interac-
tion (Fig. 4) between the type of instruction and congruency was significant,
F(2, 14) ⫽ 36.1, p ⬍. 001.
If subjects were instructed to respond to the spatial finger cue, an irrelevant
congruent finger movement led to a significant facilitation effect of 22 ms,
t(7) ⫽ 4.2, p ⬍ .01. In addition, the irrelevant-incongruent finger movement
led to a significant interference effect of 70 ms, t( 7) ⫽ 7.1, p ⬍ .001. If
participants were instructed to respond to the finger movement, a congruent
spatial finger cue had no influence on response time, but an incongruent
spatial finger cue led to an interference effect of 33 ms, t(7) ⫽ 4.4, p ⬍ .01.
Discussion
The results of Experiment 2 demonstrated that the ideomotor-compatible
finger movements dominated the less ideomotor-compatible spatial finger
cues. Again, it was found that participants were faster in the finger-movement
condition than in the spatial finger cue condition. Responses to the spatial
finger cues were facilitated when the finger movement was congruent; an
interference effect was found when the finger movement was incongruent.
Both the facilitation and the interference effects described above provided
evidence for a strong influence of movement observation on movement exe-
cution. It is interesting to note that an interference effect—but not a facilita-
tive effect—was evident within the finger-movement condition when the
134 BRASS ET AL.

spatial finger cue condition

380

360
movement onset (ms)

340

320

finger movement condition

300

280

260

240
congruent baseline incongruent

congruency

FIG. 4. RT as a function of congruency (congruent, baseline, and incongruent) and type

of instruction (finger-movement condition, and spatial finger cue condition). In the finger-cue
condition the observed finger movement was the irrelevant dimension, while in the finger-
movement condition the finger cue was the irrelevant dimension.

spatial finger cue was used as the irrelevant stimulus dimension. In other
words, a congruent spatial finger cue provides no additional benefit for re-
sponse initiation. An incongruent spatial finger cue, however, appears to neg-
atively influence response initiation processes. These results are consistent
with predictions derived from the ideomotor theory. In particular, they indi-
cate that the stimulus’ similarity to the sensory feedback from the response
determines the degree of compatibility.
Thus far, we have assumed that observed finger movements’ strong influ-
ence on movement execution could be attributed to the similarity of observed
and executed actions (i.e., the degree of ideomotor compatibility). This as-
sumption was, however, not explicitly tested in Experiments 1 and 2. Since
the executed action was always kept constant, one might as well explain the
findings with a pure perceptual account. Such a perceptual account would
state that the observed finger movement is processed faster than the symbolic
and spatial cues. In other words, not the relationship between observed and
executed actions was causing the effects found, but the results of Experi-
ments 1 and 2 can also be described at a perceptual level only. In Experiment
3 these two predictions are directly tested against each other. In order to do
so, the stimuli used in Experiment 3 were identical to the ones used in Experi-
ment 2. However, the finger movement required did not match the finger
movement observed. If the similarity between observed and executed finger
 IDEOMOTOR COMPATIBILITY 135

movements underlie the effects observed so far, a substantial change in the

pattern of results should occur. However, if the effects were caused by a
perceptual difference only, similar results as in Experiment 2 should occur.

EXPERIMENT 3
In Experiment 3, the exact same stimuli were used as in Experiment 2
(lifting finger movements and an ‘‘x’’ that appeared on the fingernail). The
participants’ required response to these stimuli was, however, different. In
contrast to the first two experiments, in which participants were required to
raise their index or middle fingers, respectively, participants now had to initi-
ate a tapping movement with the index or the middle finger. This manipula-
tion reduced the degree of ideomotor compatibility with the observed finger
movement and therefore a substantial change was expected in the pattern of
results. The facilitation effect—which had been caused by the ideomotor-
compatible irrelevant movement—should disappear. Furthermore, the inter-
ference effect should decrease, and the RT difference of the finger-movement
condition and the spatial finger-cue condition should decrease. The percep-
tual interpretation, on the other hand, predicts no substantial change in the
pattern of results.
Method
Participants
Eight students of the University of Munich participated in a 30-min session. All were right-
handed, had normal or corrected-to-normal vision, and were naive with respect to the purpose
of the study. They were paid for their participation.

Apparatus and Material

The stimulus presentation and measurement procedure was identical to Experiment 2. How-
ever, participants now were instructed to hold their index and middle fingers a few centimeters
above the surface of the table and to rest their ring and the little fingers on the table.

