ChapterPDF Available




Mirror neurons have generated intense interest since their discovery in the early 1990s because they offer a potential neural mechanism for linking the observation of a conspecific’s action to the representation of the motor plan for that action in the observer’s brain. Much progress has been made in the last two and one half decades, but much remains mysterious as well. In this chapter we discuss research in macaque monkeys and what has been revealed about the functional, anatomical and connectivity characteristics of mirror neurons. We also discuss the use of non-invasive brain imaging to measure mirror neurons, considering the pros and cons. Further discussion concerns what role mirror neurons play in action understanding as well as various models of mirror neuron function.
C. Chad Woodruff
Northern Arizona University
Mirror neurons have generated intense interest since their discovery in the early 1990s
because they offer a potential neural mechanism for linking the observation of a conspecific’s
action to the representation of the motor plan for that action in the observer’s brain. Much
progress has been made in the last two and one half decades, but much remains mysterious as
well. In this chapter we discuss research in macaque monkeys and what has been revealed about
the functional, anatomical and connectivity characteristics of mirror neurons. We also discuss the
use of non-invasive brain imaging to measure mirror neurons, considering the pros and cons.
Further discussion concerns what role mirror neurons play in action understanding as well as
various models of mirror neuron function.
While concisely defining empathy, compassion and self-compassion is difficult, it can be
said that at a minimum, they have in common the processing of the relationship between self and
other. Since their discovery in 1992, mirror neurons (MNs) have generated much interest as they
are neurons that respond both to one’s own actions as well as to the observation of others
actions. This sort of self-other overlap within individual neurons, mirror neurons, has drawn
much interest because they likely represent a neural mechanism that contributes to action
understanding a possible neural basis for the rapid transformation of social information from
perceptual to intentional.
In recent years however, much of the enthusiasm over mirror neurons (MN) has been
tempered by claims that they are not necessarily doing what many have theorized (e.g. Hickok,
2009; Southgate & Hamilton, 2008; Cook et al., 2014; Catmur, 2015). Add to this the difficulty
of noninvasively measuring MNs and we have a recipe for a multitude of claims about MNs with
likely insufficient scrutiny of many of them (Hickok, 2009; Dinstein et al., 2008; Hobson &
Bishop, 2017). For example, there are multiple studies which conclude that MN activity was
observed simply because blood oxygen changes occurred in areas believed to contain MNs (e.g.
Iacoboni et al., 1999), or oscillations of a particular frequency believed to reflect MNs were
modulated (e.g. Oberman et al., 2005; Woodruff & Maaske, 2010). On the other hand, some
papers admonishing caution, warranted as caution may be, seemed to miss some of the evidence
suggestive of a degree of correspondence between the relevant brain measures and MN activity
(e.g. Hickok, 2009; Hobson & Bishop, 2017).
This chapter will provide a brief review of the discovery of MNs, followed by a review of
some of the most recent findings about them. We will discuss recent invasive, non-human animal
research, functional Magnetic Resonance Imaging (fMRI) and electroencephalography (EEG)
studies, the role of MNs in action understanding, some theoretical models attempting to explain
MNs, and end with suggestions for future directions. Anticipating the main conclusions of this
chapter, we will argue that MNs play important roles in processes like empathy and compassion,
but that current functional neuroimaging techniques, with perhaps a few methodological
exceptions, are unable to isolate with certainty MN activity. We will also conclude that MN
activity is not sufficient for action understanding because MNs are unlikely to inherently process
self-other distinctions. However, their activity profiles do appear to be modulated by self-other
distinctions and as such likely make important contributions to action understanding.
Discovery of Mirror Neurons and Attempts to Measure Them
Mirror neurons were first discovered in premotor cortex (Fig. 1), field 5 (F5) of the
macaque monkey brain somewhat by accident when an experimenter reached toward the
monkey’s peripersonal space to exchange target objects. To the researchers’ surprise, the
recorded neuron fired, albeit less vigorously, even though the monkey remained still with no
particular intention to move (di Pellegrino et al., 1992). Eventually Giacomo Rizzolatti and
colleagues coined these neurons MNs, indicating that they not only process one’s own actions,
but mirror the actions of others as well (Gallese et al., 1996). As such, MNs are theorized to
represent a very rapid mechanism whereby one understands another’s intention because
observation of the other’s action activates that intention in the observer. Arguably, the substantial
excitement around the discovery of MNs has contributed to the emergence of Social Cognitive
Neuroscience, using functional brain measures to test social-cognitive hypotheses. Such
measures range from single-cell and linear multi-electrode probes to non-invasive hemodynamic
and electromagnetic measures to methods of manipulating MN activity (e.g. Transcranial
Magnetic Stimulation).
----Insert figure 1 about here---
Since the discovery of MNs in macaque monkey area F5, researchers have delineated two
particular regions in the monkey brain which contain them: the Inferior Arcuate Sulcus (IAS)
and the intraparietal sulcus (IPS)/inferior parietal lobule (IPL). The mirror properties of these
two regions have been studied to a greater extent than some additional regions that have been
subsequently identified as potential sites of MNs. These include medial supplementary motor
area (SMA) and the hippocampus (in humans). Due to the preponderance of invasive data from
macaque monkeys, much more is known about monkey MNs than human. Indeed, Rizzolatti and
colleagues have published a recent series of papers describing various new findings related to
MN activation profiles, location and interconnectivity between regions containing MNs (Kohler
et al., 2002; Fogassi et al., 2005; Umiltà et al., 2008; Caggiano et al., 2009; Caggiano et al.,
2011; Nelissen et al., 2011; Maeda et al., 2015; Rizzolatti and Fogassi, 2014; Giese and
Rizzolatti, 2015). We will begin with MNs in area F5.
Ventral Premotor Cortex. Area F5 lies within the banks of the inferior arcuate sulcus
(IAS) (Fig. 1) and contains neurons that appear to encode an animal’s intentions (Umiltà et al.,
2008; Caggiano et al., 2009; Caggiano et al., 2011). In a recent study using multi-electrode linear
arrays, Bonini (2014) found that approximately 41% of the neurons they recorded in area F5
were MNs. This region, which is just anterior to the frontal eye fields, has been divided into three
sub-regions: F5a and F5p in the posterior bank of the IAS and F5c in the inferior convexity (Fig.
Caggiano et al. (2009) reported newly discovered features of some F5 MNs, noting that
about half of them were selective for either peripersonal or extrapersonal space, suggesting that
these neurons are sensitive to whether an observed action occurs within the monkey’s actionable
space. Of these space-selective neurons, half of them were either coded in a metric space
(Cartesian coordinates) or in an operational space. Caggiano et al. concluded that half were
operational due to the fact that their receptive fields were dynamic. Within this class of so-called
operational MNs, when presented with an object in peripersonal space, an operational MN with a
peripersonal receptive field flipped its receptive field to peripersonal space when a transparent
barrier was introduced between the monkey and the action while those neurons that coded
peripersonal space stopped responding altogether. The critical point is that the introduced
transparent barrier did not change the metric distance between the monkey and the action, but it
did change the operational space. An action in peripersonal space that previously did not activate
an extrapersonal MN, did activate it when that peripersonal space was rendered non-operational
by the transparent screen. In other words, these neurons seem to be less about precisely how
close is the action taking place (metric distance), but is it taking place within an actionable, or
operational, distance. This observation lends further support to the idea that some MNs are
coding the meanings of actions rather than the simple spatial location of those actions.
Caggiano et al. (2011) reported one-fourth of MNs from which they recorded in ventral
F5 exhibited view-independent activity. In other words, these neurons fired to the observation of
conspecifics actions whether viewed from 0°, 90° or 180° angles. Given that an animal’s point-
of-view is irrelevant to the action-goals of the individual he is observing (save, perhaps, cases
where the observed is attempting to deceive the observer), neural activity reflecting the
intentions of the other should not change with point-of-view. Rozzi and Coudé (2015) suggested
that view-independent MNs may represent intentions at a higher-level of understanding whereas
view-dependent MNs may be more focused on connecting visual descriptions of actions to motor
goals. Following sections of this chapter will discuss the role these regions may play in action
understanding and the potential for functional brain measures to reveal MN activity.
Inferior Parietal Lobule. Subsequent to the discovery of MNs in F5, investigations have
included a focus on monkey IPL where recordings have been taken in three subdivisions, PF,
PFG and PG (Fogassi et al., 2005; Rozzi et al., 2008). Of these three, PFG appears to have been
most consistently found to contain MNs. Fogassi et al. reported approximately 75% of recorded
MNs in PFG were selective for an intention, responding when executing or observing an act of
placing food in the mouth but not when the goal was to place the food in a container mounted on
the shoulder, near the mouth. These data represent a double dissociation insofar as neurons that
were selective for observing food-to-mouth actions did not respond to place-in-container, while
those selective for placing-in-container did not respond to bringing-to-mouth, despite the fact
that the arm movements were nearly identical. These results are consistent with the claim that
IPL MNs reflect the intention of an observed action.
According to Rozzi et al. (2008), the tight correspondence between PFG MN response
preferences for a specific goal to both the executed and observed motor acts lends support to the
idea that MNs represent a direct matching between the perception of an actors actions and the
neural representation of those actions in the observer. As will be discussed later, these data are
consistent with Direct Perception models (DP) insofar as it can be argued that perception of an
action directly activates the motor plans for that action in the observer. Because PFG receives
direct projections from the superior temporal sulcus (generally construed as a visual region
selective to biological motion) and no such direct input to F5 has been found, Rozzi et al.
suggested that area PFG may represent the first stage of the MN system.
Another parietal region receiving direct inputs from superior temporal sulcus (STS) and
containing MNs is anterior intraparietal (AIP) sulcus, an area involved in coding other’s hand
grasps (Maeda et al., 2015; Nelissen et al. 2011; Peeters et al., 2009; Rizzolatti & Sinigaglia,
2010). Interestingly, some of AIP’s inputs come from a region in the lower bank of the STS that
is considered part of the inferior temporal cortex, part of the classic visual ventral, ‘what’
pathway, involved in identifying visual objects (Ungerleider & Mishkin, 1982). As discussed
later, debate exists over whether MNs themselves encode the meaning of an action or whether
this is done instead by posterior regions believed to process inferred meaning. The fact that AIP
receives input from part of the ‘what’ pathway suggests that MN coding of others intentions may
be influenced by semantic input from the ventral visual stream. Considering the PFG and AIP
findings together, Rizzolatti and Fogassi (2014) suggested that the superior temporal sulcus is the
source of visual perceptual information into the “mirror neuron circuit”.
Using Neuroimaging to Infer MN Activation
Various, relatively non-invasive, techniques have been used to explore correlates of the
MN system. These include fMRI, EEG/MEG, transcranial magnetic stimulation (TMS) and
transcranial direct current stimulation (tDCS). The bulk of this work has involved the first two
tools, fMRI and EEG, and this is where the focus of this chapter will be aimed. While these tools
are indispensable for investigating the human MN system, they are not ideal and have serious
limitations. These limitations include the inability to see activity anywhere close to the single-
cell level and hence these techniques cannot reveal whether recorded brain signals reflect
contributions from individual neurons that fire both during execution and observation of actions.
