Table 1.
Fitted parameter values for the 6 subjects.
Fig 1.
Performance measures across subjects.
Temporal evolution of (a) reaching error, and (b) straightness of trajectory (both averaged over a moving window of 10 trials) for subjects (red) and the respective fitted HML model (blue) across trials.
Fig 2.
Cursor trajectory data from the fitted model (a), (c) and human experiments (b), (d). As learning progresses through the 8 sessions, the trajectories become closer to a straight line between targets, which the proposed HML model also captures.
Fig 3.
Evolution of forward model error (FME) for the fitted model as a function of trials for all 6 subjects.
Fig 4.
Comparing HML model with Ref [10] model.
Comparing the errors in RE curve fitting from the model in Ref [10] to the HML model shows that the model in Ref [10] is not as accurate as HML model in capturing the RE for this motor learning task.
Fig 5.
Distribution of driving and exploratory effort (averaged across 128 Monte Carlo runs) with means and 95% confidence bounds across trials as η is varied around its fitted value 3.1742. While driving effort increases, exploratory effort decreases initially, and both plateau past the fitted η value. One-tailed paired t-tests over the effort values across trials reveal this plateauing effect at a significance level of p < 0.001.
Fig 6.
Speed and accuracy variation with kP.
Across trial distribution (averaged over 128 Monte Carlo runs) of speed and accuracy with means and 95% confidence bounds as kP is varied around its fitted value 1.3098. Accuracy is highest around the fitted value (p < 0.001) and past that speed increases while accuracy decreases.
Fig 7.
Probabilities of entering the targets as a function of target size and learning threshold (FME) for different trial times. For smaller trial times, lower learning thresholds (high learning accuracy) are required to achieve high success probabilities for the same target sizes. Satisficing behavior is observed at high learning accuracy (low FME) levels, where learning with higher accuracy does not necessarily increase success probabilities. The zoomed view (right) for probability curves at high learning accuracy (low FME) levels for 1.2s trial time shows the satisficing effect. Curves are average success probabilities with 95% confidence bounds over 1280 Monte Carlo runs of the HML model. Values of ρ are relative to the size of the unit cell of the game grid.
Fig 8.
FME variation with σu and number of synergies.
FME as a function of increasing σu and number of synergies used. For limited training time, using more synergies is not always the most optimal strategy. Minimum FME (blue cells) is achieved at synergies lower than 19.