A) Dual-task setup during pulsed movements. Hand position in the y-direction (toward the target) triggered perturbation force in the x-direction and/or the onset of the word presentation. The force perturbation was centered at y = 5 cm. When the hand arrived at y = 5 cm, either a concrete (blue), abstract (red), or ambiguous (green) word was presented. The word was 500 ms in duration. At the end of movement, subjects received feedback on the correctness of the movement, but no feedback on the categorization was provided. B) Average, across all dual-task conditions, prepulse movement trajectory. C) Average pulse movement trajectories for leftward (solid) and rightward (dashed) force perturbations. D) Average postpulse movement trajectories, minus the average prepulse movement trajectory, following leftward (solid) and rightward (dashed) force perturbation.

Perpendicular displacements at 3 (dotted), 5 (dashed), and 7 cm (solid) in the pulsed (A) and postpulse movements (B) versus subject categorization RT for concrete (blue), abstract (red), and ambiguous (green) word categories in the pulsed movement. A) Word categorization RT was scaled by the semantic category; subjects responded progressively faster depending on the concreteness of the word. Neither semantic category nor RT affected within movement feedback control on pulsed movements. B) Postpulse adaptation was scaled by the word categorization RT. Subjects showed significantly more adaptation following pulsed movements with concrete words than pulsed movements with ambiguous words. Error bars represent the 95% confidence interval of the mean.

Offline, subjects categorized words presented during the movement task on a scale of 1 to 5, with 1 being most concrete, and 5 being most abstract. Ratings were separated into 10 equally sized bins and averaged across subjects. Colorbar (side) represents the linear spectrum of subjects' offline word survey of concreteness from most concrete (blue) to most abstract (red). A) Online word categorization RT was the fastest for words that were best categorized into concrete or abstract words offline. Subjects were slowest for ambiguous words, which were words that subjects classified as not falling directly into concrete or abstract categories during the offline survey. B) Subjects adapted the most to words that fell best into concrete or abstract categories, while words that were ambiguous had the least adaptation.

The congruency index measured the similarity of word categorization from online to offline categorization across subjects. For each subject, words that maintained categorization were assigned a 1 and words that were switched were given a zero. Averaged across subjects, words were grouped into 10 equally sized bins dependent on congruency index. Colorbar (side) represents the linear spectrum of subjects' offline word survey of concreteness from most concrete (blue) to most abstract (red). A) The greater the congruency index (maintained categorizations), the faster the RT. These congruent words were words that best fell into either concrete or abstract categories during the offline word survey. B) The postpulse adaptation correlates with the congruency index. Subjects adapted more on congruent words (index closer to 1), than on incongruent words (closer to 0). C) Average movement trajectory preceding pulsed movements with either low congruency indices (gray), less than the mean congruency index, and movements with high congruency indices (black), greater than the mean congruency index. D) Average pulse movement trajectory for low (gray) and high (black) congruency index words for leftward (solid) and rightward (dashed) force perturbations. E) Average postpulse movement trajectory, minus average prepulse movement trajectory, following leftward (solid) and rightward (dashed) pulsed movements with low (gray) and high (black) congruency index words.

During the pulsed movement, movement errors engage the feedback control system to correct the movement online (solid route) and the errors engage cognitive control systems (dashed route). The word categorization task also engages cognitive control differentially depending on the category ambiguity of the word (gray route). Feedback error learning systems update the predictive feedforward control on the next movement (solid route). Cognitive control also influences predictive control on the next movement (dashed route), but is degraded depending on the category ambiguity of the presented word.