66.5 F
New York
Tuesday, May 28, 2024

Mirror Neurons in Macaque Monkeys Activate During Action Observation

Must read

A team of researchers has recently shown that mirror neurons in macaque monkeys activate during action observation. This discovery radically changes our understanding of the motor system’s role in social cognition.

These results indicate that mirror neurons play a central role in our ability to understand others’ actions, read their intentions and program contextually appropriate motor responses.


Focused ultrasound (FUS) has gained recognition as a non-invasive neuromodulation technique with high spatial precision. It can excite precise target neurons and evoke downstream activations in off-target somatosensory and associated brain regions, which are simultaneously detected by functional MRI. Moreover, FUS is believed to act through a synaptic pathway that is similar to that of natural tactile stimulation. We explored the neural effects of FUS on monkey touch processing.

We used a combination of acoustic and MR-guided focused ultrasound to elicit fMRI BOLD activations in macaque hemispheres that were stimulated with 8 Hz stimulus amplitudes of distal finger pads from digits 2&3 of the left hand or a region of the right hemisphere containing areas 3a/3b. The FUS-evoked fMRI BOLD activations were similar to the BOLD activations elicited by natural tactile stimuli, suggesting that FUS may be able to excitatoryally excited or inhibit neural activity in functionally related touch regions.

When monkeys performed a simple target discrimination task, LC neurons responded to target cues with short latency and high-amplitude discriminative responses. These responses were longer if the discrimination was more difficult, as was the behavioral response. These LC discriminative responses also appeared to precede behavioral RTs.

In contrast, when monkeys performed a difficult task with target and nontarget trials, LC neurons were not significantly activated by nontarget stimuli. This was not surprising, because these stimuli were presented four times more frequently than target stimuli. Nevertheless, a small modulation in LC activity was evident when all nontarget trials were included in the analysis. This modulation was not significant by the 2 SD criterion, suggesting that it was not a significant response.

Similarly, when monkeys performed a simple rejection task, nontarget stimuli that resulted in correct rejection responses produced phasic LC activity. This phasic LC activity was small, however, compared with a large phasic LC response elicited when target stimuli were presented. This phasic activity occurred only when the stimulus was presented quickly and in a short duration.

Moreover, population analyses of LC activity revealed that both target and nontarget stimuli resulting in false alarm responses produced a small, early phasic response. The early phasic response was not related to the LC discriminative response, as was previously reported. Instead, this phasic response occurred in the same time period as nontarget stimulus trials and was similar to the small, early response elicited by all trial stimuli. This suggests that the early response reflects sensory information reaching the LC.


We present evidence that SII neurons can differentiate between the self and non-self in a mirror, corroborating previous reports on the multimodal nature of this region [21,43,44]. This suggests that mirror recognition is a fundamental but rudimentary precursor to self-recognition and is one of the first areas in which whole body somatosensation is integrated.

To evaluate the ability of SII cells to distinguish between the self and non-self in ambiguous mirror images, we measured single cell activation in an awake Japanese macaque while she was watching her own image in a mirror. Neurons were activated when the experimenter touched their own cheek, or when they touched an object that the monkey could see in a mirror. The number of cells that responded to these stimuli was significantly higher in the touch condition than in the mirror condition (Figure 1).

As in the PL array, face selective cells in the ML were also sensitive to the spatial location where a face should be – when presented with complex scenes that did not contain faces but contained cues that indicated where a face should be. Interestingly, face-selective neurons in the ML were less responsive to non-face parts of these images than to face parts in isolation, but these responses were not significant compared to PL sites.

We also examined the timing of responses to different epochs of a video depicting a monkey grasping objects from first person perspective. The neurons showed a tendency to increase their discharge in the first epoch and to decrease it in the second epoch, when this epoch was obscured.

This is similar to the behavior of neurons in M1 that respond to a visuomotor task or to watching a mouse being operated by an experimenter. Moreover, we show that these neurons have stable preferred direction tuning in both the execution and observation phases of these tasks.