Design and Procedure

The experiment was divided into two blocks. In one block (spatial finger-cue condition)
the instruction was to watch for the finger with an ‘‘x’’ and to tap on the table with this finger.
In the other block (finger-movement condition), subjects were instructed to tap on the table
with the finger that was equivalent to the lifted finger on the screen.

Data Analysis
Errors. The mean error rate was 2.5%.

Response times. Errors were excluded from further analysis. An ANOVA with two within-
subject factors was computed. The first factor was type of instruction (finger movement condi-
tion vs the spatial finger-cue condition) and the second factor was congruency (baseline, con-
136 BRASS ET AL.

gruent, and incongruent). In addition a between-experiment analysis was computed using data
from Experiments 2 and 3.

Results
Error analysis revealed no significant results. In contrast to the results of
Experiments 1 and 2, no main effect was found regarding the type of instruc-
tion, F(1, 7) ⫽ 4.5, p ⫽ .07, although a small advantage for the finger-
movement condition was still present (11 ms). However, the baseline condi-
tions did not differ statistically, t(7) ⫽ .75, p ⬍ .45. The main effect for
congruency was once again significant, F(2, 14) ⫽ 75.5, p ⬍ .001. Impor-
tantly, responses in the baseline trials (282 ms) were no faster than responses
in the congruent trials (282 ms). But participants responses in both baseline
trials and congruent trials were faster, t(7) ⫽ 8.9, p ⬍ .001, than their re-
sponses in the incongruent trials (313 ms). Again, the two-way interaction
(Fig. 5) between type of instruction and congruency was significant, F(2,
14) ⫽ 9.7, p ⬍ .01. In the spatial finger-cue condition, an incongruent finger
movement led an interference effect of 43 ms in comparison with baseline
performance, t(7) ⫽ 10.4, p ⬍ .001. No facilitation effect was found for
the observed finger movement, t(7) ⫽ .5, p ⫽ .63. In the finger-movement
condition, a congruent finger cue had no significant influence on response
time. An incongruent spatial finger cue, however, led to an interference effect
of 18 ms, t(7) ⫽ 3.7, p ⬍ .01.

380

360
movement onset (ms)

340
spatial finger cue condition

320

finger movement condition

300

280

260

240
congruent baseline incongruent

congruency

FIG. 5. RT as a function of congruency (congruent, baseline, and incongruent) and type

of instruction (finger-movement condition, spatial finger-cue condition). Stimuli were identical
to Experiment 2, the participants’ required response to these stimuli was, however, different
(tapping with the index or middle finger).
 IDEOMOTOR COMPATIBILITY 137

In order to compare the effects of Experiments 2 and 3, both experiments

were combined in one ANOVA. In addition to the factors type of instruction
and congruency the factor response mapping (ideomotor-compatible vs not
ideomotor-compatible ) was added. All within-subject effects were signifi-
cant: type of instruction, F(1, 14) ⫽ 49.0, p ⬍ .001, congruency, F(2, 28) ⫽
93.9, p ⬍ 001; and type of instruction ⫻ congruency, F(2, 28) ⫽ 40.2, p
⬍ .001. The interaction between type of instruction and response mapping
was also significant, F(1, 14) ⫽ 9.0, p ⬍ .01, which indicated that the main
effect for type of instruction decreased when response mapping was not ideo-
motor-compatible. The interaction between congruency and response map-
ping was also found to be significant, F(2, 28) ⫽ 7.7, p ⬍ .01, as was the
three-way interaction between type of instruction, congruency, and response
mapping, F(2, 28) ⫽ 12.1, p ⬍ .001. In addition, the interference effect of
the incongruent finger movement decreased significantly (70 ms vs 43 ms),
t(14) ⫽ 2.5, p ⬍ .05.
Discussion
Experiment 3 showed that reducing ideomotor compatibility by changing
the relation of observed and executed action influenced the pattern of results.
This became obvious in Experiment 3 as well as in a between-experiment
comparison with Experiment 2. Importantly, the interaction of experiment
and type of instruction indicates that the advantage of movement observation
decreased significantly when the amount of ideomotor compatibility was re-
duced. In addition, the advantage of observed finger movements was largely
reduced in Experiment 3 and even disappeared completely in the baseline
conditions. This result indicated that a significant portion of the reaction-
time advantage for finger movement compared to spatial finger cues could be
attributed to the similarity of stimulus and response and not to the perceptual
advantage of finger movements compared to spatial finger cues. Observing
an irrelevant finger movement had no facilitative effect in the finger-cue
condition and the interference effect of the incongruent finger movement
was substantially reduced compared to Experiment 2. Altogether, the data
of Experiment 3 showed that the match of observed and executed action was
critical for the strong influence of observed finger movement. This finding
is in contradiction to the perceptual interpretation, which predicted no sub-
stantial changes in the pattern of results.

