Researchers have recorded activity of neurons in the monkey’s premotor cortex. These neurons fire when they see other monkeys do certain actions. The neurons are called mirror neurons and they are important for learning from others.
The neurons were tested with 12 video stimuli, eight depicting monkey and human actions. For each trial, the monkey kept its hand in its resting position and maintained fixation on a randomized window of the video.
Researchers first identified mirror neurons in the premotor cortex of macaque monkeys in the 1980s, but they only recently found that these brain cells also fire when watching other people perform a similar action. The findings suggest that these neurons play a role in understanding other people’s actions and emotions. This knowledge could help humans better communicate with each other and cooperate.
While many studies have examined mirror neuron activation in the F5 and IPL, only three have provided details about the selectivity of these neurons. One study found that, on average, about 48.9% of these neurons were broadly congruent (that is, they discharged during both execution and observation). Other studies have shown that some mirror neurons discharge only for certain types of observed actions, such as grasping.
Several studies have also shown that the selectivity of mirror neurons depends on the action’s overall goal. For example, some neurons in IPL fire only when the monkey observes a person performing a grasping action that is aimed at an object. On the other hand, other neurons in IPL fire when a person performs a grasping action that is not aimed at an object.
Other studies have found that the selectiveness of mirror neurons in IPL and F5 is modulated by hunger. In a study that appeared in Cerebral Cortex, Decety and Meltzoff showed a group of subjects videos of someone grabbing food. They found that the mirror systems of the hungry subjects were more active than those of the non-hungry subjects. This suggests that more primitive motivations such as hunger regulate the selectiveness of mirror neurons.
Some researchers are interested in whether these neurons can respond to stimuli other than hands. For example, a recent study published in Science found that mirror neurons in the frontal lobe of humans fire when they imitate or observe facial expressions. However, it is not possible to record activity from individual human brain cells using electrodes.
In addition, it is not clear whether mirror neurons in the human brain are more selective than those in the monkey brain. It is also unclear whether they can encode the exact same actions, such as hand gestures. A more promising avenue of research is the use of functional imaging to examine how different parts of the brain are connected to each other, which will provide more clues about the function of these neurons.
Stimulus-specific neurons are brain cells that respond differently to the same type of stimulus. These neurons are characterized by a high trial-by-trial correlation in their response counts, suggesting that they are selective for certain features of a stimulus. For example, one neuron may be more active when a monkey is grasping an object from first person perspective than when the object is presented in a different position or angle. The resulting neural responses could help explain how the brain makes sense of action, explains what triggers specific behaviors and how the nervous system adapts to changing situations.
Researchers have previously shown that VLPF single neurons can encode biological stimuli, such as static faces or expressive facial movements associated with vocalizations. More recently, they have demonstrated that this coding extends to the forelimb movement. Using control tasks that did not require the monkey to perform any forelimb movement, they discovered that some of these neurons were also able to code visual features of the stimuli.
To determine the preference of a particular stimulus, they recorded the activity of neurons in VLPF during passive observation of videos of human or monkey actions. These videos were divided into six epochs, with each video showing a different combination of actions and agents. In some epochs, the actions were obscured by shading. In other epochs, the action was performed by a robot that did not move. The HS neurons exhibited a preference for both videos and the HM neurons tended to prefer MGI and MGIII.
Among these, the most effective videos for activating the HS neurons were those that depicted a monkey in front of an object and reaching towards it. The HS neurons showed greater response to the first video epoch in these conditions, and their responses diminished when the second video epoch was obscured. Neurons that preferred Epoch 2 or responded equally well to both epochs are thought to be coding the context of the action and possibly its beginning and end.
The experiment was conducted on two Rhesus monkeys (M1 and M2) weighing about 4 kg. The animals were highly trained on the task prior to the start of the recording sessions, and their performance levels were above 90% correct responses. The animal handling and surgical procedures were in accordance with European guidelines for the care and use of laboratory animals.
Neurons with a preference for a particular type of stimulus
Neurons with a preference for a particular type of stimulus can be identified using a variety of methods. For example, one method involves analyzing the response to multiple stimuli and comparing their average responses. Another method uses a genetic algorithm to search for stimuli that maximize the likelihood of a response. The resulting dataset allows researchers to identify neurons with a preference for specific types of stimulation and analyze their function.
The researchers used this data to categorize the neurons into groups based on their preference for different types of stimuli. They found that the majority of neurons responded to stimuli with a human agent performing an action, while only a small number responded to monkey or object motion. These neurons are referred to as “human-oriented” (HS) neurons. The authors also analyzed the normalized neural discharge of these neurons and found that their responses to the different stimuli were highly correlated with each other, suggesting that they are highly selective for a particular stimulus.
To further investigate the preferences of these neurons, they tested their responses in a control task. In this control task, the monkey passively observed a video that featured either human or monkey actions. The researchers then obscured the first or second epoch of the video, which prevented the monkey from seeing the action itself. They found that many of these neurons showed a preference for the agent who was performing the action. These findings suggest that these VLPF neurons are not simply responding to visual descriptions of actions, but that they are actually coding the context and the beginning of the movement.
Neurons were categorized into four clusters based on their preference for different types of stimulus. Each cluster was characterized by its depth of tuning and selectivity indexes. The depth of tuning index, di, represents the proportion of the population of neurons that responds to a given stimulus. The selectivity index, si, indicates the percentage of the population that is able to discriminate a given stimulus from other stimuli. The polar plot in Figure 2 shows the correlation between these two indexes. A positive value indicates a stronger selectivity for a particular stimulus.
Neurons with a preference for a particular phase of the video
Neurons with a preference for a particular phase of the video are called “High Selectivity” (HS) neurons. In a previous study, we recorded these neurons in the prefrontal cortex of two monkeys while they passively observed videos showing seven manipulative actions performed by humans in first and third person perspectives. The data revealed that HS neurons encode a variety of observable features, including the type and location of the hand-object interaction, and that these neurons are invariant to changes in visual presentation formats.
The coding performance of HS neurons was measured by comparing psychometric and neurometric curves, which represent the average response slope and peak amplitude of the stimulus over time. The results showed that a high proportion of HS neurons elicited responses in phase with the stimulus sinusoidal wave, allowing them to detect the signal reliably. However, the mean psychometric curve was significantly steeper than the neurometric one, suggesting that additional coding codes might be involved.
To further evaluate HS neurons, we measured the epoch preference of each neuron. The epoch preference refers to the period of time that the HS neuron responds stronger to a particular phase of the video. For example, a neuron that discharges more strongly during the observation of a monkey grasping an object from a first-person perspective will decrease its response to that video epoch when it is obscured and begin to respond to the second epoch.
We also computed a depth of tuning and selectivity index for each HS neuron. The depth of tuning index represents the difference between the neuron’s maximum and minimal response normalized to its baseline. The selectivity index is the extent to which a neuron’s response deviates from maximal activity in non-preferred stimuli.
For example, a neuron that discharges strongest during the observation of a monkey grasping a food object from a first-person perspective will have a depth of tuning and selectivity index of 1. The same is true for neurons with a preference for a particular phase in the video. However, the number of neurons that display a particular preference is much smaller than the total number of neurons that respond to the various videos.