HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Full Text (PDF) All Versions of this Article: 97/4/2887    most recent 00757.2006v1 Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in ISI Web of Science Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via ISI Web of Science (5) Citing Articles via Google Scholar Google Scholar Articles by Herter, T. M. Articles by Scott, S. H. Search for Related Content PubMed PubMed Citation Articles by Herter, T. M. Articles by Scott, S. H.

Characterization of Torque-Related Activity in Primary Motor Cortex During a Multijoint Postural Task

Department of Anatomy and Cell Biology, Canadian Institute of Health Research Group in Sensory-Motor Systems, Centre for Neuroscience Studies, Queen's University, Kingston, Ontario, Canada

Submitted 21 July 2006; accepted in final form 24 January 2007

	ABSTRACT

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION GRANTS ACKNOWLEDGMENTS REFERENCES

 
The present study examined neural activity in the shoulder/elbow region of primary motor cortex (M1) during a whole-limb postural task. By selectively imposing torques at the shoulder, elbow, or both joints we addressed how neurons represent changes in torque at a single joint, multiple joints, and their interrelation. We observed that similar proportions of neurons reflected changes in torque at the shoulder, elbow, and both joints and these neurons were highly intermingled across the cortical surface. Most torque-related neurons were reciprocally excited and inhibited (relative to their unloaded baseline activity) by opposing flexor and extensor torques at a single joint. Although coexcitation/coinhibition was occasionally observed at a single joint, it was rarely observed at both joints. A second analysis assessed the relationship between single-joint and multijoint activity. In contrast to our previous observations, we found that neither linear nor vector summation of single-joint activities could capture the breadth of neural responses to multijoint torques. Finally, we studied the neurons' directional tuning across all the torque conditions, i.e., in joint-torque space. Our population of M1 neurons exhibited a strong bimodal distribution of preferred-torque directions (PTDs) that was biased toward shoulder-extensor/elbow-flexor (whole-limb flexor) and shoulder-flexor/elbow-extensor (whole-limb extensor) torques. Notably, we recently observed a similar bimodal distribution of PTDs in a sample of proximal arm muscles. This observation illustrates the intimate relationship between M1 and the motor periphery.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION GRANTS ACKNOWLEDGMENTS REFERENCES

 
It is known that primary motor cortex (M1) is an important region for executing volitional motor tasks (Hepp-Reymond 1988; Porter and Lemon 1993), although its specific role remains an area of active investigation. One key issue that remains unresolved is what portion of the motor periphery is reflected in the activity of individual neurons (Scott 2000). Specifically, is neural activity related to relatively localized motor patterns at a single joint or distributed motor patterns across multiple joints?

A number of complementary approaches have provided insight into this issue. One set of approaches examined relationships between the neural and muscle activity. Spike-triggered averaging techniques demonstrated that some M1 neurons (corticomotoneurons) synapse directly onto pools of motoneurons that innervate one or more muscles acting at multiple joints of the limb (Buys et al. 1986; Fetz and Cheney 1980; McKiernan et al. 1998). However, corticomotoneurons are relatively rare and spike-triggered averaging examines patterns of connectivity with muscles but does not inherently address whether the activity of individual neurons reflects a similar, more focal, or broader representation of the motor periphery. Another approach relates the pattern of neural activity to the pattern of muscle activity during behavior. Several previous studies showed that neural activity correlates with the activity of individual or groups of muscles acting at a single joint or across multiple joints (Bennett and Lemon 1996; Holdefer and Miller 2002) and reflect details on the timing and magnitude of muscle activity (Drew 1993; Scott 1997; Smith et al. 1975).

A second set of approaches related neural activity to kinesiological parameters such as joint motion, joint torque, or forces applied to the limb. Because the motor system contains many more muscles than mechanical degrees of freedom (df), such kinesiological parameters provide a simpler parameter space to examine the relationship between M1 activity and the motor periphery. By design, single-joint studies describe how neural activity correlates with motor patterns at a single joint (Cheney and Fetz 1980; Crutcher and Alexander 1990; Evarts 1968, 1969; Evarts et al. 1983; Fromm 1983; Kakei et al. 1999; Thach 1978), but cannot address how neural activity reflects motor patterns at multiple joints. Furthermore, most studies of volitional whole-arm control correlated neural activity to hand-based parameters (e.g., Caminiti et al. 1990; Georgopoulos et al. 1982, 1992; Kalaska et al. 1989; Moran and Schwartz 1999; Scott and Kalaska 1997; Sergio and Kalaska 2003) and thus they were not designed to identify whether neural activity is related to the shoulder, elbow, or both joints.

Perhaps the most extensive analysis of neural activity related to motor patterns at different joints was conducted by Schieber and colleagues, who studied neural activity related to individual finger movements (Poliakov and Schieber 1999; Schieber and Hibbard 1993). They found that individual neurons were generally related to independent movements of more than one digit, suggesting that neurons may reflect a substantive portion of the motor periphery, at least for the hand. However, the activity of intrinsic and extrinsic finger muscles was often modulated by movements of unspanned digits (Schieber 1995); thus neural processing could reflect motion of many digits/joints or motor patterns of a small, select group of muscles.

We introduced a planar, whole-limb paradigm that uses robotic technology to both monitor and perturb the mechanics of the shoulder and elbow joints independently (Scott 1999). Compared with the complex musculoskeletal organization of the hand, this whole-limb paradigm is relatively simple because most proximal arm muscles are monoarticular (spanning only the shoulder or elbow) with only four biarticular muscles representing about 19% of the total physiological cross-sectional area of all shoulder and elbow muscles (Cheng and Scott 2000). The objective of this whole-limb paradigm is to facilitate exploring linkages between behavior, limb mechanics, muscle activity, and neural processing.

The present study examined how neural activity in M1 reflects torques applied to the shoulder, elbow, or both joints during a whole-limb postural task. This behavioral task permitted us to examine torque-related activity, commonly recorded in previous single-joint studies, at two mechanical df spanning two joints. We directly compared neural responses to single-joint torques applied to each joint separately and show that most M1 neurons respond reciprocally to opposing torques at one or both joints. Furthermore, nearly all neurons that are coactivated or coinhibited by opposing torques at one joint nevertheless exhibit unimodal directional tuning.

A second value of this paradigm was revealed in our parallel study of EMG activity recorded from proximal arm muscles during this postural task (Kurtzer et al. 2006a). The behavioral goal of this task was to maintain the hand at a spatial target while counteracting mechanical torques applied to the shoulder and/or elbow. Across all torque conditions, this task sampled a uniform distribution of torques at the shoulder and elbow (i.e., in joint-torque space). However, the maximal activity of proximal arm muscles was biased toward either whole-limb flexor or whole-limb extensor torques, thereby creating a bimodal distribution of preferred-torque directions (PTDs) in joint-torque space. If neural processing mirrors the behavioral requirements inherent in a task, then an equal proportion of M1 neurons should have represented each joint torque in this postural task. Alternatively, if neural activity reflects the output patterns of muscles, then a greater proportion of neurons should have represented whole-limb flexor and whole-limb extensor torques in this postural task. Here we demonstrate that the maximal activity of M1 neurons exhibits the same bias toward whole-limb flexor and whole-limb extensor torques that was observed in proximal arm muscles.

The current study extends and elaborates on observations that we previously reported (Cabel et al. 2001).

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION GRANTS ACKNOWLEDGMENTS REFERENCES

 
Apparatus and task

Four male rhesus monkeys (Macaca mulatta, 6 to 12 kg) were used in this study following procedures approved by the Queen's University Animal Care Committee. They were trained to perform motor tasks with their right arm while wearing KINARM (BKIN Technologies, Kingston, Canada), a robotic exoskeleton that supports the arm, permits planar shoulder and elbow motion, and can apply mechanical torques at the shoulder and/or elbow (Scott 1999). A virtual reality system presented visual targets within the limb's movement plane while permitting the monkeys to view their entire limb.