We will discuss the virtues and limitations below as well as current research. We will conclude
that these techniques are useful for studying the MN system, but that their limitations warrant
great caution and restraint in drawing conclusions about mirror activity.
Extant research strongly suggests that EEG mu suppression, which is known to originate
from sensorimotor cortex (Pfurtscheller et al., 1997), is modulated by regions associated with
MN’s (e.g. vPMC, (Pineda, 2005)). Different labs define mu differently, with some defining it as
8-12/13Hz, being comprised of the alpha band, while others define it as including alpha as well
as 14-30Hz beta. This latter framework might make sense insofar as mu refers to rhythms
coming from the Rolandic fissure (sensorimotor cortex) and is historically associated with both
execution and observation of actions. This contrasts with alpha rhythms, which are maximal in
occipital electrodes, are not modulated by execution/observation, but are modulated by eyes
open/eyes closed.
Since rhythms coming from sensorimotor range from 8-30Hz (Pfurtscheller et al., 1997;
Hari et al., 1998), it is reasonable to consider both sensorimotor alpha and beta as part of the mu
complex (Hari & Salmelin, 1997; Cook et al., 2014). Regardless of nomenclature, significant
correlations between 8-13Hz mu and several brain regions traditionally not associated with MNs
strongly suggest that mu suppression is nothing like a pure measure of MNs (Yin et al., 2016;
Braadbaart et al., 2013; Fox et al, 2015). Rather, with potential contributions from as many as
seven brain regions, mu suppression could potentially reflect seven, or more, various cognitive
and/or emotion processes, thereby creating ambiguity with respect to correlations between mu
suppression, behavior and self-report measures. As such, future research should include efforts to
resolve this ambiguity. One way might be to “clean” data using methods like Laplacian spatial
transformation to potentially distinguish separate contributions to the mu rhythm.
Another useful approach could be to link, to the extent possible, activity in these
supposedly MN-unrelated regions to mirroring-unrelated processes. Once identified, holding
these mirror-unrelated psychological processes constant in experimental paradigms, we could
minimize expected contributions from the MN-unrelated brain regions (e.g. middle frontal gyrus
(mFG), cerebellum and thalamus).
Given that Rizzolatti and colleagues discovered MNs in the vPMC of macaque monkeys,
it seems reasonable to assume that they exist in human vPMC. Of course, this assumption is
nothing more than an untested hypothesis without supporting evidence. However, measures of
human MNs (e.g. fMRI, EEG, MEG) in most cases have not allowed the inference that observed
effects were specifically related to MNs. In all non-invasive functional brain measures, signals
are observed only with a sufficient number of neurons (greater than 10k) synchronously
undergoing post-synaptic potentials (Logothetis et al. 2001). Since a MN is defined as a neuron
that fires both during execution and observation, it requires the ability to see the activity of a
particular neuron during both conditions. Any measure, therefore, that requires 10k+ neurons to
generate an observable signal could not, strictly speaking, reveal whether a subset (no study has
reported greater than 50% of sampled neurons within a given region to be MNs) of those are
active at all, let alone whether they are active during both execution and observation.
One particular experimental paradigm, however, offers a possible means of inferring
successive activation of a specific population of neurons. Repetition Suppression (RS) is a
phenomenon observed with non-invasive functional brain measures that occurs when individual
neurons are activated by the same stimulus as in the preceding trial. RS is a measure of the
magnitude of attenuation of the response of these neurons to the second presentation of the
stimulus, relative to the response of the neurons to the initial presentation of the repeated
stimulus. The attenuation in repetition suppression derives from this reduced, or suppressed,
neural response to the second, repeated stimulus. Since the stimuli are identical, it is assumed
that the identical neurons are active in response to both the first and the second stimulus
presentation and that the suppression results from the identical neurons being less active during
the second presentation, rather than from a reduction in the total number of neurons responding
(Henson, 2003).
Regarding the measuring of MNs, RS has been used with fMRI and EEG to measure
successive activation of the same population of neurons in a cross-modal condition. To
understand, consider that motor system activity is reduced for the execution of the second of two
identical actions. If a MN is active in response to both execution and observation of the same
action, then the second activation of that MN should be suppressed, both when action execution
is followed again by execution of that action and also when action execution is followed by
observation of that action (and vice versa). In the latter, cross-modal condition, only neurons that
respond both to execution and observation should be candidates for RS. Such neurons by
definition would be MNs as they would be neurons that respond to both execution and
In a clever pair of experiments, Shiri Simon and Roy Mukamel (2015a; 2015b) separately
used EEG and fMRI to test the hypothesis that MN-related electrical activity and hemodynamic
responses, respectively, are dependent on conscious processing. Predicated on the assumption
that action understanding requires conscious understanding, they asked whether putative
measures of MN activity would show sensitivity to whether observed stimuli were perceived
consciously. Using the Continuous Flash Suppression (CFS) paradigm (Tsuchiya & Koch, 2005),
they presented to one eye a brief hand movement and to the other a Mondrian display (a
seemingly random, geometric display of multiple color swatches) each image flickering on and
off in opposite phase of one another, so that, at any given instance, only one eye was being
stimulated. With this masking technique, they were able to mask from conscious perception, the
presentation of some of the hand actions. By comparing putative MN activity elicited by
consciously perceived stimuli to that elicited by stimuli not consciously perceived, Simon and
Mukamel asked whether these apparent MN measures would be sensitive to this manipulation of
conscious perception. We will consider the EEG results, then the fMRI, followed by a discussion
of the degree to which the two data sets complement one another.
Simon and Mukamel (2015a) looked at sensorimotor suppression in both the mu (8-
13Hz) and the beta (15-25Hz) rhythms at central electrode sites C3 and C4, proximal to the
central sulcus the generator of Rolandic mu and beta rhythms (Pineda, 2005). They found
sensitivity to conscious perception both in the beta and in a sub-band of mu (8-10Hz) whereby
suppression was greater for consciously perceived stimuli relative to ones not consciously
perceived. As mentioned above, because the frequency band range of occipital alpha rhythms is
identical to mu rhythms (8-13Hz), it is difficult to know to what extent the effects observed in
central electrodes in Simon and Mukamel’s study may have been driven by alpha, rather than
mu, suppression. However, a significant interaction involving electrodes revealed that the pattern
of suppression in 8-13Hz range was different, thereby supporting the claim that the effects in
central electrodes could not be due solely to posterior alpha.
The fMRI version of this experiment was very similar in design, though it included an
additional opacity manipulation, with half of the observed hand actions being presented with low
opacity, and hence being consciously perceived less often. Focusing on the hemodynamic
response differences to consciously and not-consciously perceived stimuli, Simon and Mukamel
(2015b) found multiple regions believed to contain MNs, including dorsal PMC, primary motor
cortex (M1) and somatosensory cortex that were more active when the stimuli were consciously
perceived. Two additional and adjacent regions were also preferentially activated by consciously
perceived stimuli: 1) The pSTS, a region that is not believed to contain MNs, but is suggested to
provide the biological motion detection needed to activate the MN system, and 2) The TPJ - an
area not known to contain MNs but commonly believed to be involved in inferential processes
associated with so-called mentalizing as well as self-other discrimination. Since these five
regions were more active during conscious perception, Simon and Mukamel argue that these
regions fulfill one necessary condition of action understanding conscious perception of that
There are limitations of course to exactly what conclusions can be drawn from these two
studies. To begin with, there is great uncertainty in the degree to which either mu/beta
suppression or fMRI activation in classic MN regions reflects the activity of MNs and MNs
alone. Concerning mu suppression, we know that it is correlated with activity in PMC, consistent
with the claim that mu suppression results from the desynchronizing input of PMC to
sensorimotor cortex (Pineda, 2005), but we also know that it is correlated with various other
regions, some not obviously part of the MN system. For example, Braadbaart et al. (2013) found
that, in addition to regions commonly believed to contain MNs (iFG, PMC, IPL), mu suppression
was similarly correlated with activation in the cerebellum, middle temporal lobes, middle frontal
gyrus (mFG) and the thalamus. While the possibility exists that MNs will be discovered in these
regions, one must assume the null hypothesis for now that these areas do not contain MNs.
Therefore, it should probably be assumed that any instance of measuring mu suppression likely
reflects mirror- and non-mirror related activity. Nonetheless, the results of this pair of studies
(Simon & Mukamel, 2015a; 2015b) are intriguing in their consistency with the hypothesis that,
to the extent that MN activity is associated with action understanding, it is necessarily associated
with conscious perception.
Mizuguchi et al. (2016) provided evidence that what they refer to as the action
observation network (AON), which includes premotor and inferior parietal regions, is insensitive
to inferences about another person’s actions while TPJ and pSTS are sensitive. They presented
participants with separate videos of two different actors (one small in stature and the other large
and muscular) lifting light and heavy dumbbells, with the participants inferring that the smaller-
stature actor is working harder to lift the heavy weights. This allowed the researchers to analyze
fMRI results and to determine that the AON does not respond differentially, whereas TPJ and
pSTS do. These results show that the latter brain structures are likely involved in inferencing of
the sort characterized by inferring effort while the former appear not to be. Of course, this study
represents but only one particular instance of inference (i.e. how much effort is the actor
exerting), but it is inferencing specifically about bodily exertion and presumably not about
mentalizing. These data suggest that even inferences about motor events may not be happening
in the AON, otherwise one would expect these regions (PMC and IPL) to show differential
activation. Of course, absence of evidence is not evidence of absence, so further research should
continue to address this question, but the data seem to be in line with claims that MN regions, as
constituted by the AON, are not involved in inferential processes, and therefore are unlikely to
play a key role in inferring the meaning of an action.
Putative Measures of Mirror Neurons and Empathy
MNs are theorized to be involved in empathy and perspective-taking, supported by
research from multiple labs that have reported fMRI and EEG measures of putative MN activity
that are correlated with perspective-taking and empathy. Schülte-Ruther et al. (2007) were
among the first to report that activations in two different regions, one believed to be in the MN
system (left and right inferior frontal) and one believed to be the source of input to MNs, pSTS
(Bonini, 2016), were correlated with self-reported empathy.
Complementary to these findings, Zaki et al. (2009) used fMRI to find brain regions
activated by an empathic accuracy (EA) task in which participants gave affective ratings of
videos of an actor revealing emotional stories form her/his life. The actor in the stimulus videos
had previously given their own ratings of their affect moment by moment and this timecourse of
affective ratings was correlated with the actor’s own judgements higher correlations meant
higher empathic accuracy. Zaki et al. found that greater accuracy was associated with greater
fMRI activity in PMC, IPL, STS, as well as dmPFC and rostral mPFC. The PMC and IPL
correlations with empathic accuracy are consistent with the claim that MN function is related to
Perry, Troje and Bentin (2010) found significant mu and beta suppression responses to
point-light displays of individuals walking. Mu suppression was found to be greatest for a
condition that required inferring the intentions of the individual (walking toward or away) as
compared to identifying the gender, suggesting that mu suppression reflects activity related to
action understanding. Perry et al. were also the first to report a significant correlation between
mu suppression and self-reported empathy, as measured by the Empathy Quotient (EQ; Baron-
Cohen & Wheelwright, 2004). Surprisingly, the relationship was negative, such that greater mu
suppression was associated with lower EQ scores.