As can be seen in Figure 4, we classified HS neurons on the basis of which stimulus produced the statistically highest response (3 x 6 ANOVA followed by Newman-Keuls post hoc test, p 0.01). The most frequent coded stimuli were those representing a monkey grasping objects from first or third perspective, mimicking this motor act or moving his arm, as well as stimuli that represented a human subject grasping objects and executing a movement.


When animals shift their attention in a visual detection task, they improve performance by enhancing the processing of the information that is most relevant to the behavioral task. This covert attentional modulation is accompanied by increased spiking rates of neurons in the receptive field (RF) of the neurons affected. Interestingly, this effect is not only found in the visual cortex but also in the lateral prefrontal cortex6. Previously, it was believed that attentional modulation was task-dependent; in other words, it only affected neuronal areas specialized to process the cued feature of a stimulus.

We therefore examined whether attentional modulation affects responses in other cortical structures and whether it is dependent on changes in behavior. In order to do this, we compared the activity of single neurons and multiunit clusters in V4 during two separate trials with high (large) or low (small) rewards. Then, we assessed whether these neurons were modulated by changes in attentional intensity, as measured by the z-scored population peri-stimulus time histograms (PSTHs) of spike rates within 60-260 ms from sample stimulus onset.

During large-reward trial blocks, a higher attentional intensity (neuronal d’) was associated with significantly increased z-scored population peri-stimulus spike rates in both single units and small multiunit clusters. This is consistent with a previous report that spike rates in V4 neurons are correlated with reward size, as well as with motor action17. Moreover, these results are consistent with our finding that V4 neurons exhibit a strong modulation by changes in attentional intensity without any behavioral selectivity or response criterion change6,12.

This is also in contrast to a prior study where the same findings were reported only for single-unit data (Mirabella et al., 2007 ; Fig. 3c). These findings suggest that the V4 neurons may be able to signal absolute reward size and motor action independently of attentional performance.

In addition to the peri-stimulus PSTHs, we quantified the neuronal modulation index AI for each of the 970 units in V4. It was defined as the difference between the z-scored population peri-stimulus rates during a high and low attentional intensity condition. This index was significant for both single units and small multiunit clusters in V4 (monkey P, mean = 0.13, p = 10-31; monkey S, mean = 0.09, p = 10-127).


In a decision task, neuron activation can be collapsed into a few dimensions that carry information about the choice. A recent study found that neuron activity reflects whether a monkey has made an immediate decision, hesitated or wavered, or stuck to a previous decision. This is an important mechanism for detecting a monkey’s intentions, since it allows researchers to record neural signals during decisions.

Using arrays of electrodes, we recorded the activity of two brain regions during monkeys’ target selection tasks and were able to determine which target they favored. This is the first time that this type of information has been recorded from the monkey’s brain.

The signals were also correlated with the timing and direction of the monkey’s saccades to the targets, suggesting that these areas were involved in decision making. These signals reliably predicted which of the two targets the monkey would prefer, several hundred milliseconds before they were instructed to start moving their fingers.

We used these signals to identify the neuronal circuit that processes choice-related values. We then pharmacologically inactivated these neurons and found that the latency to make a decision increased when value differences were small (Figure 2C), indicating that these neurons code for value information that is necessary to decide which object to choose.

In addition, we observed that the level of neuronal activation representing the target and distractor locations is highly discriminable when a monkey chooses accurately. This suggests that many neurons within the superior colliculus are involved in decision-making.

This is important because many of these neurons can be located in the medial frontal cortex (MFC). We found that MFC activation reflected different processes of comparison for unfamiliar and familiar options.

These results are consistent with a model that combines a recurrent circuit with an inhibitory current (T.D. Hanks et al., 2007, Soc. Neurosci.).

The inhibitory current balances competition and cooperation between two pools of selective neurons. When the inhibitory current is sufficiently strong, one of these pools wins and ends up with a high firing rate, whereas the other pool ends up with a low firing rate, indicating that a decision state has been reached. The decision-making process then involves a gradual transition from value to choice, as dopamine and OFC neurons encode the chosen value and gradually change their activity to reflect decision behavior.

- Advertisement -

More articles

- Advertisement -

Latest article