Exp Brain Res (2004) 156: 518–523
DOI 10.1007/s00221-004-1911-4
RESEARCH NOTES
Sukhvinder S. Obhi . Patrick Haggard
Internally generated and externally triggered actions are
physically distinct and independently controlled
Received: 7 November 2003 / Accepted: 22 March 2004 / Published online: 15 May 2004
# Springer-Verlag 2004
Abstract In everyday life we must constantly balance our
intentions to act in a certain way with reactions that are
imposed upon us by the outside world. Recent neuroimaging studies have examined these classes of movement
separately but despite the fundamental requirement for us
to efficiently organize our internally generated and
externally triggered actions, few studies have examined
the relationship between these two classes of movement.
We measured EMG activity in the right first dorsal
interosseous while subjects performed right index finger
key presses either in an internally generated condition or
an externally triggered condition. In addition, in an attempt
to probe the relationship between the processing underlying these two types of action, we examined the effect on
reaction time (RT) and EMG activity in a third “truncation” condition in which subjects were forced to switch
from an intentional (internally generated) mode of
response production to an externally triggered mode.
Results indicated significantly greater muscle activation
for actions that were internally generated as compared to
externally triggered. Truncation caused responses to be
delayed by, on average, 54.7 ms as compared with simple
externally triggered responses, suggesting that the motor
system cannot take advantage of preexisting levels of
preparation when switching between internally generated
and externally triggered actions. Interestingly, the unique
EMG signatures of internally generated and externally
triggered actions were preserved in truncation. Thus,
subjects switched between the two types of action rather
than simply modifying an ongoing action. The results
provide peripheral physiological support for previous
S. S. Obhi (*)
CIHR Group on Action & Perception, Room 6246, Department
of Psychology, University of Western Ontario,
London, Ontario, N6A 5C2, Canada
e-mail: [email protected]
S. S. Obhi . P. Haggard
Institute of Cognitive Neuroscience, University College
London,
17 Queen Square,
London, WC1 N 3AR, UK
neuroimaging work suggesting that internally generated
actions are preceded by greater levels of preparation than
externally triggered actions. The present findings also raise
the interesting possibility that the motor system processes
these two classes of action separately even though the
motor output required is the same.
Keywords EMG . Externally triggered actions . Internally
generated actions . Motor system . Movement preparation
Introduction
In everyday life, in order to function successfully, we
constantly have to balance reactions to external stimuli
with our internally generated actions. A striking example
of what happens when this balance is lost comes from
individuals who exhibit utilization behavior (UB). This
disorder is often the result of bilateral damage to the
medial parts of the frontal lobe, especially the supplementary motor area (SMA) (Boccardi et al. 2002). Individuals
with UB cannot help but respond to and interact with
objects they come across in the environment around them,
even if such responses are inappropriate. One possible
account of such cases is that the lateral parts of the
premotor cortex (PMC) dominate motor planning when
the more medial SMA is damaged. The lateral PMC has
been previously shown in non-human primates to direct
movement on the basis of external cues, whereas the SMA
has been suggested to direct movements on the basis of
internal processes (e.g., Passingham et al. 1987; Halsband
et al. 1994). Despite the fundamental requirement for a
functional balance between reactions to external stimuli
and internally generated actions, most studies have
examined each of these types of action separately. That
is, surprisingly little research has focused on the interaction between these two classes of action. The aim of this
experiment was to redress this imbalance in the literature.
In contrast to non-human primates, the evidence for the
existence of two functionally specialized premotor systems in humans is less convincing. It has been shown that
519
several brain areas are involved in the production of both
internally generated actions and externally triggered
actions (Jahanshahi et al. 1995). These areas include the
dorsolateral prefrontal cortex (DLPFC), SMA, anterior
cingulate, the lateral PMC, parietal area 40, insular cortex,
the thalamus and the putamen. In addition the peak
component of the bereitschaftspotential (a movementrelated cortical potential that reflects motor preparation
and is measured over medial frontal motor structures
including the SMA) was greater in self-initiated movements compared with externally triggered movements.