On its own, the negative relationship found by Perry et al. (2010) could be considered an
anomaly. The finding does not stand alone however, as it has been reported by at least two other
labs (Woodruff & Klein, 2013; Horan et al, 2014). Woodruff, Martin and Bilyk (2011) initially
found a negative relationship between mu suppression and the perspective-taking (PT) subscale
of the Interpersonal Reactivity Index (IRI; Davis, 1983). Although the correlation was non-
significant after removing an outlier, the negative relationship pointed the researchers to the idea
that mu suppression may have a non-linear relationship to PT, whereby absence of mu
suppression (as seen in some studies of individuals with autism (Oberman et al., 2005)), relates
to poor PT, while high-levels also work against PT. It may be that, in order to take another’s
perspective, a moderate amount of mu suppression is ideal. This could be predicted by theories
of empathy that emphasize the importance of self-other discrimination (e.g. Batson et al, 1991).
This construct is argued to be necessary to keep the social observer from reacting to the
intentions of the person he is observing. Because PT implies that the observer recognizes that the
intentions he is experiencing belong to the other, not to himself, low self-other discrimination
would lead to a focus of attention on the self as though those intentions belonged to the self.
Rather, under this model of empathy, perspective-taking occurs when the observer’s brain is able
to distinguish between its own intentions and those of the person he is observing. Consequently,
Woodruff, Martin and Bilyk (2011) computed mu suppression difference scores (execution
observation) and found that these difference scores were positively related to PT, consistent with
the hypothesis that PT occurs to the extent that the observer’s brain is able to sort his own
intentions from the ones he is receiving from the observed. A similar finding was subsequently
reported by Hoenen, Shain and Pause (2013).
These experiments show a distinct relationship with empathy and perspective-taking,
albeit not as straightforward as some would have guessed. The processes reflected by mu
suppression appear to subserve empathy, not by enabling the observer to ‘become’ the observed,
but to experience some of the observed individual’s intentions, however not so much that the
observer’s brain becomes confused by whose intentions belong to whom.
Does Mu Suppression Reflect Mirror Neurons? Yes, and then Some...
A seminal paper on which much of the mu suppression research has been based was
published by Jaime Pineda (2005) in which he makes a compelling case that suppression of the
sensorimotor mu rhythm likely reflects MN input to sensorimotor cortex. Founded on the nearly
universal assumption that the EEG signal is generated by post-synaptic potentials,
desynchronization of the mu rhythm would be the result of input from premotor cortex, and the
MNs within, inducing post-synaptic potentials in sensorimotor neurons. Suppression of
oscillatory brain rhythms is believed to reflect the desynchronization of neural oscillations,
relative to those oscillations when neural activity is at baseline. This logic can be understood in
the example of a 100-member choir: the choir is louder when everyone sings together, compared
to say, a solo voice. No matter how loud the soloist sings, she will not have a louder voice than
the other 99 members of the choir, even if each choir member is singing at a lower volume than
the soloist. This of course is because, when everyone sings together, the sound waves are
synchronized and add to one another. Even if everyone is singing quietly, the sound waves of 99
vocalists can add up to a larger number of decibels than a soloist singing at maximum volume.
Therefore, the activation-related desynchronization of the sensorimotor mu rhythm is
argued to result from MN input. This section will evaluate evidence relevant to the question of
whether mu suppression reflects MN activity. Research will be cited to support the role of mirror
neurons in suppressing the EEG mu rhythm, but not exclusively so. The resulting argument will
be advanced that, regardless of the extent to which mu suppression does reflect MN and non-MN
activity, the signal does appear to be related to empathic processes.
If indeed mu suppression is reflecting MN input, then it should correlate with activation
in brain regions believed to contain MNs. Yin, Liu and Ding (2016) used a combined EEG-fMRI
paradigm in which they assessed correlations between mu suppression and BOLD responses.
Using Second Order Blind Identification (Tang et al, 2005) in an attempt to disentangle mu from
alpha rhythms, they found correlations in classic MN regions (i.e. the IFG and the IPL) as well as
in an area reported to contain MNs in humans (Mukamel, 2010), namely, the SMA. Not
surprisingly, they found that mu suppression also correlated with regions not previously
identified as areas containing MNs, including superior frontal, superior parietal cortices,
superior, middle and inferior temporal gyri as well as the middle cingulate and precuneus. Those
are quite a few areas not traditionally associated with MNs whose potential contributions to mu
suppression are concerning points for research using mu suppression as an indicator of MNs.
Hobson and Bishop (2016) report results of a cleverly designed and high powered (N=61)
experiment intended to directly evaluate the extent to which mu suppression can be disentangled
from posterior alpha and to assess mu suppression’s topographic specificity. Using a baseline
(moving kaleidoscopic videos) that appears to have drawn participants’ attention more than the
typical execution and observation conditions, their results indicated that observing this relatively
salient baseline condition elicited more alpha blocking in occipital electrodes than action
execution but that this was not the case for central electrodes. If the mu suppression observed by
Hobson and Bishop were explained by occipital-derived alpha, one would expect the greater
desynchronization seen in occipital alpha to be evident in the central electrodes as well. As the
authors note, their results indicate that distinguishing sensorimotor-specific mu rhythms from
occipital alpha is possible, at least for action execution. The same pattern was not found however
for action observation; Hobson and Bishop (2016) were unable to distinguish mu reactivity from
alpha during action observation, suggesting that caution is warranted when attempting to isolate
mu suppression during action observation.
Hobson and Bishop (2017) claim that there are few published studies that assessed
whether mu suppression can be distinguished from alpha suppression. However, despite citing
Woodruff et al. (2011), they do not mention that this study found a significant interaction
between task (execution, observation) and region (central, occipital). Using execution-
observation difference scores, Woodruff et al. found that difference scores were all significantly
different from null for the central electrodes only. Occipital difference scores were not
significantly different from zero and tended toward enhancement rather than suppression (fig. 2).
As suggested by Hannah Hobson (personal communique) however, this difference could be
explained as occipital alpha, with a contribution from Rolandic mu during execution, but not
during observation. It seems unlikely however, that this alternative explanation could readily
account for why, if central electrode 8-13Hz suppression reflects Rolandic mu during execution
but posterior alpha during observation, execution observation (self other) difference scores
should correlate with self-reported PT (Woodruff et al., 2011a). Regardless, consideration of this
dissociation is important when arguing that mu suppression is not a reliable measure of mirror
neuron activity.
----Insert figure 2 about here---
An important point to consider when comparing the pattern of results from Hobson and
Bishop (2016) to most other mu suppression experiments, is that they calculated mu
desynchronization in a different, though not unprecedented (Muthukumaraswamy & Johnson,
2004), way. While mu suppression is typically computed as log(task/baseline) (Pineda, 2005),
they subtracted task from baseline (task - baseline). This computational difference not only
means that their results cannot be directly compared to results utilizing Pineda’s convention, but
also that their results may be influenced by individual differences in baseline mu suppression.
Using ratios normalizes scores so that one is looking at percentage differences rather than
absolute differences. For these reasons, caution is warranted in comparing Hobson and Bishops
results to most published mu suppression research.
Fox et al. (2015) performed a meta-analysis on 85 mu suppression studies, testing two
hypotheses in particular: 1) If mu suppression reflects MN activity and can be dissociated from
posterior alpha, then the meta-analysis should reveal topographical specificity, and 2) if mu
suppression reflects MNs, biological, compared to non-biological, motion stimuli should elicit
more mu suppression. With regard to the first hypothesis, topographic specificity was found for
mu suppression induced by action execution in that the strongest effect size was in central
electrodes with confidence intervals not overlapping with effects in frontal, parietal and occipital
electrodes. This finding indicates that for action execution, the source of mu suppression likely is
around the central sulcus, consistent with a sensorimotor source. It should be noted that
confirming exactly where the generator of mu suppression is requires a solution to the inverse
problem (Jun et al, 2005). Although the central electrodes showed the greatest effect size,
without source localization, there is no guarantee that the signal is generated in the sensorimotor
cortex. Nonetheless, the data for action execution are encouraging.
The data are less so with regard to such specificity for action observation. The confidence
intervals for the effect sizes in all five regions tested (frontal, central, parietal, occipital,
temporal) were overlapping, although the authors noted that the n for observation studies was
low and that the hypothesis test therefore may have been underpowered. The possibility should
be considered that the generator of suppression in the 8-13Hz range for action observation is
distinct from the generator for action execution. As well, it is possible that there are common
generators for execution and observation, but that observation-induced suppression may have
additional, non-sensorimotor, non-MN generators.
Considering the various findings related to mu suppression as a measure of MN activity,
it is likely that mu rhythms do include input from MNs, but also potentially from many other
brain regions not believed to contain MNs. Furthermore, not all research measuring mu
suppression argues that the signal reflects MNs. As such, future reports of mu suppression
research should not attempt to draw strong conclusions about MNs and should focus conclusions
instead on mu suppression. Regardless of the actual source of the signal, mu suppression does
appear to relate reliably to empathic processes (Perry et al., 2010; Woodruff, Martin & Bilyk,
2011; Woodruff & Klein, 2013; Hoenen, Shain & Pause, 2013).
Beta Rhythms and Mirror Neurons
There is also evidence that oscillations within the beta frequency (here defined as 14-
30Hz), recorded at central sites, have a similar potential to reflect MN modulation of
sensorimotor cortex (Pfurtscheller, Neuper & Edlinger, 1997; Hari et al., 1998;
Muthukumaraswamy & Johnson, 2004; Cheng et al., 2006). Like the 8-13Hz mu rhythms, beta
rhythms have been observed to desynchronize in reaction to execution and observation of
movements. Given this mirror-like response pattern as well as the high degree of certainty of a
sensorimotor generator of the signal, it has been argued that beta rhythms can be taken as a
marker of MN activity.
On the other hand, multiple labs have reported beta enhancement rather than suppression,
though these have tended to involve static rather than moving stimuli (Güntekin & Başar, 2007,
2009, 2010; Woodruff et al., 2011b). Using negative and neutral stimuli from the International
Affective Picture Scale (IAPS), both Güntekin and Başar and Woodruff et al., found that a
relative synchronization of the beta rhythm was elicited by negative stimuli. Woodruff et al. also
administered the IRI and found that greater beta synchronization elicited by negative stimuli was
correlated with higher dispositional PD. These findings of synchronization appear contradictory
to the desynchronization findings mentioned above (Hari et al., 1998, Muthukumaraswamy &
Johnson, 2004). But, in addition to resulting from static rather than motion stimuli, the stimuli
associated with enhancement, IAPS images of highly negative imagery, in some cases grotesque,
were highly emotional while the stimuli eliciting beta desynchronization have tended to involve
simple hand movements. This observation points to two intriguing alternative possibilities: beta
enhancement/suppression represents two states of one system, or it represents two distinct
systems, both oscillating in the beta range. The latter possibility implies that the beta oscillations
observed at central electrode sites would be a mixture of these two different beta generators.
