De novo learning versus adaptation of continuous control in a manual tracking task

Abstract

How do people learn to perform tasks that require continuous adjustments of motor output, like riding a bicycle? People rely heavily on cognitive strategies when learning discrete movement tasks, but such time-consuming strategies are infeasible in continuous control tasks that demand rapid responses to ongoing sensory feedback. To understand how people can learn to perform such tasks without the benefit of cognitive strategies, we imposed a rotation/mirror reversal of visual feedback while participants performed a continuous tracking task. We analyzed behavior using a system identification approach which revealed two qualitatively different components of learning: adaptation of a baseline controller and formation of a new, task-specific continuous controller. These components exhibited different signatures in the frequency domain and were differentially engaged under the rotation/mirror reversal. Our results demonstrate that people can rapidly build a new continuous controller de novo and can simultaneously deploy this process with adaptation of an existing controller.

Long-Term Motor Learning in the “Wild” With High Volume Video Game Data

“…Specifically, the median hits per second was 2.72 on day 1, whereas it was 3.78 on day 60 (t = −21.27, df = 83.95, p-value < 1e-6). Improvements in hits per second were non-linear (R 2 non-linear: 0.21, R 2 linear: 0.26), showing a trend similar to in-lab studies studying motor skill acquisition (Telgen et al, 2014;Yang et al, 2021) and validating our online approach.…”

Control becomes habitual early on when learning a novel motor skill

“…Given that conventional analysis methods always yield a positive gain to describe frequency-domain data, we used the method described in [37] to compute a signed gain, g , relating cursor and target movements. This was computed as the dot product between transfer functions: where a is the transfer function for a given block of interest and is the transfer function at baseline with unit length.…”

“…With practice, participants learned to reduce the positional error between the target and cursor (Fig 3B). We examined participants' tracking performance at different frequencies of movement using a system identification approach [33][34][35][36][37], which allowed us to separately examine the behavioral responses to target movements at different frequencies even though they occurred concurrently in the task. Specifically, we computed the gain and direction of cursor movement relative to target movement at each frequency, which can be interpreted analogously to the reach direction analysis for the shows the gain and direction of cursor movements in response to target movement at a particular frequency.…”

“…To examine this, we compared how strongly participants exhibited habitual behavior between the two tasks. First, we averaged the gains of each participant's tracking behavior at the highest three frequencies, given that we expected habitual behavior to be strongest at these frequencies [37]. We then correlated each subject's average gain with the proportion of habitual reaches they made in the point-to-point task, as in All groups' habits were similarly resistant to extinction An alternative way in which behavior may become habitual is by becoming more persistent, i.e., resistant to extinction.…”

Using Artificial Intelligence for Assistance Systems to Bring Motor Learning Principles into Real World Motor Tasks

“…This approach to investigate the effect of the task parameters on motor adaptation is ongoing and should be continued for all possible task parameters: error size [ 250 , 252 , 253 ], error consistency [ 254 , 255 ], feedback timing [ 256 , 257 ], dimensionality [ 239 ], inter-trial time, movement speed, degrees of freedom of motion [ 258 ], reaction time [ 259 ] or continuous vs. discrete control [ 260 ].…”