The current task was designed to examine neural activity related to static joint torques at constant limb geometry. At the beginning of each trial a constant torque was applied at the shoulder, elbow, both, or neither joint. After a short delay a visual target (6-mm radius) was displayed near the center of the arm's workspace (angles of about 30° at the shoulder and 90° at the elbow) where passive viscoelastic forces are relatively small (Graham et al. 2003). The monkeys were required to maintain their hand within an acceptance window (8-mm radius) centered on the visual target for 3,000 to 4,000 ms. The task incorporated nine torque conditions including four single-joint (shoulder flexion, SF; shoulder extension, SE; elbow flexion, EF; and elbow extension, EE), four multijoint (SF + EF; SF + EE, SE + EF, and SE + EE), and an unloaded baseline condition (Cabel et al. 2001; Kurtzer et al. 2005). Torques of ±0.12 Nm were applied to each joint in Monkeys A–C and torques of ±0.32 Nm were used in Monkey D because of his larger size. These torque conditions were presented in a pseudorandom block design with each block repeated five times for a total of 45 trials. A fixed magnitude of torque was always applied at each joint such that the total torque magnitude for multijoint conditions was 2 greater than single-joint conditions. This design feature was important because it permitted us to treat shoulder and elbow torques as independent factors.

Data collection

Neural data were collected using standard extracellular recording techniques following surgical implantation of a cylindrical recording chamber over the shoulder/elbow representation of the left M1 (Scott and Kalaska 1997). Microelectodes were advanced through M1 until neural activity was observed in response to active or passive movements of the shoulder and/or elbow. Single neurons were then isolated and neural activity was recorded from all neurons that were related to active or passive movements of the shoulder and/or elbow, but not the wrist or fingers.

Joint angles, velocities, and torques applied by the robot were recorded at 1 kHz (Monkeys A–C) or 4 kHz (Monkey D). Cartesian hand positions and tangential hand velocity were calculated from joint angles and velocities. The equations of motion for a two-joint, two df system that included the influence of the mechanical properties of the arm and robotic device were used to calculate the instantaneous torques exerted by the monkey at the shoulder and elbow joints (Cheng and Scott 2000; Scott 1999). These were used to verify that the motors were properly functioning and the monkeys were adequately maintaining static posture. Data sets were excluded from analyses if the mean torque exerted at each joint was not within ±0.02 Nm (and opposite in sign) of the commanded torque.

Data analyses

All analyses examined neural activity with respect to the joint torques that counteracted the mechanical torques applied in each condition. Specifically, we studied torque-related activity, defined as changes in neural activity during loaded conditions relative to the unloaded baseline condition. For example, SF refers to the change in neural activity relative to baseline when a shoulder-flexor torque was generated by the monkey to compensate for an applied shoulder-extensor torque.

Visual analysis of tangential hand velocity indicated that the hand decelerated for 1,000 ms after entering the target's acceptance window (Fig. 1A). These movements appeared to influence neural activity because the mean and SD of neural activity also decreased during this period (Fig. 1B). All data were therefore analyzed by using their mean values obtained during the period from 1,000 to 3,000 ms after the hand entered the target's acceptance window.

View larger version (10K): [in this window] [in a new window]   FIG. 1. Temporal dynamics after the hand entered the target's acceptance window (averaged across all torque conditions). A: tangential hand velocity (mean ± SD). B: mean (solid line) and SD (dash–dotted line) of neural activity. Vertical lines in each panel indicate the onset of the period that data were analyzed (1,000–3,000 ms after the hand entered the target's acceptance window).

M1 TOPOGRAPHY. Three-dimensional reconstructions of M1 recording sites were created for Monkeys A–C. Postmortem examination of the cortical surface was used to determine the angle of orientation of the central sulcus relative to recording chamber in each monkey. The stereotaxic coordinates of each neuron were then rotated to view the data relative to the orientation of the central sulcus. The center of mass of recorded neurons was then identified for each monkey and used to align the medial–lateral axes across monkeys. This permitted group analyses of the medial–lateral distribution of neurons related to the torques at the shoulder and/or elbow.

SINGLE-JOINT ANALYSES. These analyses uncovered important relationships between single-joint torques and torque-related activity. A series of t-tests identified neurons that were significantly modulated by their respective flexor and extensor torques, relative to the unloaded baseline condition (P < 0.05). We also created three scores ranging continuously from –1 to 1.

Reciprocal–comodulation (R-C) scores were calculated for the shoulder and elbow

(1)

(2)

These scores identified whether neurons that responded to flexor and extensor torques exhibited reciprocal modulation (i.e., equal in magnitude and opposite in direction, R-C Score = –1), unidirectional activity (i.e., related to flexor torque only or extensor torque only, R-C Score = 0), or coexcitation/coinhibition (equal in magnitude and direction, R-C Score = 1).

Flexor–extensor (F-E) scores were also calculated for the shoulder and elbow

(3)

(4)

These scores established whether torque-related activity was related to extensor torque only (F-E Score = –1), extensor and flexor torques equally (F-E Score = 0), or flexor torque only (F-E Score = 1).

A shoulder–elbow (S-E) score was calculated across both joints

(5)

This score determined whether torque-related activity was related to the elbow only (S-E Score = –1), both joints equally (S-E Score = 0), or the shoulder only (S-E Score = 1).

PREDICTING MULTIJOINT ACTIVITY. Our preliminary report based on two monkeys (Cabel et al. 2001) found that changes in neural activity related to multijoint torques could be accurately predicted by vector summation of constituent single-joint activities. We reexamined this issue in our sample of directionally tuned neurons (bootstrap statistic, P < 0.01). Changes in activity at the shoulder (S) and elbow (E) were used to predict multijoint activity using linear (MJ_l) and vector (MJ_v) summation models

(6)

(7)

DIRECTIONAL TUNING IN JOINT-TORQUE SPACE. We examined issues related to multijoint control by calculating directional tuning properties in joint-torque space. For this particular coordinate frame, neural activities related to the joint torques exerted across the shoulder and elbow joints are represented along the x and y axes, respectively. This contrasts with the mechanical torques applied by the robot across the shoulder and elbow joints, which are equal and opposite the joint torques exerted by the monkey in our static postural task. Using flexor torque as positive and extensor torque as negative, trigonometric convention yields SF = 0°, EF = 90°, SE = 180°, and EE = 270°.

Directional tuning features were calculated with the plate method, which describes several features of directional tuning without assuming an underlying tuning function (Gribble and Scott 2002). This method characterized the "mass distribution" of torque-related activity by assuming that activity changes linearly between sampled torque directions and that torque magnitude is equal for each torque direction. By maintaining constant torques at the shoulder and elbow, the overall torque magnitude was greater in the multijoint conditions (see Apparatus and tasks). Multijoint activity was consequently divided by 2 to normalize torque-related activity across all torque conditions; normalization was carried in this specific analysis with only the plate method. The lowest activity across all trials was then subtracted so that all values were 0.