While it might be worthwhile to investigate this question using methods such as blind source
separation (e.g. Tang et al., 2005), or inverse modeling (Jun et al., 2005), some extant data offer
some interesting details about enhancement/suppression, with one experiment demonstrating, for
the first time, both enhancement and suppression within the same experiment.
Our lab recently conducted three different experiments looking at beta
enhancement/suppression (Woodruff, Barbera & Von Oepen, 2016; Woodruff et al., in prep.;
Woodruff et al. in prep.). Figure 3 shows the results of three experiments that reveal three
distinct patterns of beta enhancement/suppression, divided into two hertz sub bands. Both A and
C represent two, as yet unpublished, datasets from two different empathic accuracy tasks (see
Zaki et al., 2009), while B represents data from an emotion discrimination task. While the three
tasks differ from one another in important ways, they have in common designs that allow
distinguishing self- from other-related beta reactivity. In B, we see a distinct difference between
beta elicited by participants’ judgments of their own versus others’ emotional state. When asked
how the person in a photograph felt, beta responses were negative, below zero, indicating
suppression. However, when the question for the same photographs asked, “How does the way
this person feels make you feel?” beta responses were greater than zero, indicating beta
enhancement. These data are the first to demonstrate beta enhancement and suppression within
the same task as a function of identity (Brown, Goodman & Woodruff, 2016; Woodruff,
Goodman & Brown, in preparation).
----Insert figure 3 about here---
The three datasets taken together indicate a task-related electrophysiological dissociation
of beta enhancement/inhibition. These outcomes raise the question of what socio-cognitive
processes vary across these three experiments that could explain the obtained distinct patterns of
data. We suggest an explanation based on the focus of attention to self-relevant compared to
other-relevant information. Experiment A involved an empathic accuracy task of a similar design
to experiment C with the difference being that social targets in A were much more emotionally
expressive than those in C. As concerns self-focus, because social targets were more expressive,
emotion recognition processes were not particularly taxed, leaving attentional resources available
for reflection on how the social target’s emotions are impacting the self. Indeed, experiment B, in
which participants viewed photographs of emotional facial expressions (Ekman & Friesen,
1976), supports this conjecture by showing that participants exhibit beta suppression when asked
“How does this person feel?” compared to “How does the way this person feels make you feel?”
As can be seen in figure 3b, focusing on the feelings of individuals in the photographs was
associated with suppression throughout much of the beta band range when the question required
other-focused attention. In contrast, focusing on one’s own emotional reaction to another’s
feelings was associated primarily with enhancement.
As for experiment C, the task was rather difficult for participants as can be seen by the
rather low mean EA correlations for C (r=.24) compared to A (r=.70). The reason for the
difference in task difficulty can be explained by the fact that social targets in C were less
emotionally expressive than those in A. It follows that participants in C would need to focus
more attentional resources on the target’s emotional expressions leaving fewer resources to
process the impact on the self. Once again, experiment B supports this claim as it found that
other-focused attention related to suppression while self-focused attention related to
enhancement. There are then two distinct possibilities as concerns beta enhancement vs.
suppression: 1) Beta represents one neural process that in some cases synchronizes above
baseline (self-focus) and in other cases desynchronizes (other-focus), 2) The generators of the
synchronization and desynchronization may represent distinct sets of neurons such that self-
focus and other-focus differentially engage these sets. The socio-cognitive hypothesis predicts
that the scalp distributions of the beta activity will differ significantly between tasks, suggesting
that they derive from at least partially non-overlapping areas of tissue, while the other predicts
that, because the neural tissue is assumed to be identical, no difference should obtain. These
questions could be addressed by assessing scalp potential distributions (a significant difference
would imply two distinct generators) or inverse modeling. Until then, we offer an observation
that, in experiment B, it could be argued that the task difficulty is greater for the self-focused
condition (“How does the way this person feels make you feel?”) as it requires first assessing the
emotions of the others and then how that emotion makes the participant feel. The other-focused
task by contrast only requires the first of those two processes. Therefore, instead of beta
suppression seen with the easier, other-focused task, the task-difficulty hypothesis predicts that,
despite being a task focused on others emotions (experiment C), because it is easier, it should be
associated with synchronization rather than desynchronization.
Do MNs Constitute or Contribute to Action Understanding?
This fascinating debate has taken place since at least Gregory Hickok’s (2009) sharply
and concisely titled paper, Eight Problems for the Mirror Neuron Theory of Action
Understanding in Monkeys and Humans. Perhaps one of the most productive publications
relevant to this debate is one titled Mirror Neuron Forum (Gallese et al., 2011) in which leading
researchers discussed several issues including the question of MNs relationship to action
understanding. The five contributing authors manage to discuss the question from surprisingly
“orthogonal” perspectives ranging from Vittorio Gallese’s concept that MNs represent embodied
simulation to Cecelia Heyes evolutionary/developmental account of MNs to Hickok’s claim that
MN activity does not constitute action understanding. In this section, we will consider some of
the philosophical underpinnings of various models, some of the evidence from invasive brain
recordings taken to support them, as well as non-invasive recordings (fMRI and EEG), in
monkeys and humans. We will conclude that certain questions, such as, “Does action
understanding occur in MN regions or sensory regions?” conflates levels of analysis. Action
understanding should not be considered a property unique to, or fundamentally constitutive of,
MN function, nor should it be considered to reside in sensory (Hickok, 2009) systems to the
exclusion of mirror mechanisms.
Hickok (2009) makes an argument that combines two observations: 1) the supposed lack
of a direct match between the intent of the observed action and the intention encoded by the
specific MN, and 2) with the claim that certain actions are ambiguous with respect to their
intentions with the same action being associated with multiple possible intentions. For example,
if you were in a taxi passing someone looking at the taxi and waiving her hand, it would be
difficult to determine whether the person was waving at you or hailing the taxi. Strictly speaking,
responding to the sight of an object would be visual processing and should not therefore be part
of pure motor processing. On the other hand, there exists in premotor cortex, canonical neurons
that respond to the sight of particular objects, apparently as a function of the motor affordances
these objects offer. Indeed, how could the activity of the motor system be “about” anything if it
did not have a specific relationship to sensory input and to the effects of motor output, as
evidenced by sensory feedback? It is likely untenable to view the aboutness of a motor plan as
being contained within a specific type of neuron. Equally likely untenable is the claim that the
aboutness comes simply from the sensory-evaluated consequences of the motor plan. The
aboutness instead is more likely best viewed as the property that arises with highly correlated
sensory input, motor output and sensory feedback. By this logic, MNs could not be “where”
action understanding occurs conversely, sensory systems could not be “where” action
understanding occurs. Rather, each of these components are sub-processes of action
understanding, suggesting future research should concern itself not with any constitutive action
understanding theory of MNs but with further characterizing the likely contribution they make to
generating emergent action understanding.
Interestingly, a different argument can clearly be made that in such ambiguous situations
MNs would on their own be unable to discern between the two possible intentions (hailing a cab
versus waving to a friend). Since MNs by definition respond only to perception of biological
motion stimuli, MNs should be ignorant of any non-biological stimuli, such as the taxi cab
(although, inferior temporal cortex input to AIP might suggest otherwise (Nelissen et al., 2011)).
If MN processes cannot on their own process the taxi cab, in motion or not, then they could not
on their own understand that the waving is intended to hail the taxi. Because MNs cannot
ostensibly process non-biological stimuli and because in many the meaning of action lies in
relationship to a changing world, much of it non-biological and therefore inaccessible to MNs.
While it is likely true that action understanding involves the sensory consequences of a
motor intention, the claim that action understanding is exclusively a result of processing sensory
consequences of a motor intention does not recognize the emergent nature of this construct and it
denies a likely important motor contribution. While individual MNs could not possibly be
sufficient for action understanding, it would be very surprising indeed if they played no role
whatsoever. Rather than action understanding arising simply from MN activity or from purely
sensory processing, a likely more progressive approach is to view both sorts of processing as
necessary antecedents to action understanding.
Direct-Matching and Inferential Processing Models. Various models of intention
understanding have been offered, differentiated primarily by whether they model intention
understanding as what happens when observation of another’s actions activates the
corresponding intention in the observer’s mirror system or they model it as a more ‘deliberative’
process of inference. Here we will briefly discuss two different models that can be used to
explain the MN system’s role, or lack thereof, in action understanding. The first of these models
is Direct-Matching, or Direct-Perception (DP), models which hold that MN activation is
sufficient for action understanding, claiming that “motor-perception” occurs by virtue of the
sensory perception of an action leading to the motor plan for that action (Michael, 2011; Michael
& de Bruin, 2015). Inferential Processing (IP) models are on the other end of the spectrum of
MN involvement in action understanding because they tend to deny a role of MNs in action
understanding (Csibra, 2008), and in some cases, claim MN activation requires action
understanding rather than the other way around.
Evidence in favor of DP models includes the observation that F5 neurons coding a
grasping goal do not vary as a function of the method used for grasping (left hand, right hand and
mouth). Further evidence comes from Umiltà et al. (2008) who found that some F5 neurons
responded to the monkey’s grasping of an object with a pair of pliers, even when using different
pliers engineered to require different hand movements to operate. Traditional pliers require
squeezing the handles together to grasp an object. These traditional pliers were used in addition
to a pair that required the opposite hand movement rather than squeezing, these required
opening the hand to make the pliers grasp the object. Using this clever manipulation of squeezing
vs. opening while holding the goal constant for both types of pliers (grasp the object), the
researchers were able to isolate neurons that fired to a particular goal regardless of the motor
movements used to achieve it. Taken together, these data support the claim that F5 neurons code
the intentions of the animal.
Caroline Catmur (2015) offers a more skeptical view, laying out four conditions she
believes mirror neuron activity profiles must meet in order for that activation to constitute direct
perception of another’s intention. First, she suggests that perception of another’s actions should
activate only one motor intention in the observer. This postulate might be an overly simplistic
characterization of what actually happens. Instead, understanding may NOT comprise activation
of a single intention but graded activation of multiple, highly-related processes, with the
intention associated with peak activation corresponding to the observer’s best guess about the
other’s intention. It is certainly the case that a particular intention would come to dominate this
graded activation and could be seen as Catmur’s “only one intention”. Otherwise, the first
condition may not be a necessary one.
Catmur goes on to argue that an additional condition is that the solitary intention
activated in the observer must correspond to exactly one intention in the actor. Using the
reasoning above, a more realistic view may be that the graded activation of multiple intentions
corresponds to a graded activation in the others brain, with significant overlap between the
graded distributions of intentions. While condition two seems reasonable, condition three is not
so clear. The third condition suggests that “…this mapping from motor program to intention
must be the same in the observer as in the actor…” This condition seems to ignore the central
claim of the DP models that the motor program is the intention. Admittedly, as Catmur and
others have pointed out, this reasoning may constitute circular logic insofar as proving that MNs
code intentions involves showing that MNs are active when the animal activates an intention.
