Visual features of an observed agent do not modulate human brain activity during action observation

Abstract

Recent neuroimaging evidence in macaques has shown that the neural system underlying the observation of hand actions performed by others (i.e., “action observation system”) is modulated by whether the observed action is performed by a person in full view or an isolated hand (i.e., type of view manipulation). Although a human homologue of such circuit has been identified, whether in humans the neural processes involved in this capacity are modulated by the type of view remains unknown. Here we used functional magnetic resonance imaging (fMRI) to investigate whether the “action observation system”, with specific reference to the ventral premotor cortex, responds differentially depending on type of view. We also tested this manipulation within regions of the human brain showing overlapping activity for both the observation and the execution of action (“mirror” regions). To this end, the same subjects were requested to observe grasping actions performed under the two types of view (observation conditions) or to perform a grasping action (execution condition). Results from whole-brain analyses indicate that overlapping activity for action observation and execution was evident in a broad network of areas including parietal, premotor and temporal cortices. Activity within such network was evident for both the observation of a person in full view or an isolated hand, but it was not modulated by the type of view. Similarly, results from region of interest (ROI) analyses, performed within the ventral premotor cortex, did confirm that this area responded in a similar fashion following the observation of either an isolated hand or an entire model acting. These findings offer novel insights on what the “action observation” and the “mirror” systems visually code and how the processing underlying such coding may vary across species. Further, they support the hypothesis that action goal is amongst the main determinants for the revelation of action observation activity, and to the existence of a broad system involved in the simulation of action.

Introduction

Mirror neurons are a class of visuomotor neurons activated by both the execution and the passive observation of object-related actions. Cells having this property were found in macaques within the convexity behind the arcuate sulcus (area F5c) within the premotor cortex (Di Pellegrino et al., 1992, Gallese et al., 1996, Rizzolatti et al., 1996a), and in the complex PF/PFG (PF) within the rostral part of the convexity of the inferior parietal cortex (Gallese et al., 2002, Fogassi et al., 2005).

Following this discovery, many functional resonance imaging (fMRI) studies have been performed in order to uncover a similar system in humans (for review see Dinstein et al., 2008, Turella et al., 2009). Amongst these studies the most convincing evidence of a “mirror-like” system in humans comes from a study by Gazzola et al., 2007a, Gazzola et al., 2007b and Gazzola and Keysers (2008) who tested both action execution and observation within the same individuals. In one day they asked participants to observe either a human model or an industrial robot performing a variety of actions and in a separate day to perform the actions. They found regions of overlap for action observation and execution in classic ‘mirror’ areas together with many areas which were not previously considered as mirror (Gazzola et al., 2007a, Gazzola et al., 2007b, Gazzola and Keysers, 2008). Further, another interesting finding stemming from this work is that the mirror system is similarly activated by the sight of both the human and the robotic hand. This occurred despite the movement of the two agents exhibiting dramatically different kinematics. This was taken as the evidence that action goal rather than a tight kinematic match is amongst the main determinants for the revelation of mirror activity.

In this connection, recent findings from a fMRI study investigating the neural underpinnings of action observation in monkeys add a further level of complexity regarding the visual requirements necessary to activate the ‘observation’ component of the mirror system (Nelissen et al., 2005). Monkeys observed video clips showing a full view of a person grasping an object or an isolated hand grasping objects and static single frames or scrambled videos as controls. It was found that premotor area F5c (the area in which mirror neurons were first discovered) was active only when the monkey observed a human model presented in her entirety grasping an object, but it was not active when it observed a human hand detached from the body performing the task. In the other subregions of F5 (i.e., F5a, F5p) activation due to action observation was reported for both the model and the hand alone acting. These results seem to suggest that the type of view alerts different sectors of the premotor cortex and this occurs despite the fact that the goal for the two different agents, both biological in nature and presumably showing similar kinematics, remains the same. Therefore it might well be that visual features of agents are as important as action goal for modulating action observation activity, at least within the core “mirror” area F5c. We do not know whether manipulation concerned with the type of model would have produced similar results in other areas, with or without “mirror” properties, given that in this study the investigation was restricted to the ventral premotor cortex and nearby prefrontal regions by means of regions of interest (ROI) analyses (Nelissen et al., 2005).

Here we capitalize on the above mentioned findings to investigate for the first time whether in humans, as happens in monkeys, the action observation system or part of it is differentially activated depending on the type of view irrespective of action goal. This is a reasonable question to ask considering that a number of fMRI studies in humans have shown that action observation in humans evokes widespread frontal activation, including that of premotor area 6 and of prefrontal areas 44 and 45 which may modulate, as reported in monkeys (Nelissen et al., 2005) depending on the type of view (for review see Turella et al., 2009). Further, because ‘mirror’ activity in humans is detected across a number of areas which exceed those classically considered as ‘mirror’ (Gazzola et al., 2007a, Gazzola et al., 2007b, Gazzola and Keysers, 2008), it may be relevant to test the possible differences related to the type of view at whole-brain level in terms of action observation and execution overlapping activity.

Therefore, here we asked the same individual to observe a grasping action performed either by a fully visible model or by a hand alone (action observation conditions) and to perform a visually-guided grasping action (action execution condition) while scanned. These data may allow us to identify overlapping areas for both the observation and the execution of hand actions and how such activity might be modulated by the type of model.