J Physiol 584.2 (2007) pp 651–659 651
The nature of corticospinal paths driving human
motoneurones during voluntary contractions
Jane E. Butler2, Thomas S. Larsen1, Simon C. Gandevia2 and Nicolas T. Petersen1
1Department of Exercise and Sport Sciences, and Department of Neuroscience and Pharmacology, University of Copenhagen, Denmark
2Prince of Wales Medical Research Institute and University of New South Wales, Randwick, NSW, Australia 2031
The properties of the human motor cortex can be studied non-invasively using transcranial
magnetic stimulation (TMS). Stimulation at high intensity excites corticospinal cells with fast
conducting axons that make direct connections to motoneurones of human upper limb muscles,
while low-intensity stimulation can suppress ongoing EMG. To assess whether these cells are
used in normal voluntary contractions, we used TMS at very low intensities to suppress the firing
of single motor units in biceps brachii (n = 14) and first dorsal interosseous (FDI, n = 6). Their
discharge was recorded with intramuscular electrodes and cortical stimulation was delivered at
multiple intensities at appropriate times during sustained voluntary firing at∼10 Hz. For biceps,
high-intensity stimulation produced facilitation at 17.1 ± 2.1 ms (lasting 2.4 ± 0.9 ms), while
low-intensity stimulation (below motor threshold) produced suppression (without facilitation)
at 20.2 ± 2.1 ms (lasting 7.6 ± 2.2 ms). For FDI, high-intensity stimulation produced
facilitation at 23.3 ± 1.2 ms (lasting 1.8 ± 0.4 ms), with suppression produced by low-intensity
stimulation at 25.2 ± 2.6 ms (lasting 7.5 ± 2.6 ms). The difference between the onsets of
facilitation and suppression was short: 3.1 ± 1.2 ms for biceps and 2.0 ± 1.5 ms for FDI. This
latency difference is much less than that previously reported using surface EMG recordings
(∼10 ms). These data suggest that low-intensity cortical stimulation inhibits ongoing activity
in fast-conducting corticospinal axons through an oligosynaptic (possibly disynaptic) path,
and that this activity is normally contributing to drive the motoneurones during voluntary
contractions.
(Received 10 April 2007; accepted after revision 9 August 2007; first published online 16 August 2007)
Corresponding author N. Petersen: Department of Exercise and Sport Sciences, University of Copenhagen,
Blegdamvej 3, 2200 Copenhagen, Denmark. Email: npetersen@ifi.ku.dk
The control of human voluntary movement involves
activation of both excitatory and inhibitory connections
in the motor cortex. The activity of these cortical
circuits may be investigated by direct recordings from
cortical cells during voluntary tasks in non-human
primates (e.g. Evarts, 1965; Lemon et al. 1976; Fetz &
Cheney, 1980; for review see Phillips & Porter, 1977).
In humans, the rapidly conducting component of the
corticospinal pathway and its possible involvement in
voluntary movements has been investigated indirectly
with magnetoencephalography (e.g. Salenius et al. 1997),
electroencephalography (e.g. Macefield & Gandevia, 1991;
for review see Colebatch, 2007), and stimulation of the
motor cortex (e.g. Day et al. 1989; Palmer & Ashby, 1992;
Nielsen et al. 1993).
Magnetic or electrical stimulation of the motor cortex
is a non-invasive method to assess the excitability of the
motor pathways in awake human subjects in which the
stimuli and evoked responses can be timed precisely (for
reviews see Rothwell et al. 1991; Petersen et al. 2003).
These studies have usually focused on the evoked excitatory
output from corticospinal cells. This output evokes a
short-latency response with a central conduction time of
∼5 ms for upper limb muscles (Merton & Morton, 1980;
Rothwell et al. 1991). Central conduction times measured
to motor nuclei at many spinal levels are consistent with a
rapidly conducting projection at ∼70 m s−1 (e.g. Gandevia
& Plassman, 1988). However, stimulation at intensities
subthreshold for evoking motor potentials can inhibit
the output to a second cortical stimulus (Kujirai et al.
1993). When delivered alone, very low-intensity TMS can
reduce the ongoing electromyographic activity during a
voluntary contraction (Davey et al. 1994) and during
walking (Petersen et al. 2001). This is believed to occur
by suppression of ongoing corticospinal excitation to
motoneurones by activation of intracortical inhibitory
circuits. If the low-intensity TMS activates intracortical
inhibitory cells with only one or a few synapses to the
motor cortical output cell (e.g. Kujirai et al. 1993; Fisher
et al. 2002), the suppression would occur at a latency not
C© 2007 The Authors. Journal compilation C© 2007 The Physiological Society DOI: 10.1113/jphysiol.2007.134205

652 J. E. Butler and others J Physiol 584.2
much longer than that of the short-latency facilitation
evoked by higher stimulus intensities. However, the latency
of the suppression of the surface EMG normally occurs
∼10 ms after the short-latency facilitation (Davey et al.