But the hypothesis that the motor program simply is the intention is not inherently circular, only
the notion that proving the hypothesis involves merely showing that MNs are active when the
animal has an intention is circular. In other words, if one assumes the activation of MNs to
indicate the activation of an intention, one cannot use as evidence for that assumption, the fact
that MNs are active this is circular. Instead, one must manipulate intentions to determine
whether MN activity changes, and indeed it does (e.g., Umiltà et al., 2008). If this is correct, then
Catmur’s third condition is not really a condition. Rather the mapping required by that condition
exists insofar as the motor program just is the intention. This is not a one-to-one mapping, but
rather an identity. According to direct-perception models, they are one in the same thing,
equivalent to saying that A must map directly onto A.
Condition four seems to similarly confound the issue in that it requires that the activation
of a motor plan must automatically lead to the activation of the corresponding intention. These
conditions necessarily fail to be met by direct-perception models simply because the conditions
deny the fundamental tenet of DP models - that motor programs are intentions. And, as Michael
(2011) notes, they run the risk of implying dualism because the arguments seem to rest on the
assumption that neural activity is not itself the intention, but rather somehow maps onto the
intention. That is, to say that activating a certain motor plan must activate an intention begs the
question of what the intention is. If it’s not some set of neural activity, then is it some non-
physical substance? Having said this, it is likely proper not to regard the intention as “in the
neurons” but rather as a property that emerges when the activation of these neurons has a specific
relationship to the world.
There is still one necessary condition DP models fail to meet but that does not deny the
assumption that motor programs are intentions. This condition is that activated motor programs
distinguish between self and other - understanding another’s intention necessitates understanding
that the perceived intention belongs not to the observer but to the observed. Despite the fact that
MN activation distinguishes self from other, self-other discrimination is unlikely to derive from
MN processes themselves but more likely reentrant processing from areas that infer agency. Put
differently, if MNs represent intentions, how do they represent whose intention? It is unlikely
they do. Rather, according to self-other discrimination research, this function seems to have more
to do with the temporoparietal junction than classic MN regions.
DP is a compelling idea in many ways, not the least of which is the way it offers a
parsimonious neural mechanism for embodied models of social processing. It would seem very
efficient for natural selection to have favored understanding others’ intentions with the very
same brain regions that correspond to understanding our own intentions. To have one’s brain and
affective states “resonate” with another’s would seem to be a relatively quick and relatively low-
cost process for experiencing the intentions of others.
However, it is not the case that experiencing the intentions of another is equivalent to
understanding the intentions of another. Suppose that I were to experience your intention to eat
ice cream but that I did not go through the additional, presumably cognitive (Michael, 2011),
inferential, step of recognizing that the intention I was experiencing was not my own but one that
I was experiencing because I observed you act on said intention. With the circumvention of this
inferential step, I am likely to have the experience of intending to eat the ice cream myself.
Surely it is illogical to suppose that I understand your intentions if I (erroneously) believe the
intention belongs to me. In this case, if asked, I presumably would be unable to answer what was
your intention. Rather, I would be able to report only what I falsely perceived as my own
intention, at which time I would proceed to wrestle you to the floor for the waffle cone,
especially if it were peanut butter and chocolate!
Rather, it is suggested here that strict DP models are non-starters regarding sufficiency to
explain intention understanding. To the extent that they assume that intention understanding
occurs without processing the fact that these intentions belong to the other, they deny a necessary
condition for intention understanding - agent attribution. It might be argued that MNs themselves
code information about intention attribution (Bonini, 2016). Indeed, MNs in the pre-motor cortex
(PMC) show a variety of preferences, with the preponderance of PMC neurons responding to
self-actions more than observation of others’ (Rozzi & Coudé, 2015). In other words, most PMC
neurons are more active when an animal has the intention to perform an action himself, with
significantly fewer neurons activated by observing the actions of others. Additionally, while
some projections from PMC (specifically, F5p) do extend directly to the spinal cord, it is to
propriospinal and not spinal motor neurons, and therefore not to neurons that synapse with
muscles. It is suggested that this is one means by which the observer is able to avoid
automatically acting on another person’s intentions: PMC MNs do not appear to activate spinal
motor neurons, but rather must go through one or more synapses to influence motor output.
An interesting question that emerges here begs whether such apparent self-other
discrimination is a fundamental capacity of MNs, or is MN activation modulated by efferent
connections from regions whose job is self-other differentiation (regions like, TPJ). Interestingly,
this hypothesis could be tested by exploiting the temporal resolution of EEG to determine the
timecourse of self-other differentiation in mu suppression. The fundamental capacity model
predicts that self-other differentiation should be evident from the onset of mu suppression,
whereas the efferent connections model predicts no significant difference in mu suppression early
in time, only emerging once regions such as the TPJ have settled on a conclusion about whether
the action is the observer’s or the observed. Therefore, another test of this hypothesis would
involve establishing whether MN regions receive efferent connections from areas like TPJ.
The latter model is consistent with the claim that processes such as action understanding
cannot be pointed to in the brain but rather happen as multiple parallel routes of processing
increasingly modulate one another over time. Empathic understanding of another’s actions might
be described as a process that starts with motor simulation of the other’s affective state (related
to MN activation), followed by subthreshold potentiation of peripheral muscles and resultant
changes in skin conductance, facial expression and posture (i.e., the chameleon effect: Chartrand
& Bargh, 1999). The process continues with somatic input from the periphery to the
somatosensory brain, which sends its interpretation of the signals to TPJ (Matsuhashi, Ikeda,
(Ohara, et al., 2004). The TPJ assesses the somatosensory evidence for any possibility that the
experienced intention belongs to the self. Such evidence could potentially come from the
changes in, for example, motor evoked potentials associated with mirroring, with registration of
stronger potentials being associated with stronger mirroring. This line of reasoning leads to the,
possibly, counter-intuitive idea that increased MN activation leads to increased peripheral
stimulation. This effect would result in stronger peripheral input to somatosensory cortex and
subsequent delivery of stronger signals to the TPJ which would have a resulting increased
likelihood of erroneously modeling the sensory input it is receiving as resulting from the self
having executed the action rather than appropriately attributing the action to the observed other.
It seems likely that this binary decision (i.e. either MNs constitute action understanding,
or they have nothing to do with it) is but one of many examples of what Richard Dawkins (2004)
calls the tyranny of the discontinuous mind the tendency to see continuous phenomena in
discrete units. A good example of this would be the emerging conception of gender. Historically
gender has been viewed as binary by many societies, but modern views increasingly regard male
and female but two ends of a continuum with the majority of individuals falling somewhere in
between those extremes. Similarly, intention understanding is unlikely to be based solely on one
or the other of these information processing routes. If DP models are taken to mean that mirror
neurons and only mirror neurons generate intention understanding, then such models are non-
starters. As with most psychological processes, intention understanding is unlikely to rely on
some modular node in a serial processing assembly line. Rather, any model that incorporates a
role of mirror neurons is more likely to succeed by assuming that they contribute an embodied
dimension to understanding other’s intentions.
Intention understanding implies understanding to whom the attribution should be made.
If one experiences an intention but attributes it to no individual, then one does not understand the
intention. An intention is such by virtue of it being the strategic cause of a particular person’s
behavior. If there is no associated behavior then the neural activity we would otherwise call an
intention is intention-less. Therefore, intention understanding at a minimum requires agency
attribution. Models of agency attribution are generally cognitive and inferential in nature. If
agency attribution is an inferential process and a prerequisite for intention understanding, then
strict direct perception models could not be correct. But this of course does not imply that they
do not make important contributions to action understanding, giving an embodied dimension to
one’s understanding of another.
While it is difficult to imagine that the agent-attribution necessary for action
understanding does not require inference, Michael and de Bruin (2015) discuss the observability
assumption, which they claim is an assumption made by inferential models that intentions are not
perceivable, or observable. By claiming that action understanding cannot be directly perceived,
these models are vulnerable to naïve dualism, the notion that there are physical causes to some
behaviors while others have non-physical, intentional causes. Of course, dualism is something
that does not comport with science since science depends on the assumption that the world is
systematic and orderly (e.g. obeys the laws of physics), while any possible non-physical world,
by definition, would be unsystematic and unorderly, not constrained by the laws of physics, and
therefore unknowable. Based on this, evaluation of inferential models should include questions
about whether the model is dualistic.
Self-Other Discrimination in the MN System.
Regardless of whether MNs are sufficient for action understanding, if they relate to
intentions, then having them activated by other persons poses a problem: How does the brain
determine on which intentions to act? In other words, how does the brain discriminate between
those intentions derived from one’s beliefs, goals and desires and those derived from observing a
conspecific acting on said intentions? Self-other discrimination therefore is necessary to enable
one to act on those intentions directly relevant to his survival. Indeed, it is even necessary for one
to have a sense of self-identity to have a personality. Consider that one’s personality is partially
determined by the types of intentions he tends to have in specific situations make fun of his
mother, and he’ll have the intention to smack you. If one acted not only on his own intentions but
also on those of others within his view, the intentions-related part of his personality would be
indistinguishable from the personalities of those around him. Given the importance of acting on
one’s own intentions and not on those of others, what follows is a discussion of what we know
about MNs and self-other discrimination.
The original report of the discovery of MNs demonstrated that MNs are more active
when the monkey executes, compared to observes, an action (di Pellegrino et al., 1992). A
potentially similar pattern has been reported using EEG mu suppression. Woodruff and Maaske
(2010) reported reliably more mu suppression to executed, compared to observed actions, with
mu suppression in both conditions being significantly greater than zero. Woodruff, Martin and
Bilyk (2011) found that this self-other difference in mu suppression was positively correlated
with self-reported perspective-taking as measured by the Interpersonal Reactivity Index (Davis,
Consistent with these findings, Maranesi et al. (2012) found that F5c plays a less direct
role in motor movement than other nearby regions of the vPMC. In particular, Maranesi et al.
found that electrical stimulation is less effective at exciting F5c neurons than neurons in
surrounding areas. Furthermore, while areas like F4 and F5p have relatively direct connections to
spinal motor neurons and inactivation of these regions impairs movement, F5c’s projections are
less direct and inactivation does not seem to suppress movement. These two findings together
suggest one possible neural mechanism by which one is able to avoid erroneously acting on the
intentions of others as coded by MNs. By “funneling” mirrored actions to a region without direct
access to execution stages of the motor hierarchy, risk of other’s-intended actions being executed
is minimized.
Citing evidence that F5 and AIP/PFG (as well as another potential MN region,
ventrolateral prefrontal cortex (vlPFC)) are interconnected with the putamen, part of the basal
ganglia (BG), Bonini (2016) suggests that MNs might even be found there as well. He further
posits that motor output is modulated via activation of the indirect or the hyperdirect pathways of
the BG. Activation of this cortico-BG circuit constitutes a possible mechanism whereby the
activation of the motor system by the observation of others is down-regulated to minimize
unwanted imitation. In other words, such an extended network, if it exists, may reflect self-other
discrimination and avoid the observer acting on the intentions of others.
