Human balance control is a complex sensorimotor process in which the central nervous system (CNS) uses information from various sensory systems (e.g., visual, vestibular, and somatosensory) to detect perturbations to stance, select postural strategies with which to make postural corrections, and trigger appropriate motor responses. Two models of this process have been proposed:​ modality-specific strategy selection, in which information from a specific sensory modality would be used by the CNS to select and trigger a specific postural strategy;​ and sensory-integrated strategy selection, in which information from all sensory modalities is used collectively to determine the most appropriate postural strategy.

Many perturbations to stance elicit either the ankle strategy or the hip strategy. The ankle strategy corrects posture by controlling the body as a single-segment inverted pendulum, while the hip strategy corrects posture by controlling the body as a double-segment inverted pendulum. While the ankle strategy is used to respond to many small perturbations, normal subjects employ hip strategy during more challenging postural situations, such as while standing across a narrow beam, standing on one foot, standing tandem, standing on a compliant surface, or standing with lower leg ischemia. Patients with bilateral vestibular loss fail to elicit hip strategy responses and thus have difficulty maintaining balance without stepping during such tasks.

The absence of a hip strategy in individuals lacking vestibular cues has been purported as evidence for modality-specific strategy selection. However, each of the postural tasks which has been used to study hip strategy alters one or more of the following system characteristics:​ the dynamics, the initial state (positions and velocities of body segments), the biomechanical limits of stability, or the availability of lower leg sensory input. Each of these adjustments to the system presumably prescribes corresponding alterations of the sensorimotor control laws governing normal balance;​ thus, the absence of a hip strategy in vestibular patients could also be due to a failure to properly modify any or all of these control laws.

The objectives of this thesis were thus twofold:​ first, to identify a postural task which would elicit a hip strategy in normal individuals without altering system characteristics;​ and second, to subject vestibular patients to this postural task and examine their responses for hip strategy.

Fast, rearward, support surface translations, applied during normal, erect stance, were found to elicit postural responses which exhibited early abdominal muscle bursts and large hip rotations in normal subjects, two features typically associated with a hip strategy. However, joint torque analysis was required to determine whether or not the hip rotations were initiated with active hip torques (i.e., hip strategy). Active hip torques were implicit to the hip movements observed in many of the aforementioned studies of hip strategy, because the tasks employed in these studies imposed biomechanical constraints which rendered ankle torque ineffective at controlling body posture. However, fast, rearward, support surface translations imposed no such constraints, and thus did not rule out the possibility that the onset of the hip rotations was nothing more than a product of dynamic interaction.

Because net joint torques computed with traditional inverse dynamics techniques failed to produce forward dynamic simulations of the observed movements, linear optimal control techniques were employed. Specifically, the Linear Quadratic Follower method computed net joint torques from the kinesiological data and produced stable simulations of the observed postural responses.

Joint torque analysis revealed a subject-independent, triphasic pattern of early hip torque in the responses of normal subjects to the fast, rearward support surface translations, confirming the triggering of a hip strategy. Patients who lost vestibular function as adults responded with similar torques. In contrast, patients who lost vestibular function as infants did not respond with the hip strategy, but rather with other effective strategies.

We conclude that vestibular cues are not necessary to select and trigger a hip strategy. However, vestibular function may be critical during development to establish the normal, early hip torque patterns. We postulate, therefore, that the CNS employs sensory-integrated strategy selection, and that the absence of a hip strategy in vestibular patients during other tasks may be due to inadequate sensorimotor adaptation to the alterations of the system characteristics (i.e., the initial state, the dynamics, the biomechanical limits of stability, or the availability of lower leg sensory input).