Other studies have also found a similar network of brain
areas to be active in both classes of action but a greater (or
more sustained) activation in medial frontal areas for
internally generated actions compared with externally
triggered actions (e.g., Jenkins et al. 2000; Cunnington et
al. 2002). These findings are consistent with the notion
that the SMA is somewhat more involved in preparation
for self-initiated actions than externally triggered actions,
and is thought to be the generator of the early component
of the bereitschaftspotential (Deecke 1990). Although
generally considered as an index for voluntary, selfinitiated movement, the bereitschaftspotential has been
observed in predictably cued externally triggered actions,
albeit with reduced amplitude (e.g., Jahanshahi et al.
1995).
One way to investigate the independence or otherwise
of these systems is to measure whether and how they
might interact. Consider an individual preparing to make
an internally generated key press who is suddenly cued via
an external stimulus to make the same key press. What
would happen to the reaction time (RT) of the key press as
compared to a situation in which the individual was
externally cued whilst not preparing to make an internally
generated key press? One possibility is that, if the motor
system can take advantage of the existing levels of motor
preparation, responses to external stimuli should be
facilitated as compared to when responses are made to
the same external stimuli in the absence of internal
preparation to make the same action. However, recent
work by Astor-Jack and Haggard (2004) has investigated
this question and found the opposite result. These authors
described a “truncation” task, in which preparation for a
voluntary (internally generated) action was truncated, or
interrupted by an external stimulus requiring the same
motor response that the subject was already preparing. RT
was delayed as compared to a condition in which subjects
simply responded to an auditory stimulus when not
preparing internally generated actions. This result suggests
that internally generated and externally triggered actions
are incompatible even when the motor output required is
the same. This RT cost of internal preparation was found to
be robust in several experiments and is not due to changes
in the processing of the stimulus, since stimulus-locked
evoked potentials were not delayed or attenuated.
Those authors emphasized that the subject performed
the same action, typically a button press, in both internally
generated and externally triggered conditions. While these
movements were indeed behaviorally similar, the authors
did not report the physical parameters of the movement in
detail. Other studies of internally generated and externally
triggered actions have similarly assumed that these actions
are physically comparable (e.g., Cunnington et al. 2002).
This issue is psychologically interesting for two reasons.
First, any significant physical differences between the two
classes of action would support the view that they are
controlled by separate brain circuits. Second, the physical
form of movements in the truncation condition may show
whether subjects can integrate the characteristic movement
patterns of internally generated and externally triggered
actions to create a “hybrid” movement, or whether they
switch discretely between the two patterns. This issue
would clarify the relations between these two classes of
action.
Methods
Subjects
Twelve right-handed subjects aged 25.8±3.4 years took part in the
experiment, which was approved by the local ethics committee.
Subjects gave their written consent, and local ethical guidelines were
followed. Subjects sat at a desk in a quiet room, with their right
index finger resting on the end of a lever. Each subject performed
five practice and 28 “real” experimental trials in each of three
conditions. The order of conditions was counterbalanced across
subjects.
Apparatus
Subjects made both intentional and reactive movements by flexing
their right index finger on a metal lever. EMG of the first dorsal
interosseus muscle of the right hand was measured with surface
electrodes. The stimuli for reactive movements were taps delivered
to the back of the subject’s neck in the midline by an unseen
experimenter. Tactile stimulation was recorded by a force sensor
positioned on the subject’s neck at the point of contact of the tap.
EMG and tactile stimulation traces were recorded on a computer for
off-line analysis.
