Signatures of movement variability anticipate hand speed according to levels of intent

Abstract

Doc number: 10

Abstract

Background: Complex movement sequences are composed of segments with different levels of functionality: intended segments towards a goal and segments that spontaneously occur largely beneath our awareness. It is not known if these spontaneously-occurring segments could be informative of the learning progression in naïve subjects trying to skillfully master a new sport routine.

Methods: To address this question we asked if the hand speed variability could be modeled as a stochastic process where each trial speed depended on the speed of the previous trial. We specifically asked if the hand speed maximum from a previous trial could accurately predict the maximum speed of a sub-sequent trial in both intended and spontaneous movement segments. We further asked whether experts and novices manifested similar models, despite different kinematic dynamics and assessed the predictive power of the spontaneous fluctuations in the incidental motions.

Results: We found a simple power rule to parameterize speed variability for expert and novices with accurate predictive value despite randomly instructed speed levels and training contexts. This rule on average tended to yield similar exponent across speed levels for intended motion segments. Yet for the spontaneous segments the speed fluctuations had exponents that changed as a function of speed level and training context. Two conditions highlighted the expert performance: broad bandwidth of velocity-dependent parameter values and low noise-to-signal ratios that unambiguously distinguished between training regimes. Neither of these was yet manifested in the novices.

Conclusions: We suggest that the statistics of intended motions may be a predictor of overall expertise level, whereas those of spontaneously occurring incidental motions may serve to track learning progression in different training contexts. These spontaneous fluctuations may help the central systems to kinesthetically discriminate the peripheral re-afferent patterns of movement variability associated with changes in movement speed and training context. We further propose that during learning the acquisition of both broad bandwidth of speeds and low noise-to-signal ratios may be critical to build a verifiable kinesthetic (movement) percept and reach the type of automaticity that an expert acquires.