Abstract:
About 30 years ago, neurons were found in the premotor area F5 of macaques that responded to both their own and observed grasping movements of the hand. These neurons were termed mirror neurons. Since then the key question has been what their function might be. Arguably the most influential attempt to answer this question has been the conjecture that mirror neurons help macaques to understand observed actions by simulating it internally by kickstarting the same mirror neurons that would be activated by a similar self-action. This idea could be understood as an adaption of the concept of motor imagery as a mental state that allows actors to prepare and initiate movements and to predict their effects. And indeed there are convincing arguments supporting the idea that motor imagery might be the basis of action understanding as posited by the action simulation theory of mirror neurons. Yet, a critical re-examination of the arguments put forward in support of the notion that mirror neurons are the basis of simulation sparks doubts and the need to consider alternative interpretations of their function.
The first chapter of this thesis provides a brief review of the discovery of the motor system and early descriptions of their properties against the backdrop of an introduction into the architecture of motor and premotor cortex, the latter involving area F5, in which mirror neurons were discovered. This introductory section is followed by a more detailed review of the studies that are relevant for the formulation of action simulation theory and its critical appraisal, a balancing presentation that also requires a consideration of work on mirror neurons, later found in other parts of the brain and several species other than macaque monkeys. This chapter ends with the conclusion that the assumption of congruence of observation and self-action related activity of mirror neurons, a major underpinning of the simulation theory needs to be re-examined and that an alternative functional role, namely a role in response selection should be considered. The second chapter, re-examines the viability of the assumption of congruence, whereas the third chapter addresses experiments designed to address the possibility that F5 mirror neurons might play a role in response selection.
In the experiments reported in chapter 2, we used rhesus monkeys that were trained to perform one out of three different grasping movements prompted by specific color cues, directed at objects whose shape and feel stayed constant. Next a human actor demonstrated the same three object-directed grasping movements to the experimental monkey. The question was to which extent the activity patterns of F5 mirror neurons, identified according to standard criteria, would resemble one another. Resorting to a sophisticated classification approach (Fisher’s linear discriminant analysis), we found very little evidence for a congruence of activity patterns. Only very few mirror neurons exhibited a preference for the same type of grasping movement and then also only for very restricted phases of the overall action sequences. This chapters concludes that the widespread absence of congruence is hardly in accordance with the tenets of the simulation theory.
In the third chapter, we used a novel action selection paradigm, in which the choice of the grasping action of the experimental monkey depended on the type of grasping act carried out by a demonstrator monkey, seen before in a video on a monitor in front of the experimental animal. In other words, the observer was asked to actively simulate an observed action. In a control experiment the experimental monkey was asked to choose the type of grasping action by using information provided by a color cue. Not unexpectedly, in view of the results reported in chapter 2, the observation and self-action related discharge profiles of mirror neurons tested in this overt action simulation task lacked similarity in the vast majority of cases. For instance, when observing demonstrated actions mirror neurons could exhibit a preference for action type A but for action type B in the subsequent self-action epoch, although the experimental monkey reproduced action type A. Notwithstanding the absence of congruence, a link between action observation and execution is documented by the fact that many mirror neurons exhibited ‘memory activity’, i.e. discharge in the period in which the monkey had to memorize the seen action in order to reproduce it following the delivery of a go-signal. The diversity of memory activity profiles may indicate different functions such as a memory trace of the observed action or the emergence of a motor plan. Interestingly, and at odds with the classical description of mirror neurons, many of the recorded mirror neurons responded to the simple color cues calling for a particular action with quite a few even preferring dot cues.
Taken together, the results reported in chapters 2 and 3 are hardly compatible with the action understanding theory and more in line with a neuronal system that allows agents to develop action plans considering visual and possibly in general sensory input. Sensitivity to the observation of simple color cues asking for particular self-actions does not necessarily rule out a natural preference for sensory information on others´ actions, considering that the sensitivity to color cues may be a reflection of learned associations. In any case, the absence of any simple relationship between action type preferences during observation and self-action clearly indicates that the assumed role of the mirror neuron system in using information on the other´s action in order to decide on one´s own action must be a network achievement.
Mirror neuron