Limits...
Mapping Muscles Activation to Force Perception during Unloading.

Toma S, Lacquaniti F - PLoS ONE (2016)

Bottom Line: In fact a global measure of the muscles considered was able to predict approximately 60% of the perceptual decisions total variance.Moreover the inter-subjects differences in psychophysical sensitivity showed high correlation with both participants' muscles sensitivity and participants' joint torques.Overall, our findings gave insights into both the role played by the corticospinal motor commands while performing a force detection task and the influence of the gravitational muscular torque on the estimation of vertical forces.

View Article: PubMed Central - PubMed

Affiliation: Centre of Space Bio-medicine, University of Rome Tor Vergata, Rome, Italy.

ABSTRACT
It has been largely proved that while judging a force humans mainly rely on the motor commands produced to interact with that force (i.e., sense of effort). Despite of a large bulk of previous investigations interested in understanding the contributions of the descending and ascending signals in force perception, very few attempts have been made to link a measure of neural output (i.e., EMG) to the psychophysical performance. Indeed, the amount of correlation between EMG activity and perceptual decisions can be interpreted as an estimate of the contribution of central signals involved in the sensation of force. In this study we investigated this correlation by measuring the muscular activity of eight arm muscles while participants performed a quasi-isometric force detection task. Here we showed a method to quantitatively describe muscular activity ("muscle-metric function") that was directly comparable to the description of the participants' psychophysical decisions about the stimulus force. We observed that under our experimental conditions, muscle-metric absolute thresholds and the shape of the muscle-metric curves were closely related to those provided by the psychophysics. In fact a global measure of the muscles considered was able to predict approximately 60% of the perceptual decisions total variance. Moreover the inter-subjects differences in psychophysical sensitivity showed high correlation with both participants' muscles sensitivity and participants' joint torques. Overall, our findings gave insights into both the role played by the corticospinal motor commands while performing a force detection task and the influence of the gravitational muscular torque on the estimation of vertical forces.

Show MeSH

Related in: MedlinePlus

Bootstrap and Stepwise Elimination Procedures.A) For each of the 1000 simulated experiment (resampled stimulus force order) psychometric curve was obtained by fitting a logistic function to the probability of a positive answer with respect to each external force level (x). Ψps, β1ps, β0ps indicate the absolute threshold (PSE), the slope and the intercept, respectively, characterizing the logistic function that best explains the probability of detecting an upward force given each stimulus force. B) Computation of 1000 muscle-metric curves (one per each new force presentation order) obtained by means of a weighted pool of all 8 muscles activity (PoolEMG). C) Computation of 1000 muscle-metric curves (one per each new force presentation order) obtained from each nested model of the muscles activity pooled as above (PoolEMG), whose number of terms was iteratively reduced from 7 to 1. Once 1000 simulated experiments were performed per each nested model, muscle-metric and psycho-metric curves output were compared (bottom left).
© Copyright Policy
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4816335&req=5

pone.0152552.g003: Bootstrap and Stepwise Elimination Procedures.A) For each of the 1000 simulated experiment (resampled stimulus force order) psychometric curve was obtained by fitting a logistic function to the probability of a positive answer with respect to each external force level (x). Ψps, β1ps, β0ps indicate the absolute threshold (PSE), the slope and the intercept, respectively, characterizing the logistic function that best explains the probability of detecting an upward force given each stimulus force. B) Computation of 1000 muscle-metric curves (one per each new force presentation order) obtained by means of a weighted pool of all 8 muscles activity (PoolEMG). C) Computation of 1000 muscle-metric curves (one per each new force presentation order) obtained from each nested model of the muscles activity pooled as above (PoolEMG), whose number of terms was iteratively reduced from 7 to 1. Once 1000 simulated experiments were performed per each nested model, muscle-metric and psycho-metric curves output were compared (bottom left).

Mentions: The hierarchy of the simulation was defined by two sequential steps: the first aimed to define each one of the 8 muscular nested models (i.e., muscles pooling, Step-Wise Backward Elimination procedure), the second aimed to quantify the reliability of its associated muscle-metric parameters (i.e., bootstrap). As shown in Fig 3, each iteration j of the simulation was characterized by a parametric bootstrap resampling of the experimental force presentation order from which new (i.e., 1000) psychometric curve parameters and threshold were extracted (Fig 3A). The same simulated force presentation order was used to compute the muscle-metric curve associated to the PoolEMG distribution obtained from the pattern of eight muscles (i.e., General Mdl0, Fig 3B). After 1000 simulated experiments, each of them producing both a psycho- metric and muscle-metric curve associated with the eight muscles general model, a simplified nested regression model, was obtained by eliminating the least relevant -and/or not statistically significant- muscle term (lower βn, backward elimination, Muscles Pooling in Fig 3C). Then, new 1000 experiments were simulated and muscle-metric curves were extracted by using the PoolEMG distribution obtained by those muscles composing the simplified nested model. Thus, at each loop of such a two steps simulation procedure (bootstrap and backward elimination) a new simplified muscular pattern model M, with i minus 1 number of terms was provided—i being between 2 and 8 the number of terms of the preceding model-. Simulations stopped after 1000 muscle-metric curve of the nested model with only one regression term were extracted. In order to compare the reliability of the psycho vs muscle metrics concordance among muscular regression models the same sequence (i.e., seed) of random force presentation resampling was used for all models and it was changed per each subject (i.e., rng function, twister generator, Mathworks, Natick, MA).