Task details
The three experimental conditions comprised an internally generated
movement condition, an externally triggered movement condition
and a truncation condition. In the externally triggered movement
condition, a simple RT paradigm was employed. After the
experimenter had warned the subject of the start of the trial using
a “trial starts” instruction, a tactile stimulus in the form of a finger
tap to the back of the neck was delivered. The intensity of the
imperative stimulus was registered via a force sensitive resistor that
was attached to the back of the subject’s neck. The stimulus was
delivered in such a way that subjects had no prior information about
when they would be required to respond (i.e., the experimenter was
out of view during experimental trials). The interval between the
experimenter’s verbal warning and delivery of the tactile stimulus,
i.e., fore-period, varied randomly between 3–10 s. Subjects were
instructed to respond with a right index finger press as fast as
possible upon sensing the stimulus. The subject’s finger press ended
the trial and after a short inter-trial interval the experimenter
indicated the beginning of the next trial.
In the internally generated movement condition, subjects were
instructed to make the same right index finger press at a time of their
520
own choosing. The only constraint was that they had to try and
randomly vary their movement such that it occurred within a 3–10 s
range. This is the same time range that the imperative stimulus was
delivered within in the externally triggered movement condition. In
the internally generated condition there was no external imperative
stimulus of any kind.
Lastly, in the truncation condition subjects were asked to initiate
and prepare an internally generated voluntary press with the right
index finger at a time of their own choosing. Subjects were
particularly instructed not to act in a stereotyped or rhythmic
manner. A tactile stimulus was delivered at a random time during the
trial, as in the externally triggered condition. Subjects were
instructed to respond to the tactile stimulus as fast as possible
with a right index finger press. Thus, subjects could make either an
internally generated or an externally triggered press, according to
whether or not the stimulus occurred before their internal process of
preparation had produced the movement. In either case a press with
the right index finger signified the end of trial.
In the truncation condition, to ensure that it was possible to
interrupt the internal preparation with a tactile stimulus, subjects
were instructed to “make their responses randomly within about 3–
10 s after the onset of the trial”. This 3–10 s range corresponds to the
range of times from which the delivery times of the tactile stimulus
were sampled. Thus, on some trials subjects made their internally
generated response prior to delivery of the tactile stimulus. These
trials were termed intentional truncation trials. On most other trials,
the tactile stimulus was delivered prior to subjects making their
internally generated response and subjects reacted to this stimulus.
These trials were termed reactive truncation trials. In this way the
truncation condition resulted in two sets of trials (internally
generated and reactive truncation). These subdivisions of the
truncation condition were treated as two conditions for purposes
of analysis. On a small number of remaining trials, subjects made
the response very soon after delivery of the tactile stimulus. As we
were measuring RTs with respect to EMG onset, any response made
in less than 75 ms after the stimulus was excluded from the analysis,
since it could not be conclusively classified as intentional or
reactive. To test for the effects of truncation, reactive truncation
trials were compared to trials in the externally triggered movement
condition and intentional truncation trials were compared to trials
from the internally generated movement condition.
Data analysis
After first rectifying the EMG data, the onset and offset of the main
EMG burst were selected interactively and the waveform was
integrated between these limits to represent the total muscle activity
associated with the action. The mean of this measure for each
subject in each condition was calculated and statistics were
performed on these values. The time of tactile stimulation in
reactive trials was determined interactively by inspecting the force
sensor trace and marking the onset of applied force. The interval
between stimulus onset and EMG burst onset was used as a measure
of RT. Any RTs greater than 1,000 ms were classed as missed trials
and were not analyzed.
Fig. 1 Median trials showing raw EMG in internally generated and
externally triggered conditions. Dashed line indicates time of
stimulus presentation
Main effects of type of movement
Figure 2 shows the rectified EMG activity in the first
dorsal interosseous for internally generated and externally
triggered movements. A repeated measures ANOVA
confirmed that there was a main effect of type of movement
on rectified EMG activity with internally generated
movements generating significantly greater EMG activity
than externally triggered movements (F(1,10)=5.210,
p=.046). In addition, the duration of EMG activity in
internally generated actions was longer than the duration
of EMG activity in externally triggered actions but this
difference just failed to reach statistical significance
(F(1,10)=3.861, p=.078).
