Perceiving fear in dynamic body expressions

Abstract

Characteristic fear behaviour like putting the hands in front of the face and running for cover provides strong fear signals to observers who may not themselves be aware of any danger. Using event-related functional magnetic resonance imaging (fMRI) in humans, we investigated how such dynamic fear signals from the whole body are perceived. A factorial design allowed us to investigate brain activity induced by viewing bodies, bodily expressions of fear and the role of dynamic information in viewing them. Our critical findings are threefold. We find that viewing neutral and fearful body expressions enhances amygdala activity; moreover actions expressing fear activate the temporal pole and lateral orbital cortex more than neutral actions; and finally differences in activations between static and dynamic bodily expressions were larger for actions expressing fear in the STS and premotor cortex compared to neutral actions.

Introduction

The sound of gunfire immediately causes one to bend forward and to run for cover. Such characteristic fear behaviour protects from danger and also communicates a strong fear signal to observers who may not have heard the noise themselves. Emotional body language (EBL) provides reliable cues to recognise another’s emotions even when viewed from a distance and when the facial expression is not visible. The few currently available studies show that EBL can readily be recognised whether in static postures (de Gelder et al., 2004, Sprengelmeyer et al., 1999), whole body movements (Atkinson et al., 2004, Wallbott and Scherer, 1986) or even simple dynamic point-light displays (Dittrich et al., 1996, Bonda et al., 1996). Because of its survival value, the ability to grasp EBL is likely based on processes that are rapid, highly automatic and possibly relies systems.

At present, very little is known about the neural basis of perceiving EBL. As far as visual communication of emotion is concerned, our most valuable insights are based on investigations using images of static facial expressions. Amygdala and mid-fusiform cortex have consistently been associated with viewing facial expressions of fear (Adolphs, 2002, Dolan, 2002, Haxby et al., 2002). These two brain structures appear also to be important for processing fearful EBL (Hadjikhani and de Gelder, 2003). Yet besides the fact that facial expressions and EBL may share important brain structures, seeing EBL also evokes body specific activations (de Gelder et al., 2004). This is not surprising since EBL represents not only salient visual information but requires that the observer perceives the movements represented in the dynamic images or implied in the still pictures (running away) and grasps the meaning of the action (fleeing from danger).

Some of these characteristic aspects of EBL have recently been addressed in relatively separate research domains. For example, a possible role of premotor areas in emotion recognition has been revealed in brain imaging studies using dynamic facial expressions. Viewing dynamic facial expressions as compared to static ones engages areas processing biological movements and emotion such as the superior temporal sulcus (STS) and the amygdala (AMG), but also areas involved in the production of facial expressions, in particular the parietal and premotor cortex (Decety and Chaminade, 2003, Sato et al., 2004). These results suggest that dynamic images elicit more activity in the areas representing the affective meaning of the stimulus like the amygdala because they provide more information than static ones. On the other hand, dynamic stimuli also contain explicit movement information and this presumably elicits activity in movement sensitive areas like superior temporal sulcus (STS) and in premotor areas, as is indeed the case for dynamical facial expressions (LaBar et al., 2003, Sato et al., 2004). Given the limited information presently available it is unclear whether the different patterns of activation for dynamic and still images reflect a quantitative or also a qualitative difference. In other words, it is not yet clear whether the presence of dynamic information makes a substantial difference. It may be difficult to sort this issue out in the case of faces also because movement may be perceived implicitly even when it is not represented explicitly in the stimuli.

As the examples above bring out, we often perceive EBL when watching a person engaged in one or another action, like running away but also opening the door and making angry or fearful gestures at the unexpected visitor. On some occasions the EBL is directly linked with the emotion as in the case of fear and associated flight reaction as in our previous study (de Gelder et al., 2004). In the case envisaged EBL consists of an instrumental action performed with a strong emotional overtone as when we see a hand grasping a glass angrily (Grosbras and Paus, 2006). Thus clarifying how the brain processes EBL raises issues related to action perception which one does not encounter when investigating emotion expressed in the face.

It is now well established that when one observe other people’s actions, there is activation in the STS, the parietal and the premotor cortex (see review Grèzes and Decety, 2001). This suggests that action observation automatically triggers action representations. This mechanism of shared representations was proposed as the basis of action recognition (Jeannerod, 2001, Rizzolatti et al., 2001, Gallese et al., 2004, Iacoboni, 2005) but also more recently emotion recognition (Preston and deWaal, 2002, Carr et al., 2003, Gallese et al., 2004). Yet the relation between perceiving the emotion and representing the action has not received much attention so far. Furthermore, research on emotion has predominantly used still faces while investigations of action observation most often used video clips. The few available human neuropsychological findings indicate that the dorsal system may sustain some degree of visual processing of dynamic expression of emotion. For example, patients with focal amygdala lesion are impaired in recognizing static but not dynamic facial expressions (Adolphs et al., 1994, Adolphs et al., 2003), and they are able to produce an imitation of a facial expression on demand (Anderson et al., 2000).

The present study focussed on neutral and emotionally expressive instrumental actions and aimed at clarifying the specific contribution of dynamic information to the perception of fearful body expression. Furthermore, our approach allows us to clarify the relation between processing the fear signal provided by whole body actions in emotional and motor areas. Using event-related fMRI and a factorial design, we compared neutral and fear expressing whole body actions and their scrambled counterparts presented in either still or dynamic images. Our goals were to identify the neural circuits that are specifically involved in the perception of an action involving the whole body condition specific effects of fear and the combined effect of fear and dynamic action information.