Experience modulates automatic imitation

Abstract

Action observation gives rise to activation in corresponding areas of the premotor and primary motor cortices. We tested the hypothesis that this activation depends on visual–motor connections established through correlated experience of observing and executing the same action. Previous work has shown that hand opening and hand closing gestures are facilitated when subjects observe the movement they are performing, relative to a condition in which they observe a different movement from the one they are performing. Experiment 1 replicated this finding in a simple reaction time (RT) procedure using stimulus–response (SR) movements in orthogonal planes. This implies that the effect is an example of automatic imitation, an instruction-independent tendency to execute movements that are topologically similar to those observed, and not merely an example of spatially compatible responding. In Experiment 2, the automatic imitation effect found in Experiment 1 was abolished by a brief period of training in which subjects responded to hand opening by closing their hands, and to hand closing by opening their hands. This outcome is consistent with the hypothesis that, rather than being innate, the cortical connections mediating motor activation by action observation are formed through experience.

Introduction

There is now a substantial body of evidence that action observation gives rise to matching motor activation, i.e., to activity in the premotor and primary motor cortices normally associated with execution of the observed action [2], [3], [5], [9], [10], [15], [16], [17], [21], [25]. Functional magnetic resonance imaging (fMRI) has indicated somatotopically organized activation of the premotor cortex during observation of hand, foot, and mouth movements [5]. For example, activation of the area of the premotor cortex responsible for execution of mouth movements is greater during observation of mouth movements than during observation of hand and foot movements. Evidence of primary motor cortex activation by action observation is provided by the finding that suppression of the ∼20 Hz motor cortex rebound occurs during both action execution and action observation [10], [16], [17].

Further evidence that action observation activates corresponding motor representations has come from behavioral studies demonstrating movement compatibility or ‘automatic imitation’ effects [4], [6], [27]. For example, in a simple reaction time (RT) task, Brass et al. [4] instructed subjects to make one of two finger movements, lifting or tapping, in response to video stimuli showing either a lifting or a tapping finger. They found that subjects' responses were much faster when the stimulus and response movements were the same (compatible trials) than when they were different (incompatible trials). Similarly, in a choice RT task, Stürmer et al. [27] required subjects to open or close their hand when an image of an opening or closing hand changed color. Although the type of stimulus movement (opening or closing) was irrelevant to the assigned task, subjects responded faster on compatible trials, when the stimulus and response movement were the same, than on incompatible trials. The effects reported in both of these studies suggest that action observation promotes or facilitates imitation, execution of the observed action, and that this facilitation is automatic in the sense that it does not depend on task instructions. Given their automatic character, it is likely that these imitation effects are behavioral products of premotor and primary motor cortical activation by action observation.

Although the evidence that action observation causes matching motor activation is now strong, there has been very little research examining the source of this tendency [13]. Motor activation by action observation implies that areas of the cortex, mediating visual processing of an observed action, are linked, directly or indirectly, to motor areas involved in execution of the same action. However, we do not yet know how these connections are formed. What is the process which has ensured that action observation causes activation of motor areas mediating execution of the same actions, rather than of different actions? Broadly speaking, the candidate processes are natural selection and learning; the visual–motor cortical connections may be innate or formed through experience.

Reports of imitation in newborn infants [22] may indicate that at least some of the cortical connections mediating motor activation by action observation are innate. However, these data are controversial [1], [18]. Furthermore, it has been argued that neonatal imitation is intentional, rather than automatic, and, therefore, it is unlikely to be mediated by the ‘mirror-neuron system’ [23]. In contrast, the Associative Sequence Learning (ASL) hypothesis emphasizes the role of learning in imitation. It suggests that the cortical connections mediating motor activation by action observation arise primarily through correlated experience of observing and executing the same actions [12]. This view has not been tested directly, but it is consistent with two findings. First, mirror neurons in monkey F5 do not initially respond during observation of a mechanical pincer grasping an object, but they begin to do so when the monkey has repeatedly observed the pincer action and, between observation trials, grasped the object with its own hand [24]. Second, muscle-specific facilitation of TMS-induced MEPs by finger movement observation is greater when the model's hand is presented in the orientation at which one normally views one's own hand (heading away) than when it is presented in the orientation at which one normally views another person's hand (heading toward) [20]. This finding is consistent with the Hebbian hypothesis that visual–motor cortical connections are formed through correlated experience of observation and execution because, in most environments, execution of finger movements is more likely to be accompanied by observation of one's own finger movements than by observation of another person's finger movements.

The purpose of the present study was to investigate the role of experience in producing motor activation by action observation. Specifically, we asked whether an automatic imitation effect can be reduced or abolished by prior training in which subjects have correlated experience of observing and executing different actions. This question was addressed directly in Experiment 2, using a test procedure in which subjects were required to perform a pre-specified hand movement (opening or closing) in response to hand opening and hand closing video stimuli. Experiment 1 cleared the way for Experiment 2 by checking whether the tendency to perform the same hand opening/closing movement as a model is a genuine example of automatic imitation.