1994; Petersen et al. 2001). One interpretation of this long
delay is that the cortical stimulus suppresses the output
of corticospinal cells with slowly conducting axons (or
indirect paths), and that it is these cells that drive the
voluntary contraction. The present study was designed to
resolve this paradox.
Rather than recording surface EMG, we have studied
the change in firing probability of single motor units
evoked by TMS to obtain an accurate comparison
of the latency of the initial facilitation with that of
pure suppression (in the absence of facilitation). We
hypothesized that the latency difference between the
facilitation evoked by high-intensity stimulation and the
suppression evoked by low-intensity stimulation would
be short. This would be consistent with activation
of oligosynaptic intracortical inhibitory circuits, which
reduce output of the rapidly conducting corticospinal cells
during the voluntary contraction.
Methods
Experiments were performed on six healthy adult
subjects with no history of neurological disorders. They
were studied on multiple occasions. In one subject, only
multiunit studies were performed. Subjects were seated
comfortably in a chair. All procedures were approved by the
local ethics committee and conformed to the Declaration
of Helsinki. Subjects gave informed written consent to
participate in the study.
EMG recordings
EMG activity was recorded from right biceps brachii or
right first dorsal interosseus (FDI) using a concentric
needle electrode (13R05 Dantec Medical, Denmark). The
electrode was positioned such that activity from a single
motor unit could be clearly identified during a weak
contraction (< 5% maximal contraction strength). In
some studies, multiunit activity was recorded from the
needle electrode. Auditory feedback of the EMG activity
was provided to help the subject to maintain a continuous
stable contraction throughout the experiment. EMG
signals were amplified, filtered (25 Hz – 3 kHz), sampled
(10 kHz) and stored to computer via an analog-to-digital
interface (micro 1401 and Spike2 software, Cambridge
Electronic Design, UK).
To allow on-line analysis of single motor unit firing, the
EMG signal was passed through a window discriminator
(Digitimer D130 spike processor, Welwyn Garden City,
UK), and a running histogram of interspike intervals was
displayed.
Transcranial magnetic stimulation of the motor cortex
During the sustained weak contraction of the identified
single motor unit, transcranial magnetic stimuli (TMS)
were delivered over the contralateral motor cortex
with either a round (13 cm outside diameter coil,
the current running in the clockwise direction) or a
figure-of-eight prototype coil (diameter of the wing 9 cm,
handle orientated postero-laterally) and a Magstim Rapid
stimulator (Magstim Company Ltd, Dyfed, UK). During
previous studies of suppression of EMG by cortical
stimulation, both a figure-of-eight coil and round coil have
been used (Davey et al. 1994; Petersen et al. 2001). No
obvious differences in the evoked suppression have been
noted (also confirmed here for single unit recordings).
Similarly, the rapid rate stimulator produces a biphasic
stimulus pulse. Although the effect of such a stimulus
may differ from the standard monophasic stimulus pulse,
the study by Petersen et al. (2001) used both a rapid rate
stimulator and a standard Magstim 200. In that study no
difference in the suppression of the EMG was seen. In
the present study, a rather high stimulus rate was optimal
because the recording time for the the units was limited.
Stimuli (20–65% maximal stimulator output) were
delivered at ∼1.1 s intervals with four different conditions.
These included: no stimulus (0%) as a control, a
subthreshold stimulus that was intended to produce
suppression but no initial facilitation, a slightly higher
intensity stimulus (by ∼5%) that would also produce
suppression, and a higher intensity stimulus (by ∼10%)
that would produce initial facilitation. Note that the
high-intensity stimulus does not evoke an overt motor-
evoked potential and, in the conventional sense, when
referring to excitation of the muscle it is ‘subthreshold’.
To emphasize the low intensity of stimulation used in
this study, we define ‘very low’ intensity stimulation
as the lowest level of ‘subthreshold’ stimulation which
produces pure short-latency suppression. The four
stimulus conditions were delivered in random order. Stable
recording sessions over at least 10 min were required to
obtain sufficient stimuli (> 100) in each condition to
evaluate the effect of TMS in the single motor units.
Throughout the text the intensities are given as the
percentage output on the stimulator.
Only after a large number of stimuli were obtained
could we get an indication of the success of the initial
choice of stimulus intensities. Occasionally we needed to
do an extra run to get a measure of the suppression or
of the facilitation. It has been reported that TMS at an
interstimulus interval of about 1 Hz may produce effects
on cortical excitability in hand muscles (Chen et al. 1997)
although this does not occur for elbow flexor muscles
(Martin et al. 2006). For experiments involving single
motor unit recording a high stimulus rate is preferred.
Nevertheless, a possible effect of the interstimulus
C© 2007 The Authors. Journal compilation C© 2007 The Physiological Society