The location of the center of mass, or centroid, in joint-torque space was used to calculate four directional tuning features. 1) Preferred-torque direction (PTD) describes the angle associated with the greatest increase in activity in joint-torque space. This was calculated as the direction of the centroid relative to the origin in joint-torque space. Importantly, PTDs were calculated as the average of two separate PTDs, one obtained from the four single-joint conditions and the other from the four multijoint conditions (Cabel et al. 2001). Because the multijoint PTD of each neuron was identical irrespective of whether multijoint activity was normalized, the overall PTD was unaffected by normalization. 2) Ratio of inertia (Ir) conveys information analogous to tuning widths calculated with model-based functions such as the von Mises function. This was calculated as the ratio of torque-related activity perpendicular to the PTD axes versus torque-related activity along the PTD axes yielding values ranging from 0 to 1. A cosine obtains an Ir value of 0.44, whereas smaller values denote sharper tuning and larger values denote broader tuning. 3) Torque-slope expresses the change in torque-related activity per unit torque along the PTD. This was calculated by normalizing the magnitude of the centroid (Cr), defined as the radial distance between the centroid and origin, by the torque magnitude (0.12 or 0.32 Nm). 4) Excitation–inhibition score (E-I score) describes the relationship between increases in torque-related activity at the PTD and decreases in torque-related activity opposite the PTD. Torque-related activities at and opposite the PTD (PTD and OPP, respectively) were obtained using linear interpolation and the E-I score was then calculated

(8)

E-I scores typically range from –1 to 1. Positive (negative) E-I scores occur when the magnitude of excitation at the PTD is greater (lesser) than the magnitude of inhibition opposite the PTD. Values near 0 arise when the magnitudes of excitation and inhibition are similar. E-I scores could go beyond 1 (–1) if both PTD and OPP exhibited excitation (inhibition).

The preceding measures were examined for those neurons with only statistically significant directional tuning as determined with a nonparametric "bootstrapping" test (Scott and Kalaska 1997). This statistic was calculated by comparing the actual value of Cr with bootstrap values of Cr obtained after the torque-related activity was randomly reassigned across all trials. A neuron was considered to have significant directional tuning if fewer than 100 of 10,000 bootstrap values of Cr were greater than the actual value of Cr (P < 0.01). This method produced results that were similar to those obtained from bootstrap methods using mean vector length (Cabel et al. 2001; Scott and Kalaska 1997) and F-tests from planar regression models [similar proportions (85, 89, and 90%) of torque-related units exhibited significant directional tuning].

Reliability of the plate method was verified by comparing its results with those obtained from planar regression models that did not require normalization of data. In brief, the mean values of neural activity and actual torques exerted at the shoulder and elbow (obtained from the equations of motion) were used to calculate planar regression models for each neuron. The shoulder and elbow coefficients from these models were used to calculate PTDs and torque-slopes for each neuron (see Kurtzer et al. 2005). Directly comparing the two methods indicated that they produced similar PTDs (mean absolute difference = 4°) and torque-slopes (r = 0.96).

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION GRANTS ACKNOWLEDGMENTS REFERENCES

 
Categorization and topography

In all, 359 M1 neurons were recorded in four monkeys (Monkeys A–D: 89, 61, 84, and 125, respectively). The activity of some neurons was modulated by torques at one joint only (Fig. 2A), whereas the activity of other neurons was modulated by torques at both joints (Fig. 2B).

View larger version (14K): [in this window] [in a new window]   FIG. 2. Rasters and instantaneous firing rates illustrating exemplar patterns of neural activity during the analysis epoch (1,000–3,000 ms after the hand entered the target's acceptance window). A: single-joint neuron with torque-related activity (i.e., changes in activity relative to baseline, middle subpanel) that increased for shoulder-flexor torques, decreased for shoulder-extensor torques, and was largely uninfluenced by torques at the elbow. B: multijoint neuron with torque-related activity that increased for shoulder-extensor/elbow-flexor torques and decreased for shoulder-flexor/elbow-extensor torques. SF, shoulder flexor; EF, elbow flexor; SE, shoulder extensor; EE, elbow extensor.

The relationship between the different categories of torque-related neurons and their anatomical location within M1 was investigated using three-dimensional reconstructions of M1 in Monkeys A–C (Fig. 3A). Across the three monkeys, neurons related to torques at the shoulder only (red circles), elbow only (green circles), both joints (blue circles), or neither joint (black circles) were highly intermingled across the cortical surface (Fig. 3, B–D). The only evidence of anatomical segregation was observed in Monkey B, where shoulder-only neurons were located medial to elbow-only [Kolmogorov–Smirnov (K-S) test, P < 0.001] and multijoint neurons (K-S test, P = 0.02). In Monkeys A and C, each category exhibited a similar distribution along the medial–lateral axes of M1 (K-S tests, P > 0.1). Furthermore, there was no evidence of anatomical segregation along the medial–lateral axes of M1 when data were grouped across all three monkeys (Fig. 3E) (K-S tests, P > 0.1).

View larger version (38K): [in this window] [in a new window]   FIG. 3. Recording sites. A: top-down view of the location of the shoulder–elbow region of primary motor cortex (M1, left) and the coordinates of its 3-dimensional reconstructions (right). B–D: dorsal and sulcal views depicting the anatomical locations of recording sites (Monkeys A–C, respectively) of M1 neurons related to the shoulder only (red circles), elbow only (green circles), both joints (blue circles), or neither joint (black circles). Penetration depths are displayed relative to the electrode's initial electrical contact (assumed to be the top of the dura). Grid spacing = 1 mm. E: illustration of the cumulative sum (percentage) of neurons along the medial–lateral axes of M1 grouped across monkeys. Bin width = 0.25 mm.

Of the 146 shoulder neurons, 44 (30%) were significantly modulated by flexor torque only, 30 (21%) by extensor torque only, and 26 (18%) by both flexor and extensor torques (t-test comparing single-joint flexor and extensor torques with unloaded baseline, P < 0.05). Of the 140 elbow neurons, 25 (18%) were significantly modulated by flexor torque only, 35 (25%) by extensor torque only, and 22 (16%) by both flexor and extensor torques (t-test, P < 0.05). The remaining 46 (32%) shoulder and 58 (41%) elbow neurons exhibited similar patterns of torque-related activity but failed to reach statistical criteria because of the weaker statistical power of the t-test (df = 8) compared with the ANOVA (df = 2, 36).

Relative to baseline activity in the unloaded condition, flexor and extensor torques at the shoulder (Fig. 4A) and elbow (Fig. 4B) were associated with increases and decreases in activity that ranged from –20 to 28 spikes/s at 0.12 Nm in Monkeys A–C and from –35 to 28 spikes/s at 0.32 Nm in Monkey D. Notably, the magnitude of changes in activity appears to be relatively independent of baseline activity, except that overall activity (with baseline) cannot drop below zero spikes/s (Fig. 4, A and B). As a result, changes in torque-related activity were broadly distributed with mean activities near zero (Fig. 4, C–F).

View larger version (21K): [in this window] [in a new window]   FIG. 4. Changes in neural activity related to single-joint torques. A and B: torque-related activity of shoulder (A) and elbow (B) neurons relative to their baseline activity. Neural responses to flexor and extensor torques at the shoulder (SF and SE) and elbow (EF and EF) are shown separately for each shoulder and elbow neuron. Circles, significant responses to flexor torques (P < 0.05); diamonds, significant responses to extensor torques (P < 0.05); small dots, nonsignificant responses to flexor and extensor torques (P > 0.05); closed symbols, Monkeys A–C; open symbols, Monkey D. Diagonal lines indicate the largest negative change in activity that can occur for any given baseline activity. C–F: distributions of neural responses to flexor and extensor torques at the shoulder (C and E) and elbow (D and F).

View larger version (21K): [in this window] [in a new window]   FIG. 5. Comparison of responses to single-joint flexor and extensor torques. A and D: responses to flexor vs. extensor torques at the shoulder (A) and elbow (D). Abbreviations same as Fig. 4. B and E: distributions of reciprocal–comodulation (R-C) scores for shoulder (B) and elbow (E) neurons. Reciprocal = –1, unidirectional = 0, coexcitation/coinhibition = 1. C and F: distributions of flexor–extensor (F-E) scores for shoulder (C) and elbow (F) neurons. Extensor torque only = –1, flexor and extensor torques equally = 0, flexor torque only = 1.

