Repeatable spatial maps of a few force and joint torque patterns elicited by microstimulation applied throughout the lumbar spinal cord of the spinal frog

Abstract

Motor learning and construction of novel behavior must be constrained by the spinal motor apparatus and how it supports movement. Recent work supports a modularity of spinal cord function in both mammals and lower vertebrates. We sought to extend these analyses by a complete mapping of the lumbar enlargement not previously attempted. The frog lumbar spinal cord grey matter motor responses were systematically mapped using microstimulation at a fine grain of 200 μm separation mediolaterally and in depth throughout the enlargement. The patterns of force magnitude and direction were noted. Large areas of spinal cord produced very small forces. Some areas produced strong responses. Both the strong and the weak force responses fell into a few classes. In all frogs examined the forces elicited fell into a few (5) classes of directions. These forces were expressed as joint torques by standard means. Forces were observed to comprise both pure hip torques and combined knee/hip torque patterns, but no pure knee torques. The contiguous regions producing these force directions at high magnitudes were arranged in repeating patterns. The directions of the forces elicited were strongly correlated among frogs in specific regions of spinal cord while other regions showed individual variations. The data are consistent with microstimulation recruitment of specific motor responses or force-field primitives in some spinal cord regions in the frog. Similar constraints may exist early in mammalian motor development or after spinal cord injury.

Introduction

Previous results obtained from microstimulation of frog spinal cords showed that activation of intermediate areas of spinal cord grey matter produced a few patterns of forces (Bizzi, Mussa-Ivaldi, & Giszter, 1991; Giszter, Mussa-Ivaldi, & Bizzi, 1993; Loeb, Giszter, Borghesani, & Bizzi, 1993; Mussa-Ivaldi & Bizzi, 1994). This observation suggested that there may exist a few primitives or modules in the spinal cord that are used for controlling force and movement in reflex and perhaps in voluntary movements. The attractive features of this idea are several fold. First, the use of functional or anatomical modules to coalesce joint and muscular degrees of freedom into useful assemblies directly addresses Bernstein's formulation of the degrees of freedom problem (Bernstein (1967) and Windhorst (1991); Nichols (1994) and Loeb (2000) for an alternate view). Second, the possibility of the assembly of these modules, primitives or building blocks into new behaviors provides the CNS a means to simply develop new tasks (Mussa-Ivaldi, 1992; Mussa-Ivaldi & Giszter, 1992; Mussa-Ivaldi, 1997). Developmentally the intrinsic primitives may `bootstrap' novel motor learning. The advantage of primitives is that learning begins from a collection of developmentally and evolutionarily selected assemblies of proven efficacy which will seed motor searches in useful areas of the high dimensional search space in a way not possible in tabula rasa mechanisms (Gandolfo, Mussa-Ivaldi, & Bizzi, 1996; Zaal, Daigle, Gottlieb, & Thelen, 1999). Third, modular organizations may provide stability guarantees that constrain motor search to safe regions of actuation space and control (Cannon & Slotine, 1995; Lohmiller & Slotine, 1998). Fourth, current control ideas are very clear on the need for modularity, encapsulation and stability of component subunits, modules, primitives, agents or holons in flexible reconfigurable many degrees of freedom processes (Mataric, Williamson, Demiris, & Molan, 1998; Kiehn, Hounsgard, & Sillar, 1997). If an organizational principle is useful, it seems to us it is likely that biological evolution will have discovered it or something superior.

Moving beyond microstimulation, extensive examination of reflex behaviors has shown that the analysis of these as collections of flexibly assembled primitives may offer insights and parsimonious accounts of trajectory construction and control (Kargo and Giszter, 2000a, Kargo and Giszter, 2000b; Giszter & Kargo, 2000; Tresch, Saltiel, & Bizzi, 1999). However, the detailed analysis of microstimulation responses of spinal cord may give insight into whether anatomical modularity supports the functional modularity that is becoming established. This direction is also medically important, since intraspinal FES may soon be developed and effective dynamical controls will be critical (Giszter, Grill, Lemay, Mushahwar, & Prochazka, 2000).

The previous studies using microstimulation concentrated on the spatial organization of isometric forces throughout the limb's workspace and the rules of combination for these force-fields when elicited by multiple stimulation. Fields were generally convergent and were found to be scaled in magnitude with increasing stimulation strength or duration. Further, these force-field types could usually (∼80% of trials) be combined by simple vector superposition. More recently Kargo and Giszter (2000a) showed online adjustments of aimed reflex trajectories could be expressed as force-field primitive sums.

Early work suggested that these primitives might be located in specific regions in the spinal cord (Bizzi et al., 1991). More recent testing using microiontophoresis of NMDA also showed similar results (Saltiel, Tresch, & Bizzi, 1998).These earlier maps rested on fairly sparse random samples of strong force field responses combined across several frogs. Moreover, depth of the stimulating electrode and the variations in the thresholds and sensitivity to electrical stimulation were largely neglected in the first microstimulation studies. Systematic detailed maps of individual frogs, and their comparisons have not been presented.

The goal of the present study was to validate and extend the earlier work by using a systematic measurement grid at a fine grain in individual frogs and comparing among these. The data to be described here examine the spatial variations of force production and muscle activity across the entire lumbar spinal cord in six frogs. Different regions of spinal cord are likely to have differing processing roles and inputs. We used the maximum spatial resolution compatible with excluding the activation of similar neuron cell bodies at different electrode locations using the chosen stimulation parameters. We summarized our data as three dimensional maps of the lumbar cord. In order to generate these maps we focussed on measurement of force samples at a fixed limb configuration. We chose a configuration at which the several two dimensional force-fields characterized in earlier work could be distinguished using a single measurement. The force vectors measured at this point in each of the different force fields identified previously differed from one another by more than 10° in orientation. Thus a single measurement indicated which field of the set was represented at that location, but plainly did not fully characterize the field structure. This restriction of data collection to single force vectors rather than complete fields was necessary to make the extensive mapping experiments feasible. Full force-field collection at the density of sampling and current variations used here would not have been possible due to the data explosion of at least an order of magnitude which is required for two dimensional field description. This paper expands on the prior results, by examining three-dimensional forces rather than two-dimensional, including a detailed study of three-dimensional structure of anatomical regions of strong force production due to stimulation within spinal cord and comparing the repeatability and similarity of the elicited cord responses among frogs.

The data collected here show that only a few force directions are represented in the microstimulation data at the limb configuration chosen throughout the entire spinal cord. The areas of high force production are quite localized. Further, the data also demonstrate that the spinal cord spatial distributions of force directions elicited by microstimulation are, as has been assumed in our other work, similar among frogs in many regions of spinal cord and can be related to the segmental and dorso-ventral structure of spinal cord. These suggest that some anatomical modularity may be present, and support novel medical devices which are indwelling in the isolated spinal cord of para or tetraplegic patients (Giszter et al., 2000; Mushahwar, 1996; Lemay & Grill, 1999).