As Yoshida et al. (2011) point out, MNs do not themselves process self-other distinctions
and a further mechanism to perform this discriminatory process is needed. Similarly, Rozzi and
Coudé (2015) rightly point out that the problem posed by MNs is self-other confusion: If
observation of another’s actions is transformed into an intention in the observer’s mind, the
observer is at risk of acting as though the intention were his own. Without self-other
discrimination, groups of humans might act less like individuals and more like schooling
sardines, with one person’s action resulting in the rapid transmission of the associated intention
throughout the group. Each individual human would simply experience said intention and act on
it as though it were his own. Indeed Rozzi and Coudé suggested that S-O discrimination is likely
to require multiple areas, some not known to contain MNs. Although Rozzi and Coudé appear
favorable to the theory that MNs constitute action understanding, this theory is logically
unconfirmable by virtue of its failure to account for S-O discrimination. While it seems likely
that MNs contribute to the emergent process of action understanding, it does not seem that MN
activation on its own could constitute action understanding without self-other discrimination.
However, Rozzi and Coudé (2015) suggest that F5c neurons are particularly related to
agency. Whereas F5a and F5p neurons respond to observation of videos of actions in minimal
contexts (i.e. only the hand and target object were visible), F5c appears to become involved only
when the entire person can be seen (Nelissen et al., 2005). Rozzi and Coudé take this finding to
imply that F5c activation does not simply reflect actions in general, but actions as a function of
the agent executing them. An interesting hypothesis derives from this finding and the S-O
discrimination hypothesis mentioned above, which claims that S-O discrimination in MN activity
arises only after reentrant input from regions like the TPJ. This hypothesis suggests that initial
activity in MNs does not discriminate self from other but only takes on this characteristic later in
the timecourse of the neural activity, once TPJ has had sufficient time to perform its
computations and to indicate to MNs who is most likely the agent of the action.
Conclusions and Directions for Future Research
The discovery of mirror neurons has been one of the more exciting developments in the
neurosciences, contributing to the growth of the field of social neuroscience. While much of this
excitement is warranted, the impulse to over-ascribe functions to MNs should be treated with
caution. It is argued here that mirroring a conspecific’s actions likely contributes to
understanding the intention behind those actions, but that the activation of MNs on their own
should not be considered sufficient for action understanding. Future research into the relationship
between MNs and action understanding needs to occur on multiple fronts. Theoretically, DP and
IP models need to be further developed with DP models addressing the S-O discrimination issue
and IP models addressing whether it’s reasonable to assume MNs play no role in action
As discussed here and by others, measuring MNs through non-invasive methods whose
signals sum across tens of thousands of neurons have inherent limitations in their ability to make
strong inferences about MN activity since MNs are defined by the characteristics with a single
neuron. Since available neuroimaging techniques cannot resolve the activity of a single neuron,
they cannot, strictly speaking, unambiguously measure MNs. However, use of repetition
suppression or fMRI adaptation procedures can yield much greater inferential power as regards
MNs. Future research should strive to use such methods where possible. We suggest, however,
that methods that do not admit to unequivocal inferences about MNs should not be taken as
useless. For one, methods such as LaPlacian spatial transformation and Second Order Blind
Separation can be used to clean-up mu suppression data, potentially allowing finer spatial
resolution of the distribution of rhythms across the scalp, thereby allowing greater confidence
that mu rhythms are not unduly influenced by occipital alpha. Similarly, fMRI studies that do not
lend themselves to RS could seek to anchor inference of MN activation by systematically
modulating specific characteristics of MN function, such as self-other discrimination effects.
Given the challenges of measuring MN activity with neuroimaging, it would likely be productive
to minimize conclusions about the imaging signal’s precise relationship to MNs and focus more
on the signal’s apparent relationship to the construct of interest (e.g. empathy, action
understanding, language, etc.). That is to say, that many interesting conclusions can be drawn
about this latter relationship without getting bogged down in controversy about whether the
signal specifically reflects MNs.
As for direct measures of MNs, future research should continue to assess the extent to
which MNs are necessary and/or sufficient for action understanding. It is our contention that they
are not sufficient, but may well be necessary (or at least very useful) to the kind of action
understanding that occurs when multiple levels of action representation, from visual descriptions
to motor programs, reciprocally interact to yield hypotheses, error predictions and corrections in
an ongoing, dynamic process of understanding the actions of others, generating inferences about
how those actions relate to self, and selecting appropriate responses.
Balconi, M., & Lucchiari, C. (2006). EEG correlates (event-related desynchronization) of
emotional face elaboration: A temporal analysis. Neuroscience Letters, 392, 118-123.
Baron-Cohen, S., & Wheelwright S (2004). The empathy quotient: An investigation of adults
with Asperger Syndrome or High Functioning Autism, and normal sex differences.
Journal of Autism and Developmental Disorders, 34, 163-175.
Başar, E., Özgören, M., Öniz, A., Schmiedt, C., Başar-Eroglu, C. (2007). The brain oscillations
differentiate the picture of the own grandmother. International Journal of
Psychophysiology, 64, 81-90.
Batson, C.D., Batson, J.G., Slingsby, J.K., Harrell, K.L., Peekna, H.M., & Todd, M. (1991).
Empathic joy and the empathy-altruism hypothesis. Journal of Personality and Social
Psychology, 61, 413-426.
Bonini, L. (2014). The extended mirror neuron network: Anatomy, origin, and functions. The
Neuroscientist, 23, 1-12.
Braadbaart, L., Williams, J., Waiter, G. (2013) Do mirror neuron areas mediate mu rhythm
suppression during imitation and action observation? International Journal of
Psychophysiology, 89, 99-105.
Brown. H., Woodruff, C. C., Goodman, A. (in prep). Effect of familiarity on EEG b-rhythm
Activity and empathic accuracy.
Caggiano, V., Fogassi, L., Rizzolatti, G., Pomper, J.K., Their, P., Giese, M.A., Casile, A. (2011).
View-based encoding of actions in mirror neurons of area f5 in macaque premotor cortex.
Current Biology, 21, 1448.
Caggiano, V., Fogassi, L., Rizzolatti, G., Their, P., Casile, A. (2009). Mirror neurons
differentially encode the peripersonal and extrapersonal space of monkeys. Science, 324,
Catmur, C. (2015). Understanding intentions from actions: Direct perception, inference and the
roles of mirror and mentalizing systems. Consciousness and Cognition, 36, 426-433.
Chartrand, T.L., Bargh, J.A. (1999). The chameleon effect: The perception-behavior linkin social
interaction. Journal of Personality and Social Psychology, 76, 893-910.
Cheng, Y., Tzeng, O.J.L, Decety, J., Imada, T., & Hsieh, J. (2006). Gender differences in the
human mirror system: A magnetoencephalography study. Cognitive Neuroscience and
Neurophysiology, 17, 1115-1119.
Cook, R., Bird, G., Catmur, C., Press, C., Heyes, C. (2014). Mirror neurons: From origin to
function. Behavioral and Brain Sciences, 37, 177-241.
Csibra, G. (2008). Action mirroring and action understanding: An alternative account. In P.
Haggard, Y. Rossetti and M. Kawato (Eds.), Sensorimotor foundation of higher
cognition: Attention and performance, (pp. 435458). Oxford: Oxford University Press.
Davis, M.H. (1983). Measuring individual differences in empathy: Evidence for a
multidimensional approach. Journal of Personality and Social Psychology, 44, 113-126
Dawkins, R. (2004). The ancestor’s tale: A pilgrimage to the dawn of evolution. Houghton,
Mifflin and Harcourt: NY, NY.
di Pellegrino, G., Fadiga, L., Fogassi, L., Gallese, V., & Rizzolatti, G. (1992). Understanding
motor events: A neuropsychological study. Experimental Brain Research, 91, 176 180.
Dinstein, I., Thomas, C., Behrman, M., & Heeger, D.J. (2008). A mirror up to nature. Current
Biology, 18, R13-R18.
Ekman, P., Friesen, W.V. (1976). Pictures of facial affect. Consulting Psychologists Press, Palo
Alto, CA.
Fadiga, L., Fogassi, L., Pavesi, G., & Rizzolatti, G. (1995). Motor facilitation during action
observation: A magnetic stimulation study. Journal of Neurophysiology, 73, 26082611.
Fox, N.A., Bakermans-Kranenburg, M.J., Yoo, K.H., Bowman, L.C., Vanderwert, R.E., Ferrari,
P.F., & van Ijzendoorn., M.H. (2015). Assessing human mirror activity with EEG mu
rhythm: A meta-analysis. Psychological Bulletin, 142, 291-313.
Gallese, V., Fadiga, L., Fogassi, L., Rizzolatti, G. (1996). Action recognition in the premotor
cortex. Neurology. 119, 593609.
Gallese, V. (2005). Embodied simulation: From neurons to phenomenal experience. Phenom
Cognitive Science, 4, 23-48.
Gallese, V. (2006). Intentional attunement: A neurophysiological perspective on social cognition
and its disruption in autism. Brain Research, 1079, 15-24.
Gallese, V., Gernsbacher, M.A., Heyes, C., Hickok, G., Iacoboni, M. (2011). Mirror neuron
forum. Perspectives on Psychological Science, 6, 369-407.
Giese, M.A., Rizzolatti, G. (2015). Neural and computational mechanisms of action processing:
interaction between visual and motor representations. Neuron, 88, 16780.
Güntekin, B., Başar, E. (2010). Event-related β oscillations are affected by emotional eliciting
stimuli. Neuroscience Letters 483, 173-178.
Güntekin, B., Başar, E., (2009). Facial affect manifested by multiple oscillations. International
Journal of psychophysiology 71, 31-36.
Güntekin, B., and Başar, E. (2007). Emotional face expressions are differentiated with brain
oscillations. International Journal of Psychophysiology 64, 91-100.
Güntekin, B., Tülay, E., 2014. Event related beta and gamma oscillatory responses during
perception of affective pictures. Brain Research, 1577, 1-12.
Hari, R. & Salmelin, R. (1997). Human cortical oscillations: A neuromagnetic view through the
skull. Trends in Neurosciences, 20, 44-49.
Hari, R.R., & Forss, N.N. (1998). Activation of human primary motor cortex during action
observation: A neuromagnetic study. Proceedings of the National Academy of Sciences of
the United States of America, 95, 1506115065.
Henson, R.N.A. (2003). Neuroimaging studies of priming. Progress in Neurobiology, 70, 53-81.
Heinisch, C., Krüger, M.C., & Brüne, M. (2012). Repetitive transcranial magnetic stimulation
over the temporoparietal junction influences distinction of self from famous but not
unfamiliar others. Behavioral Neuroscience, 126, 792-796.
Hickok, G. (2009). Eight problems for the mirror neuron theory of action understanding in
monkeys and humans. Journal of Cognitive Neuroscience, 21, 1229-1243.
Hobson, H.M., Bishop, D.V.M. (2016). Mu suppression - a good measure of the human mirror
neuron system? Cortex, 82, 290310.
Hobson, H.M., & Bishop, D.V.M. (2017). The interpretation of mu suppression as an index of
mirror neuron activity: past, present and future. Royal Society Open Science, 2, 1-22.
Hoenen, M., Schain, C., Pause, B.M. (2013). Down-modulation of mu-activity through empathic
top-down processes. Social Neuroscience, 8, 515-524.