View larger version (19K): [in this window] [in a new window]   FIG. 6. Comparison of responses to single-joint torques at the shoulder and elbow. A: responses of shoulder-only (red), elbow-only (green), and multijoint (blue) neurons to single-joint torques at the shoulder (SF – SE) vs. the elbow (EF – EE). Closed circles, Monkeys A–C; open circles, Monkey D. B: distribution of shoulder–elbow (S-E) scores of shoulder-only (red), elbow-only (green), and multijoint (blue) neurons. Elbow only = –1, shoulder and elbow equally = 0, shoulder only = 1. C: comparison of shoulder vs. elbow R-C scores of multijoint neurons.

An interesting question is whether neurons that exhibited coexcitation/coinhibition related to torques at one joint (Fig. 5, C and D) responded in a similar manner at both joints. Figure 6C compares shoulder and elbow R-C scores for the 76 multijoint neurons. Notably, most (n = 48, 63%) multijoint neurons exhibited reciprocal activity at both joints, another 22 (29%) displayed coexcitation/coinhibition at one joint and reciprocal activity at the other, and only six (8%) showed coexcitation/coinhibition at both joints. These values did not differ significantly from the expected values of a random distribution (² test, P > 0.1).

Predicting multijoint activity

Our initial report found the multijoint activity could be predicted from vector summation of constituent single-joint activities (Cabel et al. 2001). Here we reinvestigated this issue and found that neither linear nor vector summation models accurately predicted multijoint activity across our entire sample of directionally tuned neurons (Fig. 7, A and B). On average, linear summation overestimated multijoint activity (mean residual error = 2.6 spikes/s; t-test, P < 0.001) and vector summation underestimated multijoint activity (mean residual error = –1.6 spikes/s; t-test, P < 0.001).

View larger version (19K): [in this window] [in a new window]   FIG. 7. Prediction of multijoint activity from constituent single-joint activities. A and B: relationships of predicted multijoint activities from the linear (A) and vector (B) summation models vs. actual multijoint activity observed across all neurons with significant directional tuning. C and D: relationships of predicted vs. actual multijoint activities in the subset of multijoint neurons. Closed circles, neurons whose multijoint activity does not differ significantly from linear (A, C) and vector (B, D) summation; open circles, neurons whose multijoint activity differs significantly from the predicted activity.

Tuning properties in joint-torque space

Of the 210 torque-related neurons flagged by the ANOVA, 178 (85%) exhibited significant directional tuning in joint-torque space (plate method bootstrap statistic, P < 0.01). The various directional tuning properties in joint-torque space (see METHODS) are illustrated with four exemplar neurons in Fig. 8. Figure 8A illustrates neuron 11531b (see also Fig. 2A), which was maximally active during torque conditions that involved shoulder-flexor torque and minimally active during torque conditions that engaged shoulder-extensor torque. The pattern of torque-related activity across all joint-torque directions gave rise to a PTD of 6°, cosine-like sensitivity to changes in joint-torque direction (Ir = 0.42 compared with 0.44 for a cosine) and a torque-slope of 129 spikes·s^–1·Nm^–1 (Cr = 15.5 spikes/s, torque = 0.12 Nm). Furthermore, the magnitude of excitation at the PTD was similar to the magnitude of inhibition opposite the PTD (E-I score = –0.02).

View larger version (22K): [in this window] [in a new window]   FIG. 8. Exemplar patterns of directional tuning. A–D: 4 exemplar neurons with different patterns of directional tuning. Each panel consists of 3 different subpanels. Left subpanels: linear plots of overall activity vs. joint-torque angle. Unloaded baseline activity and cosine fits are shown as dotted and dashed lines, respectively. Middle and right subpanels, respectively: polar plots of overall activity and torque-related activity (baseline removed) in joint-torque space. SF = 0°, EF = 90°, SE = 180°, EE = 270°.

Figure 8 also illustrates the importance of removing baseline activity before calculating Ir. Neuron 11531a (A) exemplifies a case where the minimum overall activity including baseline was near 0 spikes/s (torque direction = 180°), resulting in similar values of Ir for the overall activity (middle plot) and torque-related activity (right plot) in joint-torque space (Ir values of 0.43 and 0.42, respectively). In contrast, neurons 33171b (B) and 12811a (D) had minimum overall activities that were well above 0 spikes/s (torque direction values of 0 and 180°, respectively) resulting in broader tuning for their overall activity (Ir values of 0.72 and 0.94, respectively) compared with their torque-related activity (Ir values of 0.50 and 0.82, respectively).

A number of interesting results were observed across the entire population of directionally tuned neurons. Most notably, PTDs exhibited a significant bimodal distribution (bimodal Rayleigh test, mean vector = 0.26, P < 0.001) oriented along the axes of 141–321° (Fig. 9A). Although PTDs also exhibited a significant, albeit weaker, unimodal distribution (unimodal Rayleigh test, mean vector = 0.17, P < 0.01), the mean vector for the bimodal distribution was consistently greater for each individual monkey. Furthermore, this bimodal distribution confirms the preference for whole-limb flexor and whole-limb extensor torques that we observed when single-joint torques were compared across the shoulder and elbow (Fig. 6A).

View larger version (14K): [in this window] [in a new window]   FIG. 9. Properties of directional tuning in joint-torque space. A: polar histogram showing the distribution of preferred-torque directions (PTDs) in joint-torque space. Bin width = 15°. Solid line indicates the preferred-torque axes of the bimodal distribution (141–321°). B–D: linear histograms showing the distributions of torque-slopes (B), ratio of inertia (Ir) values (C), and excitation–inhibition (E-I) scores (D). E: relationship between Ir and E-I scores. All panels include only those neurons with significant directional tuning in joint-torque space (bootstrap statistic, P < 0.01).

Because values of torque-slope and Ir were broadly distributed, we speculated that neurons with PTDs near 45 and 225° may exhibit higher torque-slopes or different tuning to compensate for their diminutive numbers. However, neither torque- slopes nor tuning widths (Ir) varied significantly between sparsely and densely populated regions of joint-torque space (Wilcoxon rank-sum tests, P > 0.1). In fact, a negative correlation between Ir and E-I scores (r = –0.49; t-test for correlation, P < 0.001) was the only significant relationship that we observed between the various properties of directional tuning (Fig. 9E). The presence of this relationship indicates that neurons with greater excitation than inhibition were sharply tuned at their PTD, whereas neurons with greater inhibition than excitation were sharply tuned opposite their PTD (sharp tuning opposite the PTD necessitates broad tuning at the PTD).

Because neural activity cannot decrease below zero spikes/s, we speculated that neurons with low baseline activity may account for the observation that increases in torque-related activity at the PTD were greater than decreases opposite the PTD (Fig. 9D). Accordingly, the mean E-I score of neurons with low baseline activity (<10 spikes/s) was significantly greater than zero (mean E-I score = 0.64; t-test, P < 0.001), whereas the mean E-I score of neurons with higher baseline activities (>10 spikes/s) was not different from zero (mean E-I score = 0.04; t-test, P > 0.1). As a result of the relationship between Ir and E-I scores (Fig. 9E), the mean Ir of neurons with low baseline activities was significantly sharper than a cosine (median Ir = 0.28; Wilcoxon signed-rank test, P < 0.001), whereas the mean Ir of neurons with higher baseline activities was not different from a cosine (median Ir = 0. 45; Wilcoxon signed-rank test, P > 0.1).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION GRANTS ACKNOWLEDGMENTS REFERENCES

 
A major goal of this study was to examine what portion of the motor periphery is reflected in the activity of individual M1 neurons by characterizing how neural activity is modulated by variations in torque exerted at two separate joints. Our approach provides a direct extension of studies that examine neural processing during single-joint motor tasks to consider multiple joints. Our methodology also shares certain similarities to previous studies that correlate load-related activity to hand-based parameters. We therefore included complementary analyses reflective of previous single-joint and hand-based studies. Furthermore, the detailed descriptions of torque-related activity provide an important foundation for developing and assessing computational models of M1 processing.

