Task

The task was to grasp the object by the handle at the base of the rod and rotate it back and forth between two angular targets (Fig. 6.1A). Subjects were instructed to generate pure rotational movement about the object’s handle while keeping the position of the handle stationary at the centre of the home region (a visually presented 1 cm radius disc). The trials consisted of alternating clockwise (CW) and counterclockwise (CCW) rotations between the targets (two 40°-apart oriented bars emanating from the home region; Fig. 6.1B). Each trial began with the handle of the object stationary within the home region and the rod of the object aligned with either the CCW target or the CW target. Movement initiation was cued by a tone and the appearance of the second target, towards which subjects rotated the object.

The trial ended when the orientation of the object was aligned with the orientation of the aimed target. Subjects were required to finish the movement within 400 ms, otherwise they were warned with a ‘two slow’ message. Movements exceeding 500 ms had to be repeated.

Rest breaks (60 s) were provided during the experiments every 3–5 minutes.

The dynamics of the object was simulated as a point mass (1% of the subject’s body mass) at the end of the rigid rod (Fig. 6.1B). Rotating the object generated forces and torques at the handle which were simulated by the manipulandum. The torque was associated with the moment of inertia of the object, and the generated force was associated with the circular motion of the mass, which could be decomposed into two components: tangential and centripetal (F in Fig. 6.1B). The direction of force depended on the orientation of the object, while its magnitude was determined by the mass and the length of the object. Critically, the

128 Adaptation to familiar dynamics

Error clamp (C) Exposure (E) Zero force (Z)

1 cm

Fig. 6.1 A. Experimental setup. Subjects grasped the handle of the WristBot and rotated a virtual hammer-like object back and forth between two angular targets. B. The object consisted of a mass attached to a stick. Subjects were asked to rotate the object around its axis (grasp point) while trying to prevent the translational movement of the handle. The targets were two rectangle bars separated 40^◦. The dynamics of the object consisted of a torque (τ) that resisted the rotation, as well as translational forces (centripetal F_Cand tangential F_T) that would perturb the handle of the object. Subjects had to learn to compensate for the torque to rotate the object withing the accepted speed, and the perturbing forces to keep the handle stationary during the rotations. C. Three types of trials were used during the experiment. In exposure trials (left) subjects experienced both the translational forces and the torque. In null trials (middle) the forces were switched off and only the torque was applied. And, in error-clamp trials the handle of the object was fixed in the home position a using simulated spring that prevented translational movement. Subjects also experienced the inertial torque during error-clamp trials.

6.2 Methods 129

force caused the handle of the object to displace unless subjects produced a compensatory force in the opposite direction.

During the experiments, we manipulated the dynamics of the object to be one of three possible trial types: exposure trial, null trial or error-clamp trial (Fig. 6.1C). On exposure trials, subjects experienced the full dynamics of the object including the force on the handle as well as the torque. In these trials, subjects had to learn to compensate for the transla-tional forces in order to keep the handle stationary during the rotation. On null trials, the manipulandum did not produce any forces and the handle was free to move. Importantly, any forces produced by subjects on null trials would cause the handle to displace. Finally, on error-clamp trials, the manipulandum simulated a stiff two-dimensional spring, centred on the handle position at the start of the trial (the spring constant was 40 N/cm). Error-clamp trials effectively eliminated kinematic errors and prevented error-driven adaptation. They also allowed the compensatory forces produced by subjects to be measured (Ingram et al., 2011).

6.2.2 Analysis

Two performance measures were used for the analysis. On exposure and null trials we measured the position of the handle (at 1000 Hz) during the rotation, and took the peak displacement (PD) of the handle on that trial with respect to the starting position. The PD was considered as the kinematic error of movement (the task was to keep the handle stationary), and thus the PD of zero meant ideal performance. Also, on each error-clamp trial, we recorded the time course of the force magnitude generated by subjects on that trial (at 1000 Hz) in compensation for the object dynamics. We also obtained the ideal force trajectory that could fully compensate for the object dynamics based on the angular velocity and acceleration of movement on that trial (that is, Fig. 6.1B). The ideal and measured force trajectories were trimmed according to the angular velocity of the object, from when the velocity exceeded 5% of its maximum value, onwards until when the velocity reduced to less than 5% of its maximum value. We then regressed the measured force onto the ideal force (without intercept) and took the regression coefficient as the adaptation index.

130 Adaptation to familiar dynamics

Overall, for each error-clamp trial we had a single measure of adaptation, and for each null or exposure trial we had a single measure of PD. We excluded from our analysis the data from the first 10 trials that followed each rest break during the experiment.