Procedure. Observability and controllability of nonrigid movements is necessitated by task constraints on such movement, and it should be revealed in an interaction between task and body segment with respect to various movement parameters. This hypothesis was tested by comparing patterns of movement at various body locations (analogous to modal analysis in mechanical engineering) during a variety of tasks. The participant stood on a moveable platform while (a) doing nothing, (b) tapping lightly at constant intervals on a keyboard, or (c) reading text that was at a fixed location and not attached to the body or the moveable platform. Performance on the tapping and reading tasks was not measured. Accelerometers were attached to the head, hip, ankle, and the platform. The accelerometers were sensitive to tilt and acceleration in the sagittal plane (e.g., anterior-posterior "sway") and, thus, they provided an estimate of the variation in gravitoinertial force on the corresponding body segments. The variation in postural configuration (accelerator outputs) was sampled lOO times per s. These postural movement patterns were measured in trials that were 2.5 s in duration (250 samples per trial). The platform moved 5 cm forward or backward atconstantvelocity for 0.4 s (from 0.5 to 0.9 s into each trial). There were twelve trials for each combination of task (nothing, tapping, or reading) and direction
of platform movement (forward or backward). Order of the six conditions was randomized.