Mirror Neurons in a New World Monkey, Common Marmoset

Abstract

Mirror neurons respond when executing a motor act and when observing others' similar act. So far, mirror neurons have been found only in macaques, humans, and songbirds. To investigate the degree of phylogenetic specialization of mirror neurons during the course of their evolution, we determined whether mirror neurons with similar properties to macaques occur in a New World monkey, the common marmoset (Callithrix jacchus). The ventral premotor cortex (PMv), where mirror neurons have been reported in macaques, is difficult to identify in marmosets, since no sulcal landmarks exist in the frontal cortex. We addressed this problem using "in vivo" connection imaging methods. That is, we first identified cells responsive to others' grasping action in a clear landmark, the superior temporal sulcus (STS), under anesthesia, and injected fluorescent tracers into the region. By fluorescence stereomicroscopy, we identified clusters of labeled cells in the ventrolateral frontal cortex, which were confirmed to be within the ventrolateral frontal cortex including PMv after sacrifice. We next implanted electrodes into the ventrolateral frontal cortex and STS and recorded single/multi-units under an awake condition. As a result, we found neurons in the ventrolateral frontal cortex with characteristic "mirror" properties quite similar to those in macaques. This finding suggests that mirror neurons occur in a common ancestor of New and Old World monkeys and its common properties are preserved during the course of primate evolution.

Keywords: New World monkey; in vivo imaging; premotor cortex; primate; superior temporal sulcus.

Figure 1

In vivo surface connection imaging and flat map of CTB-Alexa555 staining. (A) Cortical surface of the lateral frontal cortex and temporal cortex of a common marmoset after craniotomy and duratomy. Four green dots in the ventrolateral frontal cortex and STS indicate the implantation position of four shanks of linear array multicontact (32-channels) electrodes. (B) In vivo surface connection imaging of the lateral frontal cortex and temporal cortex shown in (A). CTB-Alexa555 was injected into the posterior part of STS, as determined by electrophysiological mapping. The image was adjusted for brightness and contrast for presentation purposes. (C) Top: Two-dimensional “unfolded” labeled cell density map of the cortical surface, constructed by computer graphic reconstructions of the cortex prepared with the software program CARET (Van Essen et al., 2001). Pseudocolor represents the density of labeled cells. The yellow dotted line indicates STS. Bottom: Coronal sections stained by Nissl substrate and myelin showing the areal boarder in the ventrolateral frontal cortex and distribution of the labeled cells. The asterisks indicate recording tracks of shanks of electrodes in the ventrolateral frontal cortex. D, dorsal; R, rostral.

Figure 2

Electrophysiological mapping of STS under anesthetic condition. (A) Cortical surface around STS of a common marmoset. Four green dots indicate the penetration sites of four shanks of a linear array multicontact electrode. The red line indicates the presumed cortical surface because only 4–5 bottom contacts of each shank were inserted. The yellow dotted line indicates STS. (B) Multiunit responses to the sight of grasping action of another marmoset, which were arranged by channel configuration. Rasters and peristimulus time histograms were aligned to the stimulus onset at time = 0. The bin width for the peristimulus time histogram was 20 ms. To minimize damage to the cortex caused by the electrode penetration, multiunits were recorded only from channels on the upper part of each shank. The multiunits on the bottom two channels on the second shank strongly responded to the movie. (C) Multiunit responses to the sight of human grasping action, which were arranged by channel configuration. As in (B), the multiunits on the bottom two channels on the second shank strongly responded to the movie.

Figure 3

Three examples of multiunit responses (A–C) with mirror neuron properties under observation and execution conditions in the ventrolateral frontal cortex. Rasters and peristimulus time histograms were aligned to the touch of food at time = 0. The bin width for the peristimulus time histogram was 100 ms. The left (right) column indicated the multiunit responses when an experimenter grasped food from the left (right) side of the animal. The first to fourth rows indicate the multiunit responses when an experimenter grasped (1) a piece of banana with his/her hand, (2) a piece of bun with his/her hand, (3) a piece of banana with a pair of forceps, and (4) he/she mimed to reach and grasp as if there was a piece of food, respectively. The bottom indicates the multiunit responses when the animal itself grasped a piece of banana or bun. The responses in the shaded area were used for statistical analysis. ^*p < 0.05/8 under observation condition, p < 0.05 under execution condition (paired-t test).

Figure 4

Time course of normalized activity of neuronal population for 27 mirror neurons in the ventrolateral frontal cortex. The responses were aligned at the moment when the experimenter or animal touched the food and were averaged after normalizing the responses by the peak responses. Black and gray lines indicate the responses when an experimenter grasped a piece of food under the most preferred grasping action condition and when the animal itself grasped a piece of food, respectively.

Figure 5

Two examples of single-unit responses under observation and execution conditions in the ventrolateral frontal cortex. Display formats were the same as those in Figure 3. (A) Single-unit responses that satisfied the mirror neuron criteria. (B) Magnitude of single-unit responses that significantly increased under the observation condition, but decreased under the execution condition. ^*p < 0.05∕8 under observation condition, p < 0.05 under execution condition (paired-t test).

Figure 6

An example of a mirror neuron in STS and time course of the population activity for seven mirror neurons. (A) Rasters and peristimulus time histograms of an example of STS mirror neuron responses. Display formats are the same as in Figure 3. (B) Time course of the normalized activity of neuronal population for seven mirror neurons in STS. Display formats are the same as in Figure 4. ^*p < 0.05∕8 under observation condition, p < 0.05 under execution condition (paired-t test).

Figure 7

Comparison of visual-responsive cells in STS, visual-dominant cells in the ventrolateral frontal cortex, and mirror neurons in the ventrolateral frontal cortex. (A) Time course of normalized activity of neuronal population for 30 visual-responsive multiunits in STS, 35 visual-dominant multiunits in the ventrolateral frontal cortex, and 27 mirror neurons in the ventrolateral frontal cortex. Normalized responses of the multiunits for the preferred reaching direction when an experimenter grasped a piece of banana with his/her hand (blue), a piece of bun with his/her hand (red), a piece of banana with a pair of forceps (green), and he/she mimed to reach and grasp as if there was a piece of food (purple). The responses were aligned at the moment when the experimenter or animal touched the food. (B) Response magnitudes of visual-responsive cells in STS (green), visual-dominant cells in the ventrolateral frontal cortex (red), and mirror neurons in the ventrolateral frontal cortex (blue) for each grasping action type. Error bars indicate standard error of the mean.