Main effects of truncation
There were no significant main effects of truncation on
EMG activity (F(1,10)=1.088, p=.322) or duration of EMG
activity (F(1,10)=1.365, p=.270).
Interactions between the type of movement and
truncation
Results
One subject’s data was excluded from the analysis because
of skin artifact in the EMG signal. The data from the
remaining 11 subjects was analyzed. The present experiment comprised a 2×2 factorial design with the factors of
type of movement (internally generated or externally
triggered), and truncation (truncated or not truncated).
Figure 1 shows typical (median) trials for one subject.
There were no significant interactions between the two
factors of type of movement and truncation with respect to
rectified EMG activity (F(1,10)=2.951, p=.117) or the
duration of EMG activity (F(1,10)=1.964, p=.191). In fact,
EMG activity in both classes of truncation trials was
slightly lower than in the pure internally generated and
externally triggered trials, but this attenuation did not alter
the basic difference between the characteristic EMG
signatures of these classes of action.
521
Fig. 2 Rectified EMG activity
(error bars are SE) in externally
triggered and internally generated actions. EMG activity was
greater for internally generated
than for externally triggered
actions
Statistics were performed on the mean of the trimmed RTs.
A t-test revealed that truncation caused RTs to be on
average 54.7 ms longer than RTs in the externally
triggered movement condition (T(10)=−2.227, p<.05).
Figure 3 shows the RTs in the reactive trials in the
truncation condition and the externally triggered movement condition.
tests to test for differences between the two sets of pairs. It
is interesting to note that there were no significant
differences between the relationships between stimulus
intensity and the rectified EMG (T(10)=−0.67, p=.518).
Furthermore, there were no significant differences between stimulus intensities in the simple RT (i.e., externally
triggered) condition and the truncated RT conditions
which may have led to the differences in the RTs in
these two conditions (T(10)=1.18, p=.264).
Stimulus intensity
Discussion
Since in the reactive trials the tactile stimulus was
delivered by the experimenter, it was possible that the
trial-to-trial stimulus intensity differed. This is a potential
shortfall of the present experiment. However, as stimulus
intensity was known for each trial, we performed statistical
testing. Stimulus intensity was linearly regressed with the
dependent variable in question. This was done for each
condition and subject. We then used the slopes from the
regressions as new dependent variables and performed t-
Internally generated finger presses elicited significantly
greater EMG activity in the first dorsal interosseous than
did externally triggered finger presses. The truncation
condition produced RTs that were significantly longer than
those in the externally triggered action condition. However, there was no effect of truncation on the EMG activity
underlying the action. That is, the basic characteristic
EMG signatures of internally generated and externally
triggered actions were preserved in truncation, suggesting
that subjects were indeed switching between the two
classes of action in the truncation condition, and were not
producing some form of “hybrid” action.
Reaction time analysis
Internally generated and externally triggered motor
actions
Fig. 3 Reaction times (error bars are SE) in the simple externally
triggered action condition and the externally triggered truncation
condition (reactive truncation)
One explanation for the greater EMG activity in internally
generated movements compared with externally triggered
movements could be the amount of motor preparation
involved in both types of action. Motor preparation does
not occur to the same extent in externally generated
actions as in internally generated actions (Jahanshahi et al.
1995). Therefore, the greater EMG activity in the
internally generated action condition could be due to
greater levels of preparatory processing, for example
522
within frontal motor areas such as the SMA (Passingham
1987, 1993; Jahanshahi et al. 1995).