In summary, we observed that many neurons in the shoulder/elbow region of M1 were quantitatively related to joint torques at the shoulder only (33%), elbow only (30%), or both joints (36%) and that these neurons were highly intermingled across the cortical surface. Relative to their unloaded baseline activity, most torque-related neurons were reciprocally modulated by opposing flexor and extensor torques at a single joint (Fig. 5, B and E). A small number of neurons were coexcited or coinhibited by opposing torques at one joint, but these neurons generally exhibited reciprocal modulation at the other joint (Fig. 6C). Notably, our population of M1 neurons exhibited a bimodal distribution of PTDs that was biased toward whole-limb flexor and whole-limb extensor torques (Fig. 9A). Other properties of directional tuning (torque-slope and tuning width) exhibited broad distributions (Fig. 9, B and C) that were unrelated to this bias. These observations have several implications regarding the neural basis of multijoint control, which are discussed in the following text.

M1 topography

Classical studies of M1 in humans (Penfield and Rasmussen 1950) and monkeys (Woolsey et al. 1952) observed that movements of body parts evoked by stimulation were broadly segregated along the medial–lateral axes. Specifically, the legs, trunk, arms, face, and mouth were found to be represented within anatomically distinct subregions of M1. It is debated, however, whether there is a functional organization within these major regions (reviewed in Sanes and Schieber 2001; Schieber 2001). In the arm representation, for example, some studies observed a random organization (Donoghue et al. 1992; Gould et al. 1986), whereas other studies observed that the proximal arm (shoulder–elbow) is represented within a horseshoe-shaped region surrounding the distal arm (wrist–hand) (Kwan et al. 1978; Park et al. 2001).

The current study addressed whether neurons within the shoulder–elbow region of M1 exhibit an orderly topography. Although our study was limited to neurons within the medial branch of the horseshoe organization (hand/wrist-related neurons were found laterally), we found that neurons related to the shoulder only, elbow only, or both joints were intermingled across the surface of M1 and did not exhibit distinct segregation along the medial–lateral axes of M1. Similar results were reported for the hand representation, in which single neurons are generally related to movements of multiple digits and do not exhibit an obvious somatotopic organization (Poliakov and Schieber 1999; Schieber and Hibbard 1993). These observations support the notion that M1 possesses only a broad segregation of body parts. Substantive overlap in the neural representations of the shoulder and elbow may reflect the behavioral requirements of many motor tasks in which these joints are coordinated together to move the hand in space (Sanes and Schieber 2001; Schieber 2000).

Is cocontraction reflected in a small population of torque-related neurons?

In everyday life, common motor behaviors may be affected by random mechanical perturbations. For example, our arms randomly experience clockwise and counterclockwise perturbations while driving over rough terrain. Coactivation of antagonistic muscles (i.e., cocontraction) is one strategy to reduce the influence of limb perturbations that may occur while trying to maintain a static posture (e.g., Darainy et al. 2004; Franklin and Milner 2003; Gomi and Osu 1998; McIntyre et al. 1996).

Surprisingly little is known about the neural basis of cocontraction because few studies examined neural activity during tasks that promote cocontraction. At the level of the cortex, Humphrey and Reed (1983) examined the responses of M1 neurons during slow and fast wrist perturbations that respectively produced reciprocal activation and cocontraction of opposing wrist muscles. They found two distinct populations of neurons: one that was modulated only during rapid perturbations and one that was modulated only during slow perturbations. This finding suggests that reciprocal activation and cocontraction of opposing wrist muscles might be represented by distinct neural processes in M1. However, muscles of the elbow and shoulder were also more active during rapid perturbations than slow perturbations and thus it is possible that one population was related to the wrist, whereas the other was related to the shoulder and elbow.

Indirect evidence for distinct representations of reciprocal activation and cocontraction may also be interpreted from single-joint studies of torque-related activity in M1 (Evarts et al. 1983; Werner et al. 1991). These studies found that most torque-related M1 neurons exhibited a reciprocal pattern of activity in response to opposing torques; i.e., they were excited by torques applied in one direction and inhibited by torques applied in the opposing direction. However, they also found that a considerable number of torque-related neurons were excited or inhibited by both opposing torques. Because opposing torques primarily activate antagonistic muscle groups, these results suggest that this population of neurons may be related to cocontraction.

If neurons with similar responses to opposing single-joint torques are related to cocontraction, then neural correlates of cocontraction should also be observed in whole-limb studies. Specifically, one might expect to find neurons that exhibit bimodal or uniform changes in force- or torque-related activity relative to unloaded baseline. However, two comprehensive whole-limb studies, one postural and one isometric, did not report bimodal or uniform changes in force-related activity relative to baseline (Kalaska et al. 1989; Sergio and Kalaska 2003). The current study may explain why these whole-limb studies did not report such correlates of cocontraction. We found that a sizable proportion of multijoint neurons (28/76, 37%) showed coexcitation/coinhibition at one of the two joints. If these neurons were related to cocontraction of antagonist muscles, one would expect that many would exhibit coexcitation/coinhibition at both joints and few would exhibit unimodal tuning. Contrary to these expectations, few of these neurons (6/28, 21%) exhibited coexcitation/coinhibition at both joints (Fig. 6C) and nearly all (26/28, 93%) were unimodally tuned (e.g., Fig. 8C).

Although our results suggest that torque-related M1 neurons are not related to cocontraction, it is important to recognize that our task was not specifically designed to explore the neural basis of cocontraction. Several recent studies of human behavior examined patterns of cocontraction during posture (Darainy et al. 2004; Franklin and Milner 2003) and reaching (Burdet et al. 2001; Milner and Franklin 2005) and may provide a foundation for future studies designed to explore the neural basis of cocontraction.

Relationship between M1 and muscle activations

There are many different parameters or levels of information that can be used to describe the kinetics of forces applied to a limb: 1) contact forces between the limb/hand and load, 2) joint torques representing the combined action of all muscles spanning a given joint, 3) individual muscle forces, and even 4) muscle activation patterns. The difference between these kinetic descriptions reflects how much the properties of the musculoskeletal system are incorporated into the kinetic measure. Contact forces do not consider any aspect of the limb except the site of contact. Joint torques include the influence of limb geometry and intersegmental dynamics. Muscle forces consider the mechanical action of individual muscles. Muscle activation patterns additionally include the mechanical properties of muscle including force–length and force–velocity relationships and excitation–contraction coupling.

Neural activity could be related to any of the above parameters, although it remains unclear which parameters are represented by supraspinal processes. Theoretical frameworks based on sensorimotor transformations (Hollerbach 1982; Kalaska and Crammond 1992; Soechting and Flanders 1992) indicate that all may be represented neurally, whereas other frameworks, such as optimal feedback control (Scott 2004), suggest that such a hierarchy of representations is not required.

In the present experiment, monkeys were trained to counteract a uniform distribution of torques. If neural processing reflected the range of behavioral requirements inherent in a task, one might expect that each torque direction would have been represented by an equal proportion of M1 neurons. In fact, a previous study observed that the activity of M1 neurons can reflect the range of behavioral requirements (Hepp-Reymond et al. 1999). Specifically, force-related activity of individual neurons was modulated by contextual cues indicating the upcoming range of forces and number of force steps in a grip force task. In the current study, however, we observed a bimodal distribution of PTDs. Furthermore, we did not observe any systematic relationships between PTDs, torque-slopes, and tuning widths, indicating that the relatively small proportion of neurons with PTDs near 45 and 225° did not possess higher torque-slopes or different tuning widths to compensate for the bimodal distribution of PTDs. Taken together, these findings indicate that M1 neurons as a population do not reflect the range of behavioral requirements inherent in our postural task.