GENERAL DISCUSSION
The aim of this study was to directly compare stimuli of different degrees
of ideomotor compatibility. In the first experiment, it was found that partici-
pants responded more quickly to finger movements than to symbolic cues.
In addition, responses to the symbolic cues were facilitated when the irrele-
vant finger movement was congruent and an interference effect was found
138 BRASS ET AL.

when the irrelevant finger movement was incongruent. Both the facilitation
and the interference effects of the irrelevant finger movement described
above provided evidence for an automatic influence of observed finger move-
ments on movement execution.
The second experiment compared observed finger movements with spatial
finger cues. The results of Experiment 2 demonstrated that the finger move-
ment had a stronger influence on movement execution than did the spatial
finger cue. Again, if the observed finger movement was the relevant stimulus
dimension, participants reacted faster than if the spatial finger cue was the
relevant dimension. In addition, responses to the finger cues were facilitated
when the irrelevant finger movement was congruent and an interference ef-
fect was found when the irrelevant finger movement was incongruent. In the
finger-movement condition, however, an interference effect of the incongru-
ent spatial finger cue was found. This indicated that a spatial finger cue also
influenced the response. By reducing the similarity of observed and executed
actions in Experiment 3, it was demonstrated that the strong influence of
observed finger movement on movement execution was related to ideomotor
compatibility, which contradicts a perceptual interpretation of the results.

Ideomotor Compatibility for Visual Feedback from a Response

Thus far, ideomotor compatibility has been investigated, primarily with
regard to sensory feedback that was not directly related to the movement. For
example, researchers used auditorily presented words and verbal responses
(Greenwald, 1970b; Morrin, Konick, Troxell, & McPherson, 1965; Davis,
Moray, & Treisman, 1961) or printed digits and handwriting (Greenwald,
1970b). Studies that investigated the compatibility of visual feedback from a
response (Kerzel & Bekkering, 2000; Stürmer, Aschersleben & Prinz, 1999;
Brass, Bekkering, & Prinz, in press) did not directly compare stimuli of dif-
ferent degrees of ideomotor compatibility.
By using an interference paradigm, we demonstrated that stimuli with a
high degree of ideomotor compatibility had a stronger influence on the re-
sponse than less ideomotor-compatible stimuli. From the perspective of the
ideomotor theory, a symbolic cue (like the numbers in Experiment 1) does
not share any properties with the sensory feedback of the response. Hence, no
activation of the response image is expected. A spatial finger cue, however,
specifies the response position and hence should influence response initiation
to some extent. Nevertheless, the observed finger movements were much
more similar to the sensory feedback of the response than the spatial finger
cues.

Imitation and Automatic Response Activation

Movement observation was found to exert a strong influence on movement
execution, both being the relevant stimulus dimension as well as being the
 IDEOMOTOR COMPATIBILITY 139