Results reported by Jahanshahi et al. (1995) showed that
no brain area was exclusively activated in externally
triggered motor actions. This is consistent with the
findings of Dettmers et al. (1996), who showed that
rCBF increases logarithmically as the force of a response
increases. Our data suggests that internally generated
motor actions are probably more forceful (based on greater
EMG activity) than externally triggered motor actions, and
this provides a plausible explanation for the findings of
Jahanshahi et al. (1995). Thus, our EMG data (but not our
truncation RT data) are consistent with the possibility that
internally generated and externally triggered actions are
actually planned and produced by the same central
structures but that these structures are simply less active
in production of externally triggered actions. Differential
activation may occur because it is inefficient to keep levels
of motor preparation high when there is uncertainty as to
when the response will be required. Alternatively, it is
possible that the longer preparation in internally generated
actions produces earlier activation in primary motor
cortex, which in turn produces the longer EMG durations
found in this study. More work is needed to examine these
issues further. Future neuroimaging studies in this area
should consider using EMG activity as a covariate.
Truncation
The truncation condition was designed to investigate the
relationship between the processing underlying internally
generated and externally triggered actions. Consider a
subject preparing an internally generated motor action who
is then externally prompted to produce the same motor
action. Can the motor system take advantage of the fact
that the premotor system is already preparing to make the
action? If the motor system can indeed harness existing
levels of motor preparation, one would expect that the RTs
in the reactive trials of a truncation condition would be
faster, or at least similar, to the RTs in a straightforward
externally triggered action. The results in this study show
the RTs in the truncation condition to be, on average,
55 ms longer than those in the simple externally triggered
action condition. There were no significant effects of
truncation on the EMG of responses and no significant
differences in stimulus intensity were found that may have
contributed to the increased RTs in the truncation condition.
It is possible that differences in foreperiod in the simple
externally triggered condition and the truncated externally
triggered condition may have contributed to the observed
RT cost of intention. However, a previous study by AstorJack and Haggard (2004) demonstrates that foreperiod
effects do not account for the RT cost of intention. The
authors conducted an experiment in which subjects
performed the truncation condition first and then a simple
externally triggered condition in which the foreperiods
obtained in the preceding truncation block were replayed
in random order. Even though the foreperiods used in both
blocks were identical, the authors still found a significant
RT cost of intention, suggesting that the effect is not due to
differences in foreperiod length.
It is also possible that the RT cost of intention arises
simply due to subjects having to divide their attention
between their internal processing and the expectancy of an
imperative stimulus. In this view, the fact that the subject
is intending to make a movement would perhaps interfere
with and possibly delay the processing of the imperative
stimulus, hence yielding an increased RT.
However, we do not believe this to be the case based on
two lines of evidence. Firstly, recent research suggests that
subjects can easily perform a perceptual discrimination
task which requires them to attend to a specific region of
space whilst holding an intentional manual response
(directed to a different spatial location) in preparation.
The only time that subjects found this task difficult was
when the preparation time for the intentional action was
less than 300 ms (Deubel and Schneider 2003). Put
simply, stimulus processing is not impaired whilst subjects
are holding an intentional action in preparation, as long as
sufficient (minimal) time for preparation is allowed. In our
experiment, since the minimum foreperiod was usually
3,000 ms, subjects should have been able to easily divide
their attention between their internal processing and
external stimulus detection. This is because after the first
300 ms, they would not require attention for the initiation
of their response. Hence, we do not believe that attentional
factors alone can account for the RT cost of intention in the
present experiment.
Secondly and more specifically related to our experiment, the idea that stimulus processing is impaired in
truncation compared to a simple externally triggered
condition has been investigated in a previous study and
shown to be incorrect (Astor-Jack and Haggard 2004).
Astor-Jack and Haggard (2004) used a similar experimental design to the one reported here, with the main
difference being the use of an auditory instead of a tactile
imperative stimulus. To investigate the time course of
processing of the auditory stimulus, they recorded auditory
evoked potentials. They then compared three relevant
components of the waveforms from the truncated
externally triggered trials with those from the simple
externally triggered condition. They found an increase in
the size of the N150 component and the P300 component
in the truncation condition as compared with the simple
externally triggered condition. Most interestingly though,
they found that the N150 peak occurred only very slightly
later in the truncation condition than in the simple
externally triggered condition (8 ms later). This difference
was much too small to account for the 50+ms RT cost of
intention found in their and the present experiment.