Mapping Muscles Activation to Force Perception during Unloading.

Toma S, Lacquaniti F - PLoS ONE (2016)

Bootstrap and Stepwise Elimination Procedures.A) For each of the 1000 simulated experiment (resampled stimulus force order) psychometric curve was obtained by fitting a logistic function to the probability of a positive answer with respect to each external force level (x). Ψps, β1ps, β0ps indicate the absolute threshold (PSE), the slope and the intercept, respectively, characterizing the logistic function that best explains the probability of detecting an upward force given each stimulus force. B) Computation of 1000 muscle-metric curves (one per each new force presentation order) obtained by means of a weighted pool of all 8 muscles activity (PoolEMG). C) Computation of 1000 muscle-metric curves (one per each new force presentation order) obtained from each nested model of the muscles activity pooled as above (PoolEMG), whose number of terms was iteratively reduced from 7 to 1. Once 1000 simulated experiments were performed per each nested model, muscle-metric and psycho-metric curves output were compared (bottom left).
© Copyright Policy
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4816335&req=5

pone.0152552.g003: Bootstrap and Stepwise Elimination Procedures.A) For each of the 1000 simulated experiment (resampled stimulus force order) psychometric curve was obtained by fitting a logistic function to the probability of a positive answer with respect to each external force level (x). Ψps, β1ps, β0ps indicate the absolute threshold (PSE), the slope and the intercept, respectively, characterizing the logistic function that best explains the probability of detecting an upward force given each stimulus force. B) Computation of 1000 muscle-metric curves (one per each new force presentation order) obtained by means of a weighted pool of all 8 muscles activity (PoolEMG). C) Computation of 1000 muscle-metric curves (one per each new force presentation order) obtained from each nested model of the muscles activity pooled as above (PoolEMG), whose number of terms was iteratively reduced from 7 to 1. Once 1000 simulated experiments were performed per each nested model, muscle-metric and psycho-metric curves output were compared (bottom left).
Mentions: The hierarchy of the simulation was defined by two sequential steps: the first aimed to define each one of the 8 muscular nested models (i.e., muscles pooling, Step-Wise Backward Elimination procedure), the second aimed to quantify the reliability of its associated muscle-metric parameters (i.e., bootstrap). As shown in Fig 3, each iteration j of the simulation was characterized by a parametric bootstrap resampling of the experimental force presentation order from which new (i.e., 1000) psychometric curve parameters and threshold were extracted (Fig 3A). The same simulated force presentation order was used to compute the muscle-metric curve associated to the PoolEMG distribution obtained from the pattern of eight muscles (i.e., General Mdl0, Fig 3B). After 1000 simulated experiments, each of them producing both a psycho- metric and muscle-metric curve associated with the eight muscles general model, a simplified nested regression model, was obtained by eliminating the least relevant -and/or not statistically significant- muscle term (lower βn, backward elimination, Muscles Pooling in Fig 3C). Then, new 1000 experiments were simulated and muscle-metric curves were extracted by using the PoolEMG distribution obtained by those muscles composing the simplified nested model. Thus, at each loop of such a two steps simulation procedure (bootstrap and backward elimination) a new simplified muscular pattern model M, with i minus 1 number of terms was provided—i being between 2 and 8 the number of terms of the preceding model-. Simulations stopped after 1000 muscle-metric curve of the nested model with only one regression term were extracted. In order to compare the reliability of the psycho vs muscle metrics concordance among muscular regression models the same sequence (i.e., seed) of random force presentation resampling was used for all models and it was changed per each subject (i.e., rng function, twister generator, Mathworks, Natick, MA).

Bottom Line: In fact a global measure of the muscles considered was able to predict approximately 60% of the perceptual decisions total variance.Moreover the inter-subjects differences in psychophysical sensitivity showed high correlation with both participants' muscles sensitivity and participants' joint torques.Overall, our findings gave insights into both the role played by the corticospinal motor commands while performing a force detection task and the influence of the gravitational muscular torque on the estimation of vertical forces.

View Article: PubMed Central - PubMed

Affiliation: Centre of Space Bio-medicine, University of Rome Tor Vergata, Rome, Italy.

ABSTRACT
It has been largely proved that while judging a force humans mainly rely on the motor commands produced to interact with that force (i.e., sense of effort). Despite of a large bulk of previous investigations interested in understanding the contributions of the descending and ascending signals in force perception, very few attempts have been made to link a measure of neural output (i.e., EMG) to the psychophysical performance. Indeed, the amount of correlation between EMG activity and perceptual decisions can be interpreted as an estimate of the contribution of central signals involved in the sensation of force. In this study we investigated this correlation by measuring the muscular activity of eight arm muscles while participants performed a quasi-isometric force detection task. Here we showed a method to quantitatively describe muscular activity ("muscle-metric function") that was directly comparable to the description of the participants' psychophysical decisions about the stimulus force. We observed that under our experimental conditions, muscle-metric absolute thresholds and the shape of the muscle-metric curves were closely related to those provided by the psychophysics. In fact a global measure of the muscles considered was able to predict approximately 60% of the perceptual decisions total variance. Moreover the inter-subjects differences in psychophysical sensitivity showed high correlation with both participants' muscles sensitivity and participants' joint torques. Overall, our findings gave insights into both the role played by the corticospinal motor commands while performing a force detection task and the influence of the gravitational muscular torque on the estimation of vertical forces.

Show MeSH
Related in: MedlinePlus