Another possibility is that proprioceptive feedback during the task influenced M1 activity. Because many M1 neurons respond strongly to proprioceptive input (Evarts and Tanji 1976; Flament and Hore 1988; Lemon and Porter 1976; Scott and Kalaska 1997; Wolpaw 1980; Wong et al. 1978), one might expect that a greater proportion of neurons would have preferred joint torques that reflected the mechanical action of the shoulder and elbow muscles. Although most shoulder and elbow muscles span only a single joint, the four biarticular muscles combine flexion or extension at both joints. However, these directions in joint-torque space were least represented across our population of M1 neurons.

Notably, our population of torque-related M1 neurons exhibited a bimodal distribution of PTDs that was biased toward whole-limb flexor and whole-limb extensor torques. This mirrored the pattern of torque-related EMG of a sample of proximal arm muscles in the same monkeys in this postural task (Kurtzer et al. 2006a). Other studies also described close relationships between M1 and the motor periphery by comparing patterns of activity across populations of neurons and muscles. Importantly, greater proportions of neurons and muscles are maximally activated when the hand moves toward or away from the body compared with the left or right during planar reaching movements (Kurtzer et al. 2006b; Scott et al. 2001). Biases in their preferred directions of movement (Caminiti et al. 1990; Kakei et al. 1999; Scott and Kalaska 1997) and force (Sergio and Kalaska 2003) undergo similar rotations after changes in limb geometry. Both neurons and muscles exhibit similar temporal changes in their distributions of preferred movement direction (Sergio et al. 2005; Todorov 2000). Finally, both display a coupling between their onset time and magnitude of response (Scott 1997). Each of these findings highlights the intimate relationship between M1 and the motor periphery, but they do not imply a point-to-point mapping in which M1 neurons are merely upper motor neurons. Rather, they suggest that the activities of M1 neurons and peripheral muscles are shaped by similar constraints.

Biased representations of joint torque: muscle synergies or optimal strategies?

A fundamental issue in motor control is that the redundancy of the musculoskeletal system requires the CNS to select patterns of muscle activations from an infinite set of possible solutions to execute even the simplest multijoint behaviors. Bernstein (1967) proposed that this problem could be simplified if a broad range of motor behaviors could be produced with a small number of distinct muscle synergies, sets of muscles that are collectively activated to form functional units. Accordingly, motor commands would be accomplished by activating appropriate combinations of muscle synergies.

A number of subsequent studies provided empirical evidence suggesting that the CNS uses muscles synergies. Notably, patterns of muscle activity were decomposed into a small number of muscle synergies across a broad range of tasks and species including postural control and locomotion in humans (Henry et al. 1998; Ivanenko et al. 2004) and cats (Buford and Smith 1990; Macpherson et al. 1986; Ting and MacPherson 2005), and a variety of natural behaviors in frogs (d'Avella et al. 2003). Synergistic patterns of muscle activity were also observed during reflex responses to cutaneous stimulation in spinalized frogs, suggesting that such patterns are reflected in neural circuitry at the spinal level (Hart and Giszter 2004; Tresch et al. 1999).

The present data could be viewed as providing evidence that muscle synergies are expressed by neural processes at supraspinal levels. We found that in our planar task M1 neurons tended to represent torque-related activity in one of two multijoint muscle synergies: whole-limb flexion and whole-limb extension. Because other studies also observed that M1 activity reflects various muscles synergies (Donoghue et al. 1992; Holdefer and Miller 2002; Park et al. 2004), these observations could be viewed as evidence that muscle synergies may form the basis of a common language to create complex motor patterns.

Alternatively, our results are also consistent with motor patterns arising from optimal control strategies that are unrelated to the notion of muscle synergies (Scott 2004). Similar to earlier studies of the monkey wrist (Hoffman and Strick 1999), human forearm (Buchanan et al. 1989; van Zuylen et al. 1988), and human leg (Nozaki et al. 2004), we observed that the pattern of activations of monoarticular and biarticular shoulder and elbow muscles was systematically rotated away from their anatomical action toward whole-limb flexor or whole-limb extensor torques (Kurtzer et al. 2006a). Mathematical models illustrated that minimizing global measures of muscle activity could predict the observed PTD rotations of each group of upper arm muscles: shoulder flexors and extensors, elbow flexors and extensors, and biarticular flexors and extensors (Kurtzer et al. 2006a). These models suggest that M1 may control certain motor behaviors, such as postural maintenance, by producing patterns of neural activity that achieve the task goal by minimizing muscle factors such as output noise. As a result, muscle synergies may not reflect an explicit simplifying strategy for controlling movement. Rather, they may be the appropriate pattern of muscle activities given the physical properties of the musculoskeletal system and the task goals. This raises the problem of what evidence is required to demonstrate whether muscle synergies are an explicit strategy for simplifying motor control or an emergent pattern of muscle activity suitable for optimizing task goals.

	GRANTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION GRANTS ACKNOWLEDGMENTS REFERENCES

 
The present research was supported by Canadian Institute of Health Research (CIHR) grants and a CIHR Investigator award to S. H. Scott.

	ACKNOWLEDGMENTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION GRANTS ACKNOWLEDGMENTS REFERENCES

 
We thank K. Moore, J. Swaine, and J. Green for technical support and A. Coderre, J. A. Pruszynski, and T. Lillicrap for critical reading of the manuscript.

	FOOTNOTES

 
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

Address for reprint requests and other correspondence: S. H. Scott, Center for Neuroscience Studies, Queen's University, Botterell Hall, Room 232, Kingston, Ontario, Canada K7L 3N6 (E-mail: steve{at}biomed.queensu.ca)

	REFERENCES

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION GRANTS ACKNOWLEDGMENTS REFERENCES

 
Bennett KM, Lemon RN. Corticomotoneuronal contribution to the fractionation of muscle activity during precision grip in the monkey. J Neurophysiol 75: 1826–1842, 1996.[Abstract/Free Full Text]

Bernstein N. The Coordination and Regulation of Movements. Oxford, UK: Pergamon, 1967.

Buchanan TS, Rovai GP, Rymer WZ. Strategies for muscle activation during isometric torque generation at the human elbow. J Neurophysiol 62: 1201–1212, 1989.[Abstract/Free Full Text]

Buford JA, Smith JL. Adaptive control for backward quadrupedal walking. II. Hindlimb muscle synergies. J Neurophysiol 64: 756–766, 1990.[Abstract/Free Full Text]

Burdet E, Osu R, Franklin DW, Milner TE, Kawato M. The central nervous system stabilizes unstable dynamics by learning optimal impedance. Nature 414: 446–449, 2001.[CrossRef][Medline]

Buys EJ, Lemon RN, Mantel GW, Muir RB. Selective facilitation of different hand muscles by single corticospinal neurones in the conscious monkey. J Physiol 381: 529–549, 1986.[Abstract/Free Full Text]

Cabel DW, Cisek P, Scott SH. Neural activity in primary motor cortex related to mechanical loads applied to the shoulder and elbow during a postural task. J Neurophysiol 86: 2102–2108, 2001.[Abstract/Free Full Text]

Caminiti R, Johnson PB, Urbano A. Making arm movements within different parts of space: dynamic aspects in the primate motor cortex. J Neurosci 10: 2039–2058, 1990.[Abstract]

Cheney PD, Fetz EE. Functional classes of primate corticomotoneuronal cells and their relation to active force. J Neurophysiol 44: 773–791, 1980.[Free Full Text]