Horan, W., Pineda, J., Wynn, J., Iacoboni, M., Green, M., (2014). Cognitive Affective and
Behavioral Neuroscience 4, 1049-60.
Iacoboni M, Woods R, Brass M, Bekkering H, Mazziotta J, Rizzolatti G. (1999). Cortical
mechanisms of human imitation. Science; 28,: 2526-2528.
Jun, S.C., George, J.S., Paré-Blagoev, J., Plis, S.M., Ranken, D.M., Schmidt, D.M., Wood, C.C.
(2005). Spatiotemporal Bayesian inference dipole analysis for MEG neuroimaging data.
Neuroimage, 28, 84-98.
Keysers, C., Gazzola, V. (2009). Expanding the mirror: vicarious activity for actions, emotions
and sensations. Current Opinion Neurobiology, 19, 666-671.
Kilner, J.M., Friston, K.J., Frith, C.D. (2007). The mirror-neuron system: a Bayesian perspective.
NeuroReport, 18, 619-623.
Kilner, J.M., Neal, A., Weiskopf, N., Friston, K.J., Frith, C.D. (2009). Evidence of mirror
neurons in human inferior frontal gyrus. Journal of Neuroscience, 29, 10153-10159.
Kohler E, Keysers C, Umiltà MA, Fogassi L, Gallese V, Rizzolatti G. (2002) Hearing sounds,
understanding actions: action representation in mirror neurons. Science 297, 846848.
Logothetis, N.K., Pauls, J., Augath, M., Trinath, T., Oeltermann, A. (2001). Neurophysiological
investigation of the basis of the fMRI signal. Nature, 412, 150-157.
Maeda, K., Ishida, H., Nakajima, K., Inase, M., Murata, A. (2015). Functional properties of
parietal hand manipulation-related neurons and mirror neurons responding to vision of
own hand action. Journal of Cognitive Neuroscience, 27, 56072.
Maranesi, M., Roda, F., Bonini, L., Rozzi, S., Ferrari, P.F., Fogassi, L., & Coudé, G. (2012).
Anatomo-functional organization of the ventral primary motor and premotor cortex in the
macaque monkey. European Journal of Neuroscience, 36, 3376-3387.
Matsuhashi, M., Ikeda, A., Ohara, S., Matsumoto, R, Yamamoto, J., Takayama, M., Satow, T.,
Begum, T., Usui, K., Nagamine, T., Mikuni, N., Takahashi, J., Miyamoto, S., Fukuyama,
H., & Shibasaki, H. (2004). Multisenosry convergence at human temporo-parietal
junction epicortical recording of evoked responses. Clinical Neurophysiology, 115,
Michael, J. (2011). Four models of the functional contribution of mirror systems. Philosophical
Explorations, 14, 185-194.
Michael, J., & De Bruin, L. (2015). How direct is social perception? Consciousness and
Cognition, 36, 373-375.
Mizuguchi, N., Nakata, H., Kanosue, K. (2016). The right temporoparietal junction encodes
efforts of others during action observation. Scientific Reports, 6, 1-8.
Mukamel, R., Ekstrom, A.D., Kaplan, J., Iacoboni, M., Fried, I. (2010). Single-neuron responses
in humans during execution and observation of actions. Current Biology, 20, 750-756.
Muthukumaraswamy, S., D., & Johnson, B., W., (2004). Changes in rolandic mu rhythm during
observation of a precision grip. Psychophysiology, 41, 152-156.
Muthukumaraswamy, S., Johnson, B. (2004) Primary motor cortex activation during action
observation revealed by wavelet analysis of the EEG. Clinical Neurophysiology, 115,
Nelissen, K., Borra, E., Gerbella, M., Rozzi, S., Luppino, G., Vanduffel, W., Rizzolatti, G.,
Orban, G.A. (2011). Action observation circuits in the macaque monkey cortex. Journal
of Neuroscience, 31, 374356.
Oberman, L., Hubbard, E., McCleary, J., Altschuler, E., Ramachandran, V., & Pineda, J. (2005).
EEG evidence for mirror neuron dysfunction in autism spectrum disorders. Cognitive
Brain Research, 24, 190 198.
Pfurtscheller, G., Neuper, C., Andrew, C., Edlinger, G. (1997). Foot and hand area mu rhythms.
International Journal of Psychophysiology, 26, 121-135.
Peeters, R., Simone, L., Nelissen, K., Fabbri-Destro, M., Vanduffel, W., Rizzolatti, G., & Orban,
G.A. (2009). Journal of Neuroscience, 29, 11523-11539.
Perry, A., Troje, N., F., & Bentin, S., (2010). Exploring motor system contributions to the
perception of social information: Evidence from EEG activity in the mu/alpha frequency
range. Social Neuroscience, 5, 272-284.
Pineda. J.A. (2005). The functional significance of μ rhythms: translating ‘seeing’ and ‘hearing’
into ‘doing’. Brain Research Review, 699, 5768.
Rizzolatti, G., Camarda, R., Fogassi, L., Gentilucci, M., Luppino, G., Matelli, M. (1988)
Functional organization of inferior area 6 in the macaque monkey: II. Area F5 and the
control of distal movements. Exp. Brain Res. 71, 491507.
Rizzolatti, G., Sinigaglia, C. (2010). The functional role of the parieto-frontal mirror circuit:
Interpretations and misinterpretations. Nature Reviews Neuroscience, 11, 264-274.
Rizzolatti, G., Fogassi, L. (2014). The Mirror Mechanism: recent findings and perspectives.
Philosophical Transactions of the Royal Society of London: B Biological Science. 369, 1-
Rozzi, S., Ferrari, P.F., Bonini, L., Rizzolatti, G., Fogassi, L. (2008). Functional organization of
inferior parietal lobule convexity in the macaque monkey: electrophysiological
characterization of motor, sensory and mirror responses and their correlation with
cytoarchitectonic areas. European Journal of Neuroscience, 28, 1569-1588.
Rozzi, S., Coudé, G. (2015). Grasping actions and social interaction: neural bases and anatomical
circuitry in the monkey. Frontiers in Psychology, 6, 1-19.
Schülte-Ruther, M., Markowitsch, H., Fink, G., & Piefke, M. (2007). Mirror neuron and theory
of mind mechanisms involved in face-to-face interactions: A functional magnetic
resonance imaging approach to empathy. Journal of Cognitive Neuroscience, 19, 1354-
Simon, S., Mukamel, R. (2015a). Power modulation of electroencephalogram mu and beta
frequency depends on perceived level of observed actions. Brain and Behavior, 6, 1-11.
Simon, S., Mukamel, R. (2015b). Sensitivity to perception level differentiates two subnetworks
within the mirror neuron system. Social Cognitive and Affective Neuroscience, 6, 1-11.
Southgate, V., & Hamilton, A.F. (2008). Unbroken mirrors: Challenging a theory of Autism.
Trends in Cognitive Science, 12, 225-229.
Tang, A.C., Sutherland, M.T., & McKinney, C.J. (2005).Validation of SOBI components from
high-density EEG. Neuroimage 25, 539553.
Tsuchiya, N., Koch, C. (2005). Continuous flash suppression reduces negative afterimages.
Nature Neuroscience, 8, 1096101.
Umiltà, M.A., Escola, L., Intskirveli, I., Grammont, F., Rochat, M., Caruana, F., Jezzini, A.,
Gallese, V., & Rizzolatti, G. (2008) When pliers become fingers in the monkey motor
system. Proceedings of the National Academy of Science, 105, 2209-13.
Woodruff, C.C., Barbera, D., Von Oepen, R. (2016). Task-related dissociation of EEG β
enhancement and suppression. International Journal of Psychophysiology, 99,18-23.
Woodruff, C.C., Barbera, D., Von Oepen, R. (2016). Task-related dissociation of EEG β
enhancement and suppression. International Journal of Psychophysiology, 99,18-23.
Woodruff, C. C., & Maaske, S., (2010). Action execution engages human mirror neuron system
more than action observation. NeuroReport, 21, 432-435.
Woodruff, C.C., Martin, T., Bilyk, N. (2011a). Differences in self- and other-induced mu
suppression are correlated with empathic abilities. Brain Research, 1405, 69-76.
Woodruff, C.C., Daut R, Brower M, Bragg A (2011b). EEG alpha- and beta-band correlates of
perspective-taking and personal distress. NeuroReport 22, 744-748.
Woodruff, C. C. Goodman, A., Brown, H. (in prep). Self-other differences in EEG µ-suppression
reflect familiarity and perspective-taking.
Yang, C.-Y., Decety, J., Lee, S., Chen, C., Cheng, Y. (2009). Gender differences in the mu
rhythm during empathy for pain: An electroencephalographic study. Brain Research,
1251, 176-184.
Yin, S., Liu, Y., & Mingzhou, D. (2016). Amplitude of sensorimotor mu rhythm is correlated
with BOLD from multiple brain regions: A simultaneous EEG-fMRI Study. Frontiers in
Human Neuroscience, 10, 1-12.
Zaki, J., Weber, J., Bolger, N., & Ochsner, K. (2009). The neural bases of empathic accuracy.
PNAS Proceedings of the National Academy of Sciences of the United States of America
106, 11382-11387.
Figure Captions
Figure 1. Regions in the macaque brain believed to be part of the mirror neuron system, as well
as the pSTS, believed to provide input to the mirror neuron system.
Figure 2. Mu suppression self-other difference scores from Woodruff et al. (2011a). Difference
scores were significantly different than zero for the central electrodes (all p’s < .001), but not for
occipital electrodes. Furthermore, difference scores related positively to self-reported
perspective-taking. Error bars represent 2 SEM.
Figure 3. Beta-band (14-30Hz; subdivided into 2Hz bins) suppression/enhancement from central
electrodes from three different experiments: Woodruff et al., 2016; Brown et al., in preparation;
Goodman et al., in preparation.
