Neural mechanisms involved in mental imagery and observation of gait

Abstract

Brain activity during observation and imagery of gait was investigated. Sixteen subjects were scanned with a 3-Tesla MRI scanner while viewing six types of video clips: observation of gait movement (GO) from the third-person perspective, observation of stepping movement, observation of standing posture, “virtual walking” (VW) that was observation of visual scenes mimicking the visual afferent during walking, and the scrambled version of the GO and VW stimuli. In the VW condition, moving scenes provided a virtual visual environment in which subjects easily imagined as if they were actually walking from the first-person perspective. A behavioral experiment revealed a correlation of cadence during actual walking with that during imaginary walking under the influence of the VW stimuli, indicating that a gait planning mechanism was shared by actual walking and gait imagery. The VW condition activated the dorsal premotor cortex (PMd), supplementary motor area/cingulate motor area (SMA/CMA), parahippocampal gyrus, and subcortical nuclei. The GO stimuli yielded activation of the SMA, PMd, inferior frontal gyrus, and inferior parietal lobule. Moreover, the conjunction null test of GO and VW revealed common activity in the SMA/CMA and PMd, which were reportedly active during actual gait movement, in addition to visual areas. Detailed analyses of activity during stepping or standing observation supported the specificity of the SMA and PMd to GO. These findings suggest that motor planning centers of gait, including the SMA and PMd, are activated during both imagination (first-person perspective) and observation (third-person perspective) of gait behaviors.

Introduction

Walking movement requires coordinated patterns of muscle activation, anti-gravity control of the moving body mass, balance, and modulation of these controls to meet environmental demands (Patla, 1991). The central pattern generators (CPG) in the spinal cord are believed to elicit rhythmic muscle activation (Grillner, 2006). Similarly, brainstem locomotor regions and related structures in the reticular formation probably mediate gait movement in both quadrupeds and bipeds (Takakusaki et al., 2003).

However, several lines of evidence suggest that the driving of those basic locomotor modules is strongly dependent on cortical input in humans. First, patients with complete spinal cord transection are not able to walk spontaneously (Bussel et al., 1996, Calancie et al., 1994). Second, the supplementary motor area (SMA) has been reported to be involved in initiation and termination of gait (Crenna and Frigo, 1991, Jian et al., 1993, McFadyen and Winter, 1991), and patients with premotor cortex (PM) lesions exhibit locomotor difficulty when turning or negotiating narrow passages (Nutt et al., 1993). In addition, although patients with “gait apraxia” exhibit no paresis or sensory deficit, they cannot perform normal walking movement. Although the site of lesions is generally difficult to be determined in gait apraxia, a higher-order cortical center such as the SMA is reported to be responsible (Della Sala et al., 2002). Third, neuroimaging studies with near-infrared spectroscopy (NIRS) and single photon emission computed tomography (SPECT) have shown that human gait is associated with widespread cortical brain activity involving the SMA, PM, primary motor cortex, and primary somatosensory cortex, in addition to subcortical structures (Fukuyama et al., 1997, Hanakawa et al., 1999a, Hanakawa et al., 1999b, Miyai et al., 2001).

Furthermore, the interaction of gait control mechanisms with visual information from the environment likely occurs at the cortical level. Visual information from the environment significantly affects posture and gait especially during infancy and childhood (Sparto et al., 2006, Sundermier and Woollacott, 1998). Visual information sometimes ameliorates gait disturbance in Parkinson disease, and this “paradoxical gait” phenomenon is probably mediated by PM activity (Hanakawa et al., 1999a). In the real-life walking environment, we often need to modify our own gait plans by observing other people's walking behavior. In this regard, there should be a system mediating the interaction of walking behavior among people through the visual information. Interestingly, mere observation of actions involving various body parts elicits activation in motor-related areas such as the PM in a somatotopically organized fashion (Buccino et al., 2001). Similar to action observation, motor imagery shares substantial functional circuits with motor execution and involves the PM, SMA, basal ganglia, and cerebellum (Grezes and Decety, 2002, Hanakawa et al., 2003b). The cortical motor areas exhibit somatotopically aligned activity during motor imagery, as during motor execution (Ehrsson et al., 2003, Hanakawa et al., 2005).

To explore higher-level cortical gait planning centers in humans, we investigated brain activity in healthy volunteers, using functional magnetic resonance imaging (fMRI). Since the MRI environment excluded direct assessment of gait movement, we used a paradigm assessing activation of the motor network during imaginary gait (Jahn et al., 2004, Malouin et al., 2003, Sacco et al., 2006). Recently, a virtual reality environment that supplies realistic visual feedback mimicking that during real walking has been used to produce imaginary locomotion (Buekers et al., 1999, Fung et al., 2006, Hollman et al., 2006). Slobounov et al. combined such a virtual reality environment with fMRI in order to create the environment that is impossible to materialize in the MRI scanner (Slobounov et al., 2006). In the present study, we employed visual stimuli that appeared to elicit mental imagery of gait by non-interactive, passive viewing (“virtual walking”). We hypothesized that such stimuli would help invoke motor planning programs for the subject's own gait and produce active mental rehearsal of motor acts (gait imagery from the first-person perspective). This idea was tested in a behavioral experiment separately from the fMRI experiment. Other experimental conditions included observation of other people's walking-related behavior (“gait observation”), which should concern visual recognition of gait movement performed by others. The recognition and understanding of other people's walking behavior may be mediated by a sort of gait imagery from the third-person perspective and may also involve higher-order gait planning centers in the cerebral cortex. It was hypothesized that such gait planning centers could be situated in the posterior frontal lobe, particularly the SMA and PM, which have been consistently shown to be activated during actual walking.