Cheng EJ, Scott SH. Morphometry of Macaca mulatta forelimb. I. Shoulder and elbow muscles and segment inertial parameters. J Morphol 245: 206–224, 2000.[CrossRef][Web of Science][Medline]

Crutcher MD, Alexander GE. Movement-related neuronal activity selectively coding either direction or muscle pattern in three motor areas of the monkey. J Neurophysiol 64: 151–163, 1990.[Abstract/Free Full Text]

Darainy M, Malfait N, Gribble PL, Towhidkhah F, Ostry DJ. Learning to control arm stiffness under static conditions. J Neurophysiol 92: 3344–3350, 2004.[Abstract/Free Full Text]

d'Avella A, Saltiel P, Bizzi E. Combinations of muscle synergies in the construction of a natural motor behavior. Nat Neurosci 6: 300–308, 2003.[CrossRef][Web of Science][Medline]

Donoghue JP, Leibovic S, Sanes JN. Organization of the forelimb area in squirrel monkey motor cortex: representation of digit, wrist, and elbow muscles. Exp Brain Res 89: 1–19, 1992.[CrossRef][Web of Science][Medline]

Drew T. Motor cortical activity during voluntary gait modifications in the cat. I. Cells related to the forelimbs. J Neurophysiol 70: 179–199, 1993.[Abstract/Free Full Text]

Evarts EV. Relation of pyramidal tract activity to force exerted during voluntary movement. J Neurophysiol 31: 14–27, 1968.[Free Full Text]

Evarts EV. Activity of pyramidal tract neurons during postural fixation. J Neurophysiol 32: 375–385, 1969.[Free Full Text]

Evarts EV, Fromm C, Kroller J, Jennings VA. Motor cortex control of finely graded forces. J Neurophysiol 49: 1199–1215, 1983.[Free Full Text]

Evarts EV, Tanji J. Reflex and intended responses in motor cortex pyramidal tract neurons of monkey. J Neurophysiol 39: 1069–1080, 1976.[Abstract/Free Full Text]

Fetz EE, Cheney PD. Postspike facilitation of forelimb muscle activity by primate corticomotoneuronal cells. J Neurophysiol 44: 751–772, 1980.[Free Full Text]

Flament D, Hore J. Relations of motor cortex neural discharge to kinematics of passive and active elbow movements in the monkey. J Neurophysiol 60: 1268–1284, 1988.[Abstract/Free Full Text]

Franklin DW, Milner TE. Adaptive control of stiffness to stabilize hand position with large loads. Exp Brain Res 152: 211–220, 2003.[CrossRef][Web of Science][Medline]

Fromm C. Changes of steady state activity in motor cortex consistent with the length-tension relation of muscle. Pfluegers Arch 398: 318–323, 1983.[CrossRef][Web of Science][Medline]

Georgopoulos AP, Ashe J, Smyrnis N, Taira M. The motor cortex and the coding of force. Science 256: 1692–1695, 1992.[Abstract/Free Full Text]

Georgopoulos AP, Kalaska JF, Caminiti R, Massey JT. On the relations between the direction of two-dimensional arm movements and cell discharge in primate motor cortex. J Neurosci 2: 1527–1537, 1982.[Abstract]

Gomi H, Osu R. Task-dependent viscoelasticity of human multijoint arm and its spatial characteristics for interaction with environments. J Neurosci 18: 8965–8978, 1998.[Abstract/Free Full Text]

Gould HJ, Cusick CG, Pons TP, Kaas JH. The relationship of corpus callosum connections to electrical stimulation maps of motor, supplementary motor, and the frontal eye fields in owl monkeys. J Comp Neurol 247: 297–325, 1986.[CrossRef][Web of Science][Medline]

Graham KM, Moore KD, Cabel DW, Gribble PL, Cisek P, Scott SH. Kinematics and kinetics of multijoint reaching in nonhuman primates. J Neurophysiol 89: 2667–2677, 2003.[Abstract/Free Full Text]

Graham KM, Scott SH. Morphometry of Macaca mulatta forelimb. III. Moment arm of shoulder and elbow muscles. J Morphol 255: 301–314, 2003.[CrossRef][Web of Science][Medline]

Gribble PL, Scott SH. Method for assessing directional characteristics of non-uniformly sampled neural activity. J Neurosci Methods 113: 187–197, 2002.[CrossRef][Web of Science][Medline]

Hart CB, Giszter SF. Modular premotor drives and unit bursts as primitives for frog motor behaviors. J Neurosci 24: 5269–5282, 2004.[Abstract/Free Full Text]

Henry SM, Fung J, Horak FB. EMG responses to maintain stance during multidirectional surface translations. J Neurophysiol 80: 1939–1950, 1998.[Abstract/Free Full Text]

Hepp-Reymond M. Functional organization of motor cortex and its participation in voluntary movements. In: Comparative Primate Biology: Neurosciences, edited by Steklis HD, Erwin J. New York: A. R. Liss, 1988, vol. 4, p. 501–624.

Hepp-Reymond M, Kirkpatrick-Tanner M, Gabernet L, Qi HX, Weber B. Context-dependent force coding in motor and premotor cortical areas. Exp Brain Res 128: 123–133, 1999.[CrossRef][Web of Science][Medline]

Hoffman DS, Strick PL. Step-tracking movements of the wrist. IV. Muscle activity associated with movements in different directions. J Neurophysiol 81: 319–333, 1999.[Abstract/Free Full Text]

Holdefer RN, Miller LE. Primary motor cortical neurons encode functional muscle synergies. Exp Brain Res 146: 233–243, 2002.[CrossRef][Web of Science][Medline]

Hollerbach JM. Computers, brains and the control of movement. Trends Neurosci 5: 189–192, 1982.[CrossRef][Web of Science]

Humphrey DR, Reed DJ. Separate cortical systems for control of joint movement and joint stiffness: reciprocal activation and coactivation of antagonist muscles. Adv Neurol 39: 347–372, 1983.[Medline]

Ivanenko YP, Poppele RE, Lacquaniti F. Five basic muscle activation patterns account for muscle activity during human locomotion. J Physiol 556: 267–282, 2004.[Abstract/Free Full Text]

Kakei S, Hoffman DS, Strick PL. Muscle and movement representations in the primary motor cortex. Science 285: 2136–2139, 1999.[Abstract/Free Full Text]

Kalaska JF, Cohen DA, Hyde ML, Prud'homme M. A comparison of movement direction-related versus load direction-related activity in primate motor cortex, using a two-dimensional reaching task. J Neurosci 9: 2080–2102, 1989.[Abstract]

Kalaska JF, Crammond DJ. Cerebral cortical mechanisms of reaching movements. Science 255: 1517–1523, 1992.[Abstract/Free Full Text]

Kurtzer I, Herter TM, Scott SH. Random change in cortical load representation suggests distinct control of posture and movement. Nat Neurosci 8: 498–504, 2005.[Web of Science][Medline]

Kurtzer I, Herter TM, Scott SH. Non-uniform distribution of reach-related and torque-related activity in upper arm muscles and neurons of primary motor cortex. J Neurophysiol 96: 3220–3230, 2006b.[Abstract/Free Full Text]

Kurtzer I, Pruszynski JA, Herter TM, Scott SH. Primate upper limb muscles exhibit activity patterns that differ from their anatomical action during a postural task. J Neurophysiol 95: 493–504, 2006a.[Abstract/Free Full Text]

Kwan HC, MacKay WA, Murphy JT, Wong YC. Spatial organization of precentral cortex in awake primates. II. Motor outputs. J Neurophysiol 41: 1120–1131, 1978.[Free Full Text]

Lemon RN, Porter R. Afferent input to movement-related precentral neurones in conscious monkeys. Proc R Soc Lond B Biol Sci 194: 313–339, 1976.[Medline]

