Your mind's hand: Motor imagery of pointing movements with different accuracy

Abstract

Jeannerod (2001) postulated that motor control and motor simulation states are functionally equivalent. If this is the case, the specifically relevant task parameters in online motor control should also be represented in motor imagery. We tested whether the different spatial accuracy demands of manual pointing movements are reflected on a neural level in motor imagery. During functional magnetic resonance imaging (fMRI) scanning, 23 participants imagined hand movements that differed systematically in terms of pointing accuracy needs (i.e., none, low, high). In a low-accuracy condition, two big squares were presented visually prior to the imagery phase. These squares had to be pointed at alternately on a mental level. In the high-accuracy condition, two little squares had to be hit. As expected on the basis of speed–accuracy trade-off principles, results showed that participants required more time when accuracy of the imagined movements increased. The fMRI results showed a stepwise increase in activation in the anterior cerebellum and the anterior part of the superior parietal lobe (SPL) with rising accuracy needs. Moreover, we found increased activation of the anterior part of the SPL and of the dorsal premotor cortex (dPMC) when imagery included a square (i.e., in the low- and high-accuracy conditions) compared to the no-square condition. These areas have also been discussed in relation to online motor control, suggesting that specific task parameters relevant in the domain of motor control are also coded in motor imagery. We suggest that the functional equivalence of action states is due mostly to internal estimations of the expected sensory feedback in both motor control and motor imagery.

Introduction

One striking characteristic of human nature is our ability to consciously use imagination to simulate our own acts and their outcome independently from our actual physical environment. This may well represent a clear advantage for the acting individual (Desmurget and Grafton, 2000, Wolpert et al., 1995). During the last decade, such action simulation has become a major topic in cognitive neuroscience (Zentgraf et al., 2005, Decety and Grèzes, 2006, Munzert and Zentgraf, 2009). Within this discussion, Jeannerod's (2001) simulation theory has postulated a functional equivalence between imagining and executing an action. It proposes that every action involves a covert stage, and that this covert state spans the goal of the action, the means to reach it, and its sensory consequences. One situation corresponding to covert actions is the conscious simulation of one's own actions, that is, motor imagery (MI). MI is defined as an internal rehearsal of movements from a first-person perspective without any overt physical movement (Crammond, 1997, Decety and Jeannerod, 1996, Hanakawa et al., 2008, Jeannerod, 1994; see, for a review, Munzert et al., 2009). The past 20 years has seen an accumulation of behavioral findings showing that imagined actions follow the same constraints as their corresponding executed actions (Decety and Jeannerod, 1996, Ehrsson et al., 2003, Sirigu et al., 1995). For example, imagined tapping on targets of varying size retains the same temporal regularities as actual tapping on the same targets (Sirigu et al., 1995). As a result, it has been proposed that MI is a simulation that uses the motor system as a substrate (De Lange et al., 2006, Jeannerod, 2001). This is supported by several neuroimaging studies showing that roughly the same brain areas are involved in both motor execution and in MI (Decety et al., 1994, Deiber et al., 1992, Hanakawa et al., 2008, Lotze et al., 1999, Porro et al., 1996). Hence, imagined actions may be considered as actions as well; the only difference being that they are not executed (Jeannerod, 2001).

In the framework of motor control, motor skills can be classified on the basis of action functions (e.g., object manipulations) and the environmental context (see, for a taxonomy, Gentile, 2000). Taking the underlying processes involved in pointing movements as an example, motor control can be partitioned into several basal problems: trajectory planning, target-related processing, transformation of target localization into movement parameters, and selection of adequate motor efferents (Jeannerod et al., 1995, Wolpert et al., 1998). It has been argued that so-called internal models provide a computational foundation for these relevant parameters (Miall and Wolpert, 1996, Wolpert and Flanagan, 2001). A “planner” (i.e., the inverse model) provides the motor efferents that cause a desired change in state and environment, and a predictive forward model estimates the anticipated sensory outcome. For instance, a forward model calculates an estimate of the movement end-point location as output, and this can be compared to the target location. In case of a discrepancy between the two signals, an error signal is given and the motor command is corrected. The underlying internal representations and sensorimotor contingencies are built up through an individual's continuous interaction with the environment, and these support subsequent interactions with that environment, especially by providing accurate predictions of the required movement parameters (Desmurget and Grafton, 2000; Johansson and Flanagan, 2009).

Two brain regions are considered to contain inverse and forward models for the motor system: the cerebellum and posterior parietal regions, in particular, the superior parietal lobe (SPL) (Culham and Valyear, 2006; Desmurget and Grafton, 2000; Diedrichsen et al., 2005, Evangeliou et al., 2009, Sirigu et al., 1996; Blakemore and Sirigu, 2003; Wolpert et al., 1998). This is supported by research on brain-damaged patients: Cerebellar damage leads to severe deficits in fine motor adjustment and in generating precise and well-timed hand movements (Babin-Ratté et al., 1999, Ivry et al., 1988). For example, Ivry et al. (1988) reported that cerebellar patients exhibit hypermetric movement overshoots and fail to hit a target. Damage to posterior parietal sites results in a lack of sensorimotor integration accompanied by a poor representation of one's own body and the outer world (Wolpert et al., 1998).

Blakemore and Sirigu (2003) suggested that internal models are also used in MI, and that they require the retrieval of the stored forward model of the particular movement (Blakemore and Sirigu, 2003). This indicates that sensorimotor prediction might be the mechanism that makes MI and motor control equivalent.

The present study investigated whether brain areas coding for specific motor parameters within motor control (Blakemore and Sirigu, 2003, Desmurget and Grafton, 2000) are also involved in motor imagery. More precisely, we tried to elucidate the functional equivalence of motor control and motor imagery for pointing movements of varying accuracy. We applied a design with three experimental conditions asking participants to perform MI of hand movements with identical movement trajectories but different accuracy demands. Spatial accuracy was manipulated stepwise. If MI really is a covert stage of an action in its specific environment, we expect accuracy to impact on neural activation in areas associated with forward modeling, sensorimotor prediction, and the planning of an adequate movement trajectory—for example, posterior parietal areas and the cerebellum.