irrelevant stimulus dimension. When finger movement was the relevant stim-
ulus dimension, the task was an imitative one in the sense that one does what
one sees. The results revealed that the finger movement condition led to
faster responses than did the nonimitative instructions (symbolic cue and
spatial finger cue). The suggestion made at the beginning of this article, that
imitating a movement is an easier task than responding to an abstract instruc-
tion, seems to be confirmed by the present results, at least at a very basic
motor control level.
But the influence of observed finger movement on movement execution
was not restricted to imitation of the finger movement. Even when the finger
movement was the irrelevant dimension it strongly affected movement exe-
cution. This finding supports the idea that movement observation exerts an
automatic influence on movement execution. If the observed finger move-
ment is congruent to the proper response, this response is activated. If the
finger movement is incongruent the incorrect response alternative is activated
and must be inhibited. Although automatic response activation has also been
demonstrated for spatial finger cueing (e.g., Kornblum & Lee, 1995), the
results of Experiments 2 and 3 indicate that the strength of the activation
processes depends on the degree of its ideomotor compatibility.
Recently, Craighero et al. (in press) proposed a motor-visual priming
mechansim which may serve as an alternative account for our results. They
found that the preparation of a specific response facilitates the detection and
discrimination of objects that have an orientation which is congruent with
the prepared response. Such a facilitation effect was also found when sub-
jects finally needed to respond with different effectors, indicating that the
improved object detection was due to the prepared congruent response and
not due to a priming of the executed response by the visual object, as was
previously demonstrated by Craighero et al. (1996, 1998). A motor-visual
priming mechanism, however, is unlikely to be able to explain fully the RT
patterns of the present experiments. First of all, in the present experiments,
subjects were not able to prepare only one response in advance. Second,
even when motor-visual priming for both possible finger movements exists,
how would this influence be able to explain the effects found for finger move-
ments on the irrelevant dimension?
Another interesting aspect that emphasizes the special role of ideomotor
compatibility is the fact that the additional information provided by the ob-
served finger movement was not relevant for the selection of the correct
response. That is, in order to select the correct response, participants only
had to specify the finger to be moved. The type of movement (lifting or
tapping) was preinstructed and remained constant within each experiment. In
other words, the observed finger movement was not informative for response
selection. For models that explain SRC effects with a controlled S-R transfor-
mation operation and that cast doubt any on kind of automatic response acti-
vation (e.g., Hasbroucq, Guiard, & Ottomani, 1990; Hasbroucq & Guiard,
140 BRASS ET AL.

1991), this finding is difficult to explain. Ideomotor theory, however, has

no difficulties explaining such a finding. It assumes that observing a finger
movement preactivates a response image of the corresponding response. This
activation process influences response initiation even if the response is prede-
fined. Therefore, observing a finger movement should affect movement exe-
cution even in a task containing no response uncertainty. In a recent study,
Brass, Bekkering, and Prinz (in press) explicitly investigated this point using
a simple response task. Indeed, they found a substantial compatibility effect
in a task in which participants always had to execute the same response.
Their findings clearly support the view that highly ideomotor-compatible
stimuli influence response initiation independent of response-selection pro-
cesses.

Ideomotor Compatibility and Recent Theoretical Developments

Ideomotor theory seems capable of explaining the results of our experi-
ments. The results are, however, also consistent with more recent theories
of compatibility, such as the model of dimensional overlap (Kornblum, Has-
broucq & Osman, 1990). This model states that ‘‘. . . SRC is the direct
consequence of the degree to which the stimulus and the response sets of a
stimulus–response (S-R) ensemble are perceptually, conceptually, or struc-
turally similar’’ (p. 855, Kornblum & Lee, 1995). The notion of perceptual
similarity is very similar to the notion of ideomotor compatibility. One’s
perception of a body movement is similar to what one perceives when exe-
cuting that movement. The theoretical advantage of the ideomotor theory is
that it provides a functional mechanism of how the similarity between stimu-
lus and response can influence response initiation by emphasizing the role
of the response image in control of voluntary action.
This notion of ideomotor compatibility fits well in the more general frame-
work of the relationship between perception and action as elaborated recently
in the Theory of Event Coding (Hommel, Müsseler, Aschersleben, & Prinz,
submitted). This theory states that ‘‘. . . perceptual contents and action goals
are cognitively represented by distributed, yet integrated codes of their distal
features, that is, perceived and to-be-produced events are coded within a
common representational medium’’ (p. 3). Because an observed action
shares a lot of features with its distal representation, the perception–action
link is very close for observed and executed actions. This view is also sup-
ported by recent neurocognitive findings.

Neurocognitive Support for the Theory of Ideomotor Action

As described in the Introduction neuropsychological research supports the
idea that a specific mechanism is involved in imitative action. Recently, some
neurophysiological and brain-imaging studies also support this view. For
example, so called ‘‘mirror neurons’’ were found in monkeys’ premotor cor-
 IDEOMOTOR COMPATIBILITY 141