Macpherson JM, Rushmer DS, Dunbar DC. Postural responses in the cat to unexpected rotations of the supporting surface: evidence for a centrally generated synergic organization. Exp Brain Res 62: 152–160, 1986.[Web of Science][Medline]

McIntyre J, Mussa-Ivaldi FA, Bizzi E. The control of stable postures in the multijoint arm. Exp Brain Res 110: 248–264, 1996.[Web of Science][Medline]

McKiernan BJ, Marcario JK, Karrer JH, Cheney PD. Corticomotoneuronal postspike effects in shoulder, elbow, wrist, digit, and intrinsic hand muscles during a reach and prehension task. J Neurophysiol 80: 1961–1980, 1998.[Abstract/Free Full Text]

Milner TE, Franklin DW. Impedance control and internal model use during the initial stage of adaptation to novel dynamics in humans. J Physiol 567: 651–664, 2005.[Abstract/Free Full Text]

Moran DW, Schwartz AB. Motor cortical representation of speed and direction during reaching. J Neurophysiol 82: 2676–2692, 1999.[Abstract/Free Full Text]

Nozaki D, Nakazawa K, Akai M. Muscle activity determined by cosine tuning with a nontrivial preferred direction during isometric force exertion by lower limb. J Neurophysiol 93: 2614–2624, 2005.[Abstract/Free Full Text]

Park MC, Belhaj-Saif A, Cheney PD. Properties of primary motor cortex output to forelimb muscles in rhesus macaques. J Neurophysiol 92: 2968–2984, 2004.[Abstract/Free Full Text]

Park MC, Belhaj-Saif A, Gordon M, Cheney PD. Consistent features in the forelimb representation of primary motor cortex in rhesus macaques. J Neurosci 21: 2784–2792, 2001.[Abstract/Free Full Text]

Penfield W, Rasmussen T. The Cerebral Cortex of Man. New York: Macmillan, 1950.

Poliakov AV, Schieber MH. Limited functional grouping of neurons in the motor cortex hand area during individuated finger movements: a cluster analysis. J Neurophysiol 82: 3488–3505, 1999.[Abstract/Free Full Text]

Porter R, Lemon RN. Cortical Function and Voluntary Movement. New York: Oxford Univ. Press, 1993.

Sanes JN, Schieber MH. Orderly somatotopy in primary motor cortex: does it exist? NeuroImage 13: 968–974, 2001.[Web of Science][Medline]

Schieber MH. Muscular production of individuated finger movements: the roles of extrinsic finger muscles. J Neurosci 15: 284–297, 1995.[Abstract]

Schieber MH. Constraints on somatotopic organization in the primary motor cortex. J Neurophysiol 86: 2125–2143, 2001.[Abstract/Free Full Text]

Schieber MH, Hibbard LS. How somatotopic is the motor cortex hand area? Science 261: 489–492, 1993.[Abstract/Free Full Text]

Scott SH. Comparison of onset time and magnitude of activity for proximal arm muscles and motor cortical cells prior to reaching movements. J Neurophysiol 77: 1016–1022, 1997.[Abstract/Free Full Text]

Scott SH. Apparatus for measuring and perturbing shoulder and elbow joint positions and torques during reaching. J Neurosci Methods 89: 119–127, 1999.[CrossRef][Web of Science][Medline]

Scott SH. Role of motor cortex in coordinating multi-joint movements: is it time for a new paradigm? Can J Physiol Pharmacol 78: 923–933, 2000.[CrossRef][Web of Science][Medline]

Scott SH. Optimal feedback control and the neural basis of volitional motor control. Nat Rev Neurosci 5: 532–546, 2004.[Medline]

Scott SH, Gribble PL, Graham KM, Cabel DW. Dissociation between hand motion and population vectors from neural activity in motor cortex. Nature 413: 161–165, 2001.[CrossRef][Medline]

Scott SH, Kalaska JF. Reaching movements with similar hand paths but different arm orientations. I. Activity of individual cells in motor cortex. J Neurophysiol 77: 826–852, 1997.[Abstract/Free Full Text]

Sergio LE, Hamel-Paquet C, Kalaska JF. Motor cortex neural correlates of output kinematics and kinetics during isometric-force and arm-reaching tasks. J Neurophysiol 94: 2353–2378, 2005.[Abstract/Free Full Text]

Sergio LE, Kalaska JF. Systematic changes in motor cortex cell activity with arm posture during directional isometric force generation. J Neurophysiol 89: 212–228, 2003.[Abstract/Free Full Text]

Smith AM, Hepp-Reymond M-C, Wyss UR. Relation of activity in of precentral cortical neurons to force and rate of force change during isometric contractions of finger muscles. Exp Brain Res 23: 315–332, 1975.[Web of Science][Medline]

Soechting JF, Flanders M. Moving in three-dimensional space: frames of reference, vectors, and coordinate systems. Annu Rev Neurosci 15: 167–191, 1992.[CrossRef][Web of Science][Medline]

Thach WT. Correlation of neural discharge with pattern and force of muscular activity, joint position, and direction of intended next movement in motor cortex and cerebellum. J Neurophysiol 41: 654–676, 1978.[Abstract/Free Full Text]

Ting LH, Macpherson JM. A limited set of muscle synergies for force control during a postural task. J Neurophysiol 93: 609–613, 2005.[Abstract/Free Full Text]

Todorov E. Direct cortical control of muscle activation in voluntary arm movements: a model. Nat Neurosci 3: 391–398, 2000.[CrossRef][Web of Science][Medline]

Tresch MC, Saltiel P, Bizzi E. The construction of movement by the spinal cord. Nat Neurosci 2: 162–167, 1999.[CrossRef][Web of Science][Medline]

van Zuylen EJ, Gielen CC, Denier van der Gon JJ. Coordination and inhomogeneous activation of human arm muscles during isometric torques. J Neurophysiol 60: 1523–1548, 1988.[Abstract/Free Full Text]

Werner W, Bauswein E, Fromm C. Static firing rates of premotor and primary motor cortical neurons associated with torque and joint position. Exp Brain Res 86: 293–302, 1991.[Web of Science][Medline]

Wolpaw JR. Correlations between task-related activity and responses to perturbation in primate sensorimotor cortex. J Neurophysiol 44: 1122–1138, 1980.[Abstract/Free Full Text]

Wong YC, Kwan HC, MacKy WA, Murphy JT. Spatial organization of precentral cortex in awake primates. I. Somatosensory inputs. J Neurophysiol 41: 1107–1119, 1978.[Free Full Text]

Woolsey CN, Settlage PH, Meyer DR, Sencer W, Pinto Hamuy T, Travis AM. Patterns of localization in precentral and "supplementary" motor areas and their relation to the concept of a premotor area. Res Publ Assoc Res Nerv Ment Dis 30: 238–364, 1952.[Web of Science][Medline]

This article has been cited by other articles:

				T. M. Herter, T. Korbel, and S. H. Scott Comparison of Neural Responses in Primary Motor Cortex to Transient and Continuous Loads During Posture J Neurophysiol, January 1, 2009; 101(1): 150 - 163. [Abstract] [Full Text] [PDF]

				S. H. Scott Inconvenient Truths about neural processing in primary motor cortex J. Physiol., March 1, 2008; 586(5): 1217 - 1224. [Abstract] [Full Text] [PDF]

This Article Abstract Full Text (PDF) All Versions of this Article: 97/4/2887    most recent 00757.2006v1 Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in ISI Web of Science Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via ISI Web of Science (5) Citing Articles via Google Scholar Google Scholar Articles by Herter, T. M. Articles by Scott, S. H. Search for Related Content PubMed PubMed Citation Articles by Herter, T. M. Articles by Scott, S. H.