Tool use linked to mirror neurons?

1 minute read

Michael Balter reports on recent research by Giacomo Rizzolatti, of "mirror neuron" fame. They taught some macaques to use pliers, and recorded what their neurons were doing:

Rizzolatti and his co-workers conclude that when learning to use a tool, the pattern of neuronal activity is somehow transferred from the hand to the tool, "as if the tool were the hand of the monkey and its tips were the monkey's fingers." As for how the same neurons could affect both the opening and the closing of the hand, the team speculates that they may be connected with other sets of neurons that more directly control these movements. The authors also point out that area F5 is rich in so-called mirror neurons, a type of nerve cell discovered earlier by Rizzolatti that fires both when a primate performs an action and when it observes another individual doing the same thing (ScienceNOW, 13 July 2007). Mirror neurons in F5, the authors suggest, may be involved in this transfer process as a monkey learns how to use a tool by watching others.

This is pretty cool, but I think its surprisingness has been overstated. A bit of reflection (har!) shows that something like mirror neurons ought to exist -- it would be far harder to build a brain capable of learning by observation if some of the same neurons weren't involved in both observing and doing. Likewise, it would be a lot more complicated to learn to use a tool if the same neurons weren't involved in planning the action using the hands compared to planning the action using a tool. So there ought to be a whole lot of neural overlap.

Now, the fact that these are in a particular region (F5) is structurally important, since this region has other functions whose interrelations aren't obvious logical necessities. And the fact that tool use is another example of a behavior that maps onto this region (at least in terms of learning) is also pretty important. It will be nice to see whether this "transfer" ability -- from hand to tool -- is a primate feature, or if instead it may be more widely distributed.