Composition and decomposition of visuomotor maps during manual tracking

Abstract

Adapting hand movements to changes in our body or the environment is essential for skilled motor behavior, as is the ability to flexibly combine experience gathered in separate contexts. However it has been shown that when adapting hand movements to two different visuomotor perturbations in succession, interference effects can occur. Here we investigate whether these interference effects compromise our ability to adapt to the superposition of the two perturbations. Participants tracked with a joystick a visual target that followed a smooth but unpredictable trajectory. Four separate groups of participants (total n = 83) completed one block of 50 trials under each of three mappings: one in which the cursor was rotated by 90° (ROTATION), one in which the cursor mimicked the behavior of a mass-spring system (SPRING), and one in which the SPRING and ROTATION mappings were superimposed (SPROT). The order of the blocks differed across groups. Although interference effects were found when switching between SPRING and ROTATION, participants who performed these blocks first performed better in SPROT than participants who had no prior experience with SPRING and ROTATION (i.e., composition). Moreover, participants who started with SPROT exhibited better performance under SPRING and ROTATION than participants who had no prior experience with each of these mappings (i.e., decomposition). Additional analyses confirmed that these effects resulted from components of learning that were specific to the rotational and spring perturbations. These results show that interference effects do not preclude the ability to compose/decompose various forms of visuomotor adaptation.

Asymmetrical transfer of adaptation between reaching and tracking: implications for feedforward and feedback processes

“…We averaged this error across time for each trial while excluding the first second of data from each trial to minimize transients due to initial cursor positioning [30][31][32]. To assess the presence of a possible bias in hand movement direction, we rotated the cursor data (while also excluding the first sec of data) around the center and determined which angle allowed to minimize the Euclidian distance [32]. This angle was called directional error.…”