Furthermore, the fact that there was an enhancement of
the N150 suggests that processing of the auditory stimulus
was actually enhanced rather than reduced in the
truncation condition. This result clearly suggests that
differences in the processing of the external stimulus due
523
to divided attention cannot account for the RT cost of
intention.
We suggest that the RT cost of intention is consistent
with the notion that a “motor switch” occurs in truncation
in which the mode of movement production changes from
internally generated to an externally triggered one. The
increased RTs suggest that this “motor switch” from
internally generated to externally triggered modes of
action production takes time. Furthermore, internally
generated and externally triggered actions each have a
distinctive physiological pattern. These patterns are
preserved in the truncation condition but a delay in the
production of the action is inserted. Lastly, the fact that we
did not observe a hybrid EMG pattern in truncated trials,
but instead found a preservation of the characteristic EMG
pattern for internally generated and externally triggered
trials in truncation, further supports the idea that subjects
do actually switch between modes of response production.
The switch-cost in the truncation condition remains to
be explained. In cognitive psychological models, switching between two processes typically occurs when the two
processes cannot simultaneously access a limited-capacity
information-processing channel (Broadbent 1958). This
study did not aim to identify the locus of this limitation,
but we note that intentional actions are generally
performed serially, rather than in parallel. Further work
must address the underlying causes of this delay in
responding when subjects are forced to switch from
internally generated to externally triggered action production.
In summary, we have shown that i) muscle activation in
the primary agonist producing index finger presses is
significantly higher in internally generated actions than in
externally triggered actions and ii) truncation causes
actions to be delayed but does not alter the physiological
characteristics of those actions. We have suggested that
these quantitative differences should also be considered by
neuroimaging studies investigating possible qualitative
differences between brain processes underlying these two
types of actions.
References
Astor-Jack T, Haggard P (2004) Intention and reactivity. In:
Humphreys GW, Riddoch MJ (eds) Attention for action.
Elsevier (in press)
Boccardi E, Della Sala S, Motto C, Spinnler H (2002) Utilisation
behaviour consequent to bilateral SMA softening. Cortex
38:289–308
Broadbent DE (1958) Perception and communication. Pergamon,
Oxford
Cunnington R, Windischberger C, Deecke L, et al (2002) The
preparation and execution of self-initiated and externallytriggered movement: a study of event-related fMRI. Neuroimage 15:373–385
Deecke L (1990) Electrophysiological correlates of movement
initiation. Rev Neurol (Paris) 146:612–619
Dettmers C, Ridding MC, Stephan KM, Lemon RN, Rothwell JC,
Frackowiak RS (1996) Comparison of regional cerebral blood
flow with transcranial magnetic stimulation at different forces. J
Appl Physiol 81:596–603
Deubel H, Schneider WX (2003) Delayed saccades, but not delayed
manual aiming movements, require visual attention shifts. Ann
N Y Acad Sci 1004:289–296
Halsband U, Matsuzaka Y, Tanji J (1994) Neuronal activity in the
primate supplementary, pre-supplementary and premotor cortex
during externally and internally instructed sequential movements. Neurosci Res 20:149–155
Jahanshahi M, Jenkins H, Brown RG, Marsden CD, Passingham
RE, Brooks DJ (1995) Self-initiated versus externally triggered
movements. 1. An investigation using measurement of regional
cerebral blood-flow with pet and movement-related potentials
in normal and Parkinsons-disease subjects. Brain 118:913–933
Jenkins IH, Jahanshahi M, Jueptner M, Passingham RE, Brooks DJ
(2000) Self-initiated versus externally triggered movements II.
The effect of movement predictability on regional cerebral
blood flow. Brain 123:1216–1228
Passingham RE (1987) Two cortical systems for directing movements (Review). Ciba Foundation Symposium 132:151–164
Passingham RE (1993) The frontal lobes and voluntary action.
Oxford Psychology Series No. 21, Oxford University Press