The effect of observed biological and non biological movements on action imitation: An fMRI study

Abstract

Past research has indicated that when individuals observe biological movements many areas in the observer's motor system become active. Nonetheless, recent behavioral evidence showed that observed movements can interfere with execution of incompatible movements, especially the biological ones. However, the hypothesis that the interference originates within a common neural network, encoding both movement observation and execution, and responding preferentially to biological movements, still awaits confirmation. To test this hypothesis, in the present fMRI study we compared patterns of activation obtained when participants executed finger-movements after having observed either a biological or a non biological model performing compatible (imitative) or incompatible (non imitative) movements. Moreover, we tested the possibility that imitative responses are influenced by the emotional facial expression (sad, neutral, angry) presented before the observed movement. Behaviorally, participants showed a marginally larger compatibility effect (compatible movements faster than incompatible movements) in the biological condition than in the non biological condition. In the imaging data, the interaction testing for areas more active when the observed model was biological (compared with non biological) and performed compatible movements (compared with incompatible movements), activated a network including the motor, premotor and parietal cortices. Notably, the interaction was significant for the neutral and sad facial expressions only. We showed that observing biological movements modulates the activation of motor-related regions, by facilitating the execution of compatible movements and/or interfering with the execution of incompatible movements.

Introduction

Different research strands converge in supporting the argument that imitation is a contagious and automatic process (see Rumiati and Tessari, 2007, for a review). The most convincing evidence of the automaticity of imitation can be found in behavioral studies in which a very simple paradigm has been used (see Brass and Heyes, 2005, for a review). For instance, Brass et al. (2001) demonstrated that when participants are pre-instructed to perform a tapping movement in one block of trials, and a lifting movement in another block, independently of whether the movement they saw as prime was tapping or lifting movement, they were faster when seen and performed movements were the same (compatible trials) compared with when they were different (incompatible trials). This automatic imitation has been held to be an instance of the stimulus–response compatibility (SRC) (Brass et al., 2000) and has been interpreted in terms of a “direct matching” between the observation of a movement and its execution (see Iacoboni et al., 1999, Prinz, 1997). Direct matching theories need to be considered within the more general theoretical framework of ideomotor theory (Greenwald, 1970; see Brass and Heyes, 2005, for a recent review); the key concept of these types of theories is that observing the effect of an action facilitates its execution because perception and action planning share the same representations. On this account, imitation is achieved by activation of motor representations through action observation; the ease with which a stimulus is transformed into an action would depend on the similarity between the observed and the executed action (see Brass et al., 2000).

There is now increasing evidence that, in humans, action observation and imitation share a broad network of brain regions, including the inferior frontal gyrus, the ventral and dorsal premotor cortex, and the superior and inferior parietal lobes (see Brass and Heyes, 2005, for a review; see also Iacoboni et al., 1999). This network roughly overlaps with the regions in the monkey's brain, in which neurons fire when the animal sees as well as when it performs a purposeful action (i.e., the Mirror Neuron System, MNS, di Pellegrino et al., 1992, Rizzolatti and Sinigaglia, 2010, for a recent review).

Consistent with the view that imitation relies on shared representations for perceived and executed actions, it has been shown that the mere observation of biological actions (e.g., realistic actions performed by a human agent) is especially effective in activating the regions of a putative human MNS, possibly because this type of actions is already present in the imitator's motor repertoire (e.g., Perani et al., 2001, Stevens et al., 2000, Tai et al., 2004). Likewise, several behavioral studies have reported larger visuomotor priming effects (i.e., the generation of motor activation by visual stimuli) for biological stimuli (such as moving hands or limbs) compared with non biological stimuli (such as moving objects or robotic limbs), hence suggesting that human perception–action coupling may be tuned to biological actions (e.g., Bird et al., 2007, Brass et al., 2001, Castiello et al., 2002, Jonas et al., 2007, Liepelt and Brass, 2010, Press et al., 2005, Press et al., 2006, Press et al., 2007; but see Gazzola et al., 2007, Newman-Norlund et al., 2010, for a different account). Furthermore, behavioral studies have reported specific interference effects of observed movements on action execution (Kilner et al., 2003, Kilner et al., 2007). In particular, in Kilner et al. (2003), the interference was only found when observed movements were incompatible (i.e. non imitative) with the executed movements and the former were performed by a biological model, compared with a non biological model. The interference was held to be mediated by an activation of the parietal and premotor cortices, namely of a common neural network that would encode both observed and executed movements and respond preferentially to biological actions.

To the best of our knowledge, no past fMRI study has been designed to investigate the neural correlates of the interference effect of observed biological and non biological movements on executed actions. In the present fMRI study we used a paradigm similar to that employed by Brass et al. (2001, experiment 2). We compared patterns of activation observed when participants executed finger movements (e.g., tapping) after having observed either a hand or a moving dot (biological/non biological movement) performing compatible/imitative (i.e., tapping or downward movements for biological and non biological movements, respectively) or incompatible/non imitative movements (lifting or upward movements, respectively). Our main prediction concerned the interaction term testing for areas more active when participants saw biological model compared with non biological model and performed imitative movements vs. non imitative movements. Activation in the parietal and motor/premotor regions would suggest that observed biological movements affect action production, by facilitating the execution of imitative movements and/or interfering with the execution of non imitative movements.

We also aimed at examining whether the type of emotional facial expression presented before the observed movement influences imitative responses. In a previous study (Grecucci et al., in press; see also Grecucci et al., 2009), some of us examined how emotional pictures, presented as primes, affect imitative tendencies using a Brass et al.'s (2001) modified compatibility paradigm. The key result was that when seen index finger movements (tapping or lifting) and pre-instructed finger movements (tapping or lifting) were the same (tapping–tapping or lifting–lifting, compatible trials), subjects were faster than when they were different (lifting–tapping or tapping–lifting, incompatible trials). This compatibility effect was enhanced when the seen finger movement was preceded by negative primes compared with positive or neutral primes; we proposed that this could be due to negative stimuli being particularly effective in rapidly preparing the organism for actions in potential flight-or-fight situations (Grecucci et al., in press).

In the present study, we used instead facial expressions (neutral, sad and angry) because of their known role in inducing empathic and imitative reactions (e.g., Carr et al., 2003), and predicted that participants would show a stronger imitative tendency following sad but also neutral facial expressions than angry facial expressions. This is because while we are inclined to empathize with a person expressing sadness (Blair, 2003, Blair et al., 1999), we are not likely to do so with a person expressing anger: the former case elicits prosocial behavior and may involve mirroring the expression (Chakrabarti et al., 2006), while the latter can be perceived as a threat and thus trigger avoidance behavior in the observer (e.g., Pichon et al., 2008, Pichon et al., 2009; see also social response reversal, Blair, 2003, Blair et al., 1999).

As far as the brain activations are concerned, we expected the activity of parietal and motor/premotor regions during imitation of biological vs. non biological movements to be particularly influenced by an emotional context promoting empathy (i.e. with sad and neutral facial expressions). Premotor and motor areas were found to be robustly activated when participants observed emotional facial expressions, in particular in the imitative condition (Carr et al., 2003); however, based on their study, it is not possible to establish the contribution of the different expressions presented or their effects on imitative responses as they were not measured. Nevertheless, it was later showed that empathy may rely on a perception–action mechanism that shares with hand imitation a common neural network involving motor/premotor brain regions (Leslie et al., 2004).

Overall, the present experimental design involved the within-subjects variables of compatibility (compatible vs. incompatible), observed movement (biological vs. non biological), and the facial expression (neutral, sad, and angry) presented before the observed biological or non biological movement (see Fig. 1).