Full-text available
Death of the shame Summary Shame and guilt together with pride create a control system - a deference-emotion system - serving to protect social relations. The reasons for shame are to a large extent learned, specific to a given culture and can be regarded as a memetic phenomenon. The causes of shame can often be reduced to deviations from the culturally accepted norm or loss of intimacy. Thus, the loss of shame - the social death of shame - can have undesirable consequences for the functioning of individuals and society. The neurobiological background of shame is based on the theory of mind as a psychic function proper to man and perhaps a few species of animals. It is expanded by intentionality, understood as the ability to multi-level, recurrent (chain) modelling of the state of mind of subsequent individuals. The basis of these functions are the areas of the large cognitive circuit of the brain including the prefrontal cortex, the cingulum, some areas of the parietal, occipital and temporal cortex and the insula and amygdala. The shame and related guilt reveal themselves in the different activities of these structures. The reasons for the death of shame in modern society can be seen in four hypothetical areas of social relations, provisionally described as: 1. Ease and cheapness of nonsense; 2. Demystification of elite behavioural patterns; 3. Devaluation of education and culture; 4. Erroneus addressing the message and long maturation of critical wisdom. Examples are commonly observed and discussed phenomena, such as: much greater propagation of false and harmful information but affecting emotion; revealing in social communication the truth about the cultural patterns of the elite in connection with the resignation from intimacy; lowering the standards of requirements in education due to its easy and mass availability; presentation in works of art, mainly in cinema pictures, forms, contents and problems to which some of the recipients have not been matured, which results in changes in sensitivity and ethical attitudes. Key words signs of shame, reasons of shame, neurobiology of shame, immodesty, hypothes on reasons of the immodesty
Full-text available
Mirror neurons are a subset of brain cells that discharge during action execution and passive observation of similar actions. An open question concerns the functional role of their ability to match observed and executed actions. Since understanding of goals requires conscious perception of actions, we expect that mirror neurons potentially involved in action goal coding, will be modulated by changes in action perception level. Here, we manipulated perception level of action videos depicting short hand movements and measured the corresponding fMRI BOLD responses in mirror regions. Our results show that activity levels within a network of regions, including the sensorimotor cortex, primary motor cortex, dorsal premotor cortex and posterior superior temporal sulcus, are sensitive to changes in action perception level, whereas activity levels in the inferior frontal gyrus, ventral premotor cortex, supplementary motor area and superior parietal lobule are invariant to such changes. In addition, this parcellation to two sub-networks manifest as smaller functional distances within each group of regions during task and resting state. Our results point to functional differences between regions within the mirror neurons system which may have implications with respect to their possible role in action understanding.
Full-text available
Mu suppression studies have been widely used to infer the activity of the human mirror neuron system (MNS) in a number of processes, ranging from action understanding, language, empathy and the development of autism spectrum disorders (ASDs). Although mu suppression is enjoying a resurgence of interest, it has a long history. This review aimed to revisit mu's past, and examine its recent use to investigate MNS involvement in language, social processes and ASDs. Mu suppression studies have largely failed to produce robust evidence for the role of the MNS in these domains. Several key potential shortcomings with the use and interpretation of mu suppression, documented in the older literature and highlighted by more recent reports, are explored here.
Full-text available
Smooth social interactions require a deep understanding of others’ intentions and feelings. In the present study, to investigate brain regions that respond to inference of others’ effort level, we recorded brain activity during action observation of different effort levels using functional magnetic resonance imaging (fMRI). We used a dumbbell curl movement to depict a movement requiring effort. To dissociate the factors of effort level of the actor and weight of the dumbbell, we used four combinations of dumbbell weight and actor physique: a thin actor or a built actor lifting a heavy or light dumbbell. During observation of dumbbell curls, the bilateral front-parietal action observation network (AON) was activated. This included the premotor cortices, parietal cortices, visual areas 5/superior temporal cortices (STS), amygdalae, hippocampi, right dorsolateral and ventrolateral frontal cortices. When we evaluated brain regions associated with the actor’s effort level, activity in the right temporoparietal junction (TPJ) and STS was observed. However, activity in the front-parietal AON was independent of the actor’s effort during action observation. This finding suggests that the right TPJ and STS play an important role in the inference of others’ effort levels during the observation of others’ movements.
Full-text available
Introduction The ability to understand actions and intentions of others is of great importance to social relationships and is associated with the mirror neuron system of the human brain. Whether conscious perception of specific actions is necessary to trigger activity in this system, or alternatively whether this response is independent of conscious perception is not known. Methods We addressed this issue by rendering videos of right hand movements invisible to conscious perception, and measuring electroencephalogram (EEG) power suppression in the mu (8–13 Hz) and beta (15–25 Hz) range as index corresponding to the magnitude of mirror neuron activity. Results In the beta range over bilateral sensorimotor sites, we find that suppression indices follow the reported perceptual level of subjects with stronger suppression for consciously perceived trials. Furthermore, in the nonperceived trials, oscillation power is significantly suppressed relative to baseline. In the low mu range (8–10 Hz), oscillation power over the left sensorimotor site is significantly more suppressed in the consciously perceived versus nonperceived trials. Conclusions Our data suggest that the intensity of mirror system responses during action observation decreases with the observers' perception level yet remains significant during observation of invisible actions. Such subliminal activity could help explain phenomena such as covert imitation.
Full-text available
Mu suppression has been proposed as a signature of the activity of the human mirror neuron system (MNS). However the mu frequency band (8–13 Hz) overlaps with the alpha frequency band, which is sensitive to attentional fluctuation, and thus mu suppression could potentially be confounded by changes in attentional engagement. The specific baseline against which mu suppression is assessed may be crucial, yet there is little consistency in how this is defined. We examined mu suppression in 61 typical adults, the largest mu suppression study so far conducted. We compared different methods of baselining, and examined activity at central and occipital electrodes, to both biological (hands) and non-biological (kaleidoscope) moving stimuli, to investigate the involvement of attention and alpha activity in mu suppression. We also examined changes in beta power, another candidate index of MNS engagement. We observed strong mu suppression restricted to central electrodes when participants performed hand movements, demonstrating that mu is indeed responsive to the activity of the motor cortex. However, when we looked for a similar signature of mu suppression to passively observed stimuli, the baselining method proved to be crucial. Selective suppression for biological versus non-biological stimuli was seen at central electrodes only when we used a within-trial baseline based on a static stimulus: this method greatly reduced trial-by-trial variation in the suppression measure compared with baselines based on blank trials presented in separate blocks. Even in this optimal condition, 16–21% of participants showed no mu suppression. Changes in beta power also did not match our predicted pattern for MNS engagement, and did not seem to offer a better measure than mu. Our conclusions are in contrast to those of a recent meta-analysis, which concluded that mu suppression is a valid means to examine mirror neuron activity. We argue that mu suppression can be used to index the human MNS, but the effect is weak and unreliable and easily confounded with alpha suppression.
Full-text available
Illusions that produce perceptual suppression despite constant retinal input are used to manipulate visual consciousness. Here we report on a powerful variant of existing techniques, continuous flash suppression. Distinct images flashed successively at approx10 Hz into one eye reliably suppress an image presented to the other eye. The duration of perceptual suppression is at least ten times greater than that produced by binocular rivalry. Using this tool we show that the strength of the negative afterimage of an adaptor was reduced by half when it was perceptually suppressed by input from the other eye. The more completely the adaptor was suppressed, the more strongly the afterimage intensity was reduced. Paradoxically, trial-to-trial visibility of the adaptor did not correlate with the degree of reduction. Our results imply that formation of afterimages involves neuronal structures that access input from both eyes but that do not correspond directly to the neuronal correlates of perceptual awareness.
The mu rhythm is a field oscillation in the ∼10Hz range over the sensorimotor cortex. For decades, the suppression of mu (event-related desynchronization) has been used to index movement planning, execution, and imagery. Recent work reports that non-motor processes, such as spatial attention and movement observation, also desynchronize mu, raising the possibility that the mu rhythm is associated with the activity of multiple brain regions and systems. In this study, we tested this hypothesis by recording simultaneous resting-state EEG-fMRI from healthy subjects. Independent component analysis (ICA) was applied to extract the mu components. The amplitude (power) fluctuations of mu were estimated as a time series using a moving-window approach, which, after convolving with a canonical hemodynamic response function (HRF), was correlated with blood-oxygen-level-dependent (BOLD) signals from the entire brain. Two main results were found. First, mu power was negatively correlated with BOLD from areas of the sensorimotor network, the attention control network, the putative mirror neuron system, and the network thought to support theory of mind. Second, mu power was positively correlated with BOLD from areas of the salience network, including anterior cingulate cortex and anterior insula. These results are consistent with the hypothesis that sensorimotor mu rhythm is associated with multiple brain regions and systems. They also suggest that caution should be exercised when attempting to interpret mu modulation in terms of a single brain network.
To facilitate a multidimensional approach to empathy the Interpersonal Reactivity Index (IRI) includes 4 subscales: Perspective-Taking (PT) Fantasy (FS) Empathic Concern (EC) and Personal Distress (PD). The aim of the present study was to establish the convergent and discriminant validity of these 4 subscales. Hypothesized relationships among the IRI subscales between the subscales and measures of other psychological constructs (social functioning self-esteem emotionality and sensitivity to others) and between the subscales and extant empathy measures were examined. Study subjects included 677 male and 667 female students enrolled in undergraduate psychology classes at the University of Texas. The IRI scales not only exhibited the predicted relationships among themselves but also were related in the expected manner to other measures. Higher PT scores were consistently associated with better social functioning and higher self-esteem; in contrast Fantasy scores were unrelated to these 2 characteristics. High EC scores were positively associated with shyness and anxiety but negatively linked to egotism. The most substantial relationships in the study involved the PD scale. PD scores were strongly linked with low self-esteem and poor interpersonal functioning as well as a constellation of vulnerability uncertainty and fearfulness. These findings support a multidimensional approach to empathy by providing evidence that the 4 qualities tapped by the IRI are indeed separate constructs each related in specific ways to other psychological measures.
Empathy is an essential part of normal social functioning, yet there are precious few instruments for measuring individual differences in this domain. In this article we review psychological theories of empathy and its measurement. Previous instruments that purport to measure this have not always focused purely on empathy. We report a new self-report questionnaire, the Empathy Quotient (EQ), for use with adults of normal intelligence. It contains 40 empathy items and 20 filler/control items. On each empathy item a person can score 2, 1, or 0, so the EQ has a maximum score of 80 and a minimum of zero. In Study 1 we employed the EQ with n = 90 adults (65 males, 25 females) with Asperger Syndrome (AS) or high-functioning autism (HFA), who are reported clinically to have difficulties in empathy. The adults with AS/HFA scored significantly lower on the EQ than n = 90 (65 males, 25 females) age-matched controls. Of the adults with AS/HFA, 81% scored equal to or fewer than 30 points out of 80, compared with only 12% of controls. In Study 2 we carried out a study of n = 197 adults from a general population, to test for previously reported sex differences (female superiority) in empathy. This confirmed that women scored significantly higher than men. The EQ reveals both a sex difference in empathy in the general population and an empathy deficit in AS/HFA.
A fundamental issue in cognitive neuroscience is how the brain encodes others' actions and intentions. In recent years, a potential advance in our knowledge on this issue is the discovery of mirror neurons in the motor cortex of the nonhuman primate. These neurons fire to both execution and observation of specific types of actions. Researchers use this evidence to fuel investigations of a human mirror system, suggesting a common neural code for perceptual and motor processes. Among the methods used for inferring mirror system activity in humans are changes in a particular frequency band in the electroencephalogram (EEG) called the mu rhythm. Mu frequency appears to decrease in amplitude (reflecting cortical activity) during both action execution and action observation. The current meta-analysis reviewed 85 studies (1,707 participants) of mu that infer human mirror system activity. Results demonstrated significant effect sizes for mu during execution (Cohen's d = 0.46, N = 701) as well as observation of action (Cohen's d = 0.31, N = 1,508), confirming a mirroring property in the EEG. A number of moderators were examined to determine the specificity of these effects. We frame these meta-analytic findings within the current discussion about the development and functions of a human mirror system, and conclude that changes in EEG mu activity provide a valid means for the study of human neural mirroring. Suggestions for improving the experimental and methodological approaches in using mu to study the human mirror system are offered. (PsycINFO Database Record