Even though there is a growing consensus that mirror neurons are not the basis of action understanding (or language, theory of mind, autism, etc.), the question remains, What are mirror neurons doing? There are now a couple of good theories. One is the idea promoted by Cecelia Heyes that observation of self-action results in an association between executed and observed actions, which is then generalized to others' actions. Another is the view proposed by Michael Arbib that mirror neurons start out as part of a motor control circuit, providing visual feedback on ongoing actions and then later get recruited to bolster action understanding.
Here's another idea: mirror neurons are not naturally occurring in wild macaques. Instead, they are trained into the lab animals by pairing experimenter object-related actions with the monkey’s own object directed actions.
Consider this quote about the training procedure from a 2001 report (Umlita et al. 2001):
Before starting the neurophysiological experiments, the monkeys were habituated to the experimenters. They were seated in a primate chair and trained to receive food from the experimenters. Monkeys received pieces of food of different size, located in different spatial locations. This pretraining was important for subsequent testing of the neuron’s motor properties … and for teaching the animal to pay attention to the experimenters.
It appears that the Parma group’s training procedures involved exposing the monkeys to a good deal of human actions that are directly related to the monkey’s task of reaching for and interacting with objects.
Why do I suspect that mirror neurons didn’t exist before this training? Because there is strong evidence for the development of mirror-like responses where previously they did not exist.
In the foundational Gallese et al. 1996 report, mirror neurons did not respond to grasping with pliers. This was viewed as evidence for mirror action specificity and biological relevance: monkeys don’t grasp with pliers so they can’t mirror that action. However, a later study showed that tool-related mirror neurons do exist, after the monkey is repeatedly exposed to humans picking up and poking items with tools. I quote from that report,
…observations made in our laboratory showed that at the end of a relatively long period of experiments, it was possible to find mirror neurons responding also to actions made by the experimenter with tools
Did this lead to a new understanding of tool use in the macaque observers? Not so much. After training, the monkey was presented with food that was just out of reach along with the stick that the experimenters had used to poke food during training (and which generated tool-related mirror neuron responses):
the monkey never attempted to use the tool for reaching food, although in the first minutes after the stick was available, the monkey grasped it and bit it.
I suggest (Hickok, 2014) that the action-related "canonical" mirror neurons discovered in the 1990s and that formed the foundation of new theories of action understanding, language, theory of mind, autism and the rest, formed in the same way, by associations built-up in the sensorimotor system via the training procedure. This hypothesis is easy to test.
Gallese, V., Fadiga, L., Fogassi, L., & Rizzolatti, G. (1996). Action recognition in the premotor cortex. Brain, 119 ( Pt 2), 593-609.
Hickok, G. (2014). The myth of mirror neurons: The real neuroscience of communication and cognition. New York: Norton.
Ferrari, P. F., Rozzi, S., & Fogassi, L. (2005). Mirror neurons responding to observation of actions made with tools in monkey ventral premotor cortex. Journal of Cognitive Neuroscience, 17(2), 212-226.
Umiltà, M., Kohler, E., Gallese, V., Fogassi, L., Fadiga, L., Keysers, C., & Rizzolatti, G. (2001). I know what you are doing. a neurophysiological study. Neuron, 31, 155-165.