tex (di Pellegrino, Fadiga, Fogassi, Gallese, & Rizzolatti, 1992). These neu-
rons are active both during specific goal-directed behavior and during the
observation of the same behavior. This finding was interpreted as a neural
correlate of an observation/execution matching system. Such a system is
thought to directly link performed actions with the recognition of observed
actions. It has been argued that an action observation/execution matching
system should also exist in human primates (Gallese, Fadiga, Fogassi, &
Rizzolatti, 1996). For instance, Fadiga, Fogassi, Pavesi, and Rizzolatti (1995)
showed that the observation of goal-directed arm movements is followed by
selective EMG activation if the primary motor cortex is concurrently stimu-
lated by transcranial magnetic stimulation. It was concluded that movement
observation leads to an activation of response-relevant cortical structures.
This view is also supported by a number of recent brain-imaging studies
(Decety et al., 1994, 1997; Rizzolatti et al., 1996; Krams et al., 1998).
Even if it is difficult to transfer these neurocognitive findings to the behav-
ioral level, the converging evidence of behavioral, neuropsychological, neu-
rophysiological, and brain-imaging data support the assumption that there is
a close relationship of action observation and action execution as proposed
by ideomotor theory.

REFERENCES
Alexander, M. P., Baker, E., Naeser, M. A., Kaplan, E., & Palumbo, C. (1992). Neuropsycho-
logical and neuroanatomical dimensions of ideomotor apraxia. Brain, 115, 87–107.
Brass, M., Bekkering, H., & Prinz, W. (in press). Movement observation affects movement
execution in a simple response task. Acta Psychologica.
Craighero, L., Fadiga, L., Rizzolatti, G., & Umilta, C. (1998). Visuomotor priming. Visual
Cognition, 5(1-2), 109–125.
Craighero, L., Fadiga, L., Rizzolatti, G., & Umiltá, C. (in press). Action for perception: A
motor-visual attentional effect. Journal of Experimental Psychology: Human Perception
and Performance.
Craighero, L., Fadiga, L., Umilta, C. A., & Rizzolatti, G. (1996). Evidence for visuomotor
priming effect. Neuroreport: An International Journal for the Rapid Communication of
Research in Neuroscience, 8(1), 347–349.
Davis, R., Moray, N., & Treisman, A. (1961). Imitating responses and the rate of gain of
information. Quarterly Journal of Experimental Psychology, 13, 78–89.
De Renzi, E., Motti, F., & Nichelli, P. (1980). Imitating gestures: A quantitative approach to
ideomotor apraxia. Archives of Neurology, 37, 6–10.
Decety, J., Grezes, J., Costes, N., Perani, D., Jeannerod, M., Procyk, E., Grassi, F., & Fazio,
F. (1997). Brain activity during observation of actions: Influence of action content and
subject’s strategy. Brain, 120(10), 1763–1777.
Decety, J., Perani, D., Jeannerod, M., Bettinardi, V., Tadary, B., Woods, R., Mazziotta,
J. C., & Fazio, F. (1994). Mapping motor representations with positron emission tomo-
graphy. Nature, 371, 600–602.
di Pellegrino, G., Fadiga, L., Fogassi, L., Gallese, V., & Rizzolatti, G. (1992). Understanding
motor events: A neurophysiological study. Experimental Brain Research, 91, 176–180.
142 BRASS ET AL.

