Poor matching of action codes challenges ‘mirroring’ in macaque F5 mirror neurons

The discharge of F5 mirror neurons of a monkey observing another individual performing an action has been interpreted as a motor representation, stimulated by sensory input, ultimately serving action understanding. This hypothesis requires mirror neurons to exhibit an action tuning that is matched between action observation and execution. However, the evidence for this is insufficient, mainly because previous work has suffered from a limited exploration of action space tuning and an object vs. action vision confound. To overcome these limitations, we conducted an experiment in which identical objects had to be manipulated in three different ways in order to serve distinct action goals. We show that the population of F5 mirror neurons can dissociate the three actions almost perfectly, during both observation and execution. However, the population code for action execution was only poorly matched to the code for observation. Just a few neurons exhibited only matched codes, yet, always confined to short, varying segments of the overall action. These findings challenge the hypothesis that an observer understands the actions of others by the activation of motor representations of the observed actions. Rather they support the alternative notion that F5 might use information on observed actions to select behavioral responses.