Measuring the Mirror Neuron System: A Combined EEG/fMRI Study

By simply observing others, humans can infer a great deal about their goals,
sensations, and emotions. The neural basis of this "mind reading" ability is
poorly understood, but the discovery of mirror neurons in macaque monkeys was
a great step forward for a scientific account of social cognition. Mirror
neurons fire either when a macaque performs an action, such as grasping, or
when the macaque observes another human or macaque performing this action. It
has therefore been proposed that mirror neurons serve to translate observed
actions into the neural activity associated with the macaque's own actions,
therefore turning the problem of understanding others' actions into the easier
problem of understanding one's own actions.

While mirror neurons have been invoked to explain an impressive variety of
human cognitive functions, including language and imitation in addition to
action understanding, studying the mirror neuron system in human subjects
remains elusive. fMRI researchers often study mirror neurons by finding "shared
voxels"--small volumes of brain tissue which are active during both action
execution and observation In contrast, EEG researchers usually quantify mirror
neuron activity using the suppression of the rolandic alpha (8-13 Hz) and beta
(13-30 Hz) rhythms during action observation and execution, suppressions which
are believed to reflect modulation of the primary sensorimotor cortex by
premotor mirror neurons. While these two approaches, shared voxels in fMRI and
rolandic rhythm suppression in EEG, are assumed to both reflect activity in the
mirror neuron system, to date there is no evidence that they measure the same
underlying neural activity. To address this issue, the present study was
designed to determine the relationship between shared voxels and rolandic alpha
and beta suppression.

In this experiment, we intend to determine how shared voxels, as measured with
fMRI, are related to rolandic alpha and beta suppression, as measured by EEG.
Because they are believed to both measure mirror neuron activity, we
hypothesize that rolandic alpha and beta will correlate with activity in shared
voxels such that rolandic alpha and beta suppression will be enhanced on blocks
in which shared voxels are particularly active.

During the experiment, fMRI BOLD signal and EEG will be recorded simultaneously
to be analyzed off-line.

In complex action observation blocks, participants will be shown videos in
which an actor will perform a goal-directed hand action with an object, such as
spreading jam on bread or placing a flower in a vase. Only the actor's arm and
the object will be visible. The complex action condition will have two control
conditions: complex control, in which the videos will show an arm moving in a
similar trajectory in an identical environment, but without a clear goal and no
object interaction; and static control, in which the videos will show an arm in
an identical environment but with no movement.

Four additional conditions are needed to isolate regions that respond to the
execution of actions, and to disenntangle the contribution of motor and
proprioceptive brain regions to the rolandic alpha and beta rhythms. In complex
action and control blocks, a small green or red circle will alert participants
that a complex action or control block, respectively, is starting. Following
the initial circle, large green circles or red circles will be presented and
will subsequently shrink and eventually disappear. In the complex action
blocks, the location of the circle will indicate which object (a bowl and
spoon, a wine glass, or a mug) participants should interact with and the
shrinking of the circle will indicate how long to interact with the object. In
control blocks, participants will fixate on the red circles without performing
actions. Additionally, there will be two simple action execution conditions. In
each block of the voluntary movement condition, participants will squeeze a
foam ball. In involuntary movement blocks, however, the experimenter will
squeeze the participant*s hand around the foam ball. A small green or red
circle will alert participants that a voluntary or involuntary movement block
is, respectively, is starting, and shrinking green or red circles will indicate
how long the participant should interact with the foam ball.

A period of resting state, during which participants will passively watch a
fixation cross for 12 minutes, will also be acquired after the observation and
execution sessions to estimate the connectivity between brain regions as their
correlations within the frequency range below 0.1Hz [6]. Finally, an 8-minute
anatomical scan will be acquired at the end of the experiment.

Thus, there will be four types of sessions (as well as an anatomical scan):
complex action observation, complex action execution, simple action execution,
and resting state. Because the Keysers lab has used the stimuli previously in
fMRI studies but not in EEG studies, 30-60 minute piloting will be conducted in
5 healthy participants using EEG only in the dummy scanner while the scanner
noise is played from a loudspeaker to determine how many trials are necessary
to reliably record rolandic alpha and beta suppression, how long the pauses
must be between trials to ensure that the rhythms return to baseline levels,
and whether it is preferable to utilize a block design with three
movies/complex actions per block or if rolandic alpha and beta suppression are
observed more reliably when pauses break up the movies/complex actions into
single events. The final design of the study will be such that participants
spend no more than 1.5 hours in the scanner (including anatomical scanning),
and we anticipate that a scanner session of 1.0 hours will likely be
sufficient.

Additionally, prior to combined EEG/fMRI testing, the first five subjects will
be pretested with EEG only while they perform the observation and execution
sessions (but not the resting-state session) in the dummy scanner while the
scanner noise is played. After the first five subjects, we will evaluate
whether the pretesting is useful for predicting whether participants display
rolandic alpha and beta suppression during the combined EEG/fMRI experiment. If
the pretesting proves useful (i.e., subjects who display rolandic alpha and
beta suppression during pretesting also reliably display this suppression
during the combined EEG/fMRI experiment), then we will pretest the remaining
subjects. But if the pretesting is not useful (i.e., there are participants who
display rolandic alpha and beta suppression during pretesting but not during
the combined EEG/fMRI experiment), then we will cease to pretest participants.

Note that the piloting is distinct from the pretesting. The purpose of the
piloting is to collect the data needed to fine-tune the procedure, and
therefore certain features of the experiment, such as the length of the pauses
between blocks, will be varied systematically. The pretesting, however, aims to
determine whether a particular participant reliably displays rolandic alpha and
beta suppression, and therefore the final procedure will be used.

The experiments will not entail more than minimal risk or burden to the
participants. EEG, which will be measured across 32 channels embedded in an
elastic cap, is a routine and noninvasive technique used in neuroscience
research. In order to collect simultaneous fMRI data, subjects will be exposed
to a magnetic field of 3 Tesla and rapidly alternating magnetic gradients and
radio frequency fields. This field strength is used routinely in fMRI and MRI
research. On rare occasions, a peripheral nerve in the abdomen in stimulated by
the changing magnetic gradients, which is innocuous but results in an itching
sensation. While fMRI requires that participants lie still in a confined space,
our experience is that most participants can easily make it for 90 minutes and
are very glad to participate. Additionally, short breaks will be provided
between sessions at the request of the participant or if the experimenter
believes it is necessary. Combined EEG-fMRI research has already been conducted
at UMCG and it has been verified that the EEG apparatus is MRI-compatible and
poses no risks to participants.

The study is not intended to benefit the subjects directly, but our experience
suggests that individual volunteers often feel rewarded from participating in
this sort of study. Moreover, the data collected during this study could
improve our understanding of the mirror neuron system. In particular, this
investigation could help clarify the relationship between the fMRI phenomenon
of shared voxels and the EEG phenomenon of rolandic alpha and beta suppression,
which have each been used individually to study the mirror neuron system.