Fadiga, L., Fogassi, L., Pavesi, G., & Rizzolatti, G. (1995). Motor facilitation during action
observation: A magnetic stimulation study. Journal of Neurophysiology, 73(6), 2608–
2611.
Fitts, P. M., & Deininger, R. L. (1954). S-R compatibility: Correspondence among paired
elements within stimulus and response codes. Journal of Experimental Psychology, 48,
483–492.
Fitts, P. M., & Seeger, C. M. (1953). S-R compatibility: Spatial characteristics of stimulus
and response codes. Journal of Experimental Psychology, 46, 199–210.
Gallese, V., Fadiga, L., Fogassi, L., & Rizzolatti, G. (1996). Action recognition in the premotor
cortex. Brain, 119, 593–609.
Gibson, J. J. (Ed.). (1986). The ecological approach to visual perception. Hillsdale, NJ: Erl-
baum.
Greenwald, A. G. (1970a). Sensory feedback mechanisms in performance control: With special
reference to the ideo-motor mechanism. Psychological Review, 77, 73–99.
Greenwald, A. G. (1970b). A choice reaction time test of ideomotor theory. Journal of Experi-
mental Psychology, 80(1), 20–25.
Hasbroucq, T., Guiard, Y., & Ottomani, A. (1990). Principles of response determination: The
list-rule model of SR compatibility. Bulletin of the Psychonomic Society, 28(4), 327–
330.
Hasbroucq, T., & Guiard, Y. (1991). Stimulus response compatibility and the Simon effect:
Toward a conceptual clarification. Journal of Experimental Psychology: Human Percep-
tion and Performance, 17(1), 246–266.
Heilman, K. M., & Rothi, L. J. G. (1985). Apraxia. In K. M. Heilman & E. Valenstein (Eds.),
Clinical neuropsychology . New York: Oxford Univ. Press.
Hommel, B., & Prinz, W. (1997). Theoretical issues in stimulus-response compatibility: Edi-
tors’ Introduction. In B. Hommel & W. Prinz (Eds.), Theoretical issues in stimulus–
response compatibility. Amsterdam: Elservier. Pp. 3–11.
Hommel, B., Müsseler, J., Aschersleben, G., & Prinz, W. (2000). The theory of event coding
(TEC). Submitted for publication.
Jeannerod, M. (1994). The representing brain: Neural correlates of motor intention and imag-
ery. Behavioral & Brain Sciences, 17(2), 187–245.
Kerzel, D., & Bekkering, H. (1999). Motor activation from visible speech: Evidence from
stimulus response compatibility. Journal of Experimental Psychology: Human Perception
and Performance, 26(2), 634–648.
Kornblum, S., & Lee, J.-W. (1995). Stimulus-response compatibility with relevant and irrele-
vant stimulus dimensions that do and do not overlap with the response. Journal of Experi-
mental Psychology: Human Perception & Performance, 21(4), 855–875.
Kornblum, S., Hasbroucq, T., & Osman, A. (1990). Dimensional overlap: Cognitive basis for
stimulus–response compatibility: A model and taxonomy. Psychological Review, 97(2),
253–270.
Krams, M., Rushworth, M. F. S., Deiber, M.-P., Frackowiak, R. S. J., & Passingham, R. E.
(1998). The preparation, execution and supression of copied movements in the human
brain. Experimental Brain Research, 120, 386–398.
Lhermitte, F., Pillon, B., & Serdaru, M. D. (1986). Human autonomy and the frontal lobes.
Part I. Imitation and utilization behavior: A neuropsychological study of 75 patients.
Annals of Neurology, 19, 335–343.
Mehler, M. F. (1987). Visuo-imitative apraxia. Neurology, 37, 129.
Morrin, R. E., Konick, A., Troxell, N., & McPherson, S. (1965). Information and reaction
time for ‘‘naming’’ responses. Journal of Experimental Psychology, 70, 309–314.
 IDEOMOTOR COMPATIBILITY 143

Ochipa, C., Rothi, L. J. G., & Heilman, K. M. (1994). Conduction apraxia. Journal of Neurol-
ogy, Neurosurgery & Psychiatry, 57(10), 1241–1244.
Prinz, W. (1987). Ideo-motor action. In H. Heuer & A. F. Sanders (Eds.), Perspectives on
perception and action. Hillsdale: Lawrence Erlbaum. Pp. 47–76.
Prinz, W. (1990). A common coding approach to perception and action. In O. Neumann &
W. Prinz (Eds.), Relationships between perception and action. Berlin: Springer-Verlag.
Rizzolatti, G., Fadiga, L., Matelli, M., Bettinardi, V., Paulesu, E., Perani, D., & Fazio, F.
(1996). Localization of grasp representations in humans by PET. Experimental Brain
Research, 111, 246–252.
Rothi, L. J., Ochipa, C., & Heilman, K. M. (1991). A cognitive neuropsychological model of
limb praxis. Cognitive Neuropsychology, 8(6), 443–458.
Roy, E. A., & Hall, C. (1992). Limb apraxia: A process approach. In L. Proteau & D. Elliott
(Eds.), Advances in psychology, Vol. 85: Vision and motor control. Amsterdam: North-
Holland. Pp. 261–282.
Schnider, A., Hanlon, R. E., Alexander, D. N., & Benson, D. F. (1997). Ideomotor apraxia:
Behavioral dimensions and neuroanatomical basis. Brain and Language, 58(1), 125–136.
Stürmer, B., Aschersleben, G., & Prinz, W. (in press). Effects of correspondence between
complex stimulus and response patterns. Journal of Experimental Psychology: Human
Perception and Performance.
Tucker, M., & Ellis, R. (1998). On the relations between seen objects and components of
potential actions. Journal of Experimental Psychology: Human Perception & Perfor-
mance, 24(3), 830–846.