14 Citations (Scopus)

Abstract

The time course of grip force from object contact to onset of manipulation has been extensively studied to gain insight into the underlying control mechanisms. Of particular interest to the motor neuroscience and clinical communities is the phenomenon of bell-shaped grip force rate (GFR) that has been interpreted as indicative of feedforward force control. However, this feature has not been assessed quantitatively. Furthermore, the time course of grip force may contain additional features that could provide insight into sensorimotor control processes. In this study, we addressed these questions by validating and applying two computational approaches to extract features from GFR in humans: 1) fitting a Gaussian function to GFR and quantifying the goodness of the fit [root-mean-square error, (RMSE)]; and 2) continuous wavelet transform (CWT), where we assessed the correlation of the GFR signal with a Mexican Hat function. Experiment 1 consisted of a classic pseudorandomized presentation of object mass (light or heavy), where grip forces developed to lift a mass heavier than expected are known to exhibit corrective responses. For Experiment 2, we applied our two techniques to analyze grip force exerted for manipulating an inverted T-shaped object whose center of mass was changed across blocks of consecutive trials. For both experiments, subjects were asked to grasp the object at either predetermined or self-selected grasp locations ('constrained' and 'unconstrained' task, respectively). Experiment 1 successfully validated the use of RMSE and CWT as they correctly distinguished trials with versus without force corrective responses. RMSE and CWT also revealed that grip force is characterized by more feedback-driven corrections when grasping at self-selected contact points. Future work will examine the application of our analytical approaches to a broader range of tasks, e.g., assessment of recovery of sensorimotor function following clinical intervention, interlimb differences in force control, and force coordination in human-machine interactions.

Original languageEnglish (US)
Article number7004021
Pages (from-to)1363-1375
Number of pages13
JournalIEEE Transactions on Biomedical Engineering
Volume62
Issue number5
DOIs
StatePublished - May 1 2015

Fingerprint

Mean square error
Wavelet transforms
Force control
Experiments
Feedforward control
Point contacts
Feedback
Recovery

Keywords

  • Continuous wavelet transform (CWT)
  • feedback
  • feedforward
  • hand
  • sensorimotor memory.

ASJC Scopus subject areas

  • Biomedical Engineering

Cite this

Extraction of time and frequency features from grip force rates during dexterous manipulation. / Mojtahedi, Keivan; Fu, Qiushi; Santello, Marco.

In: IEEE Transactions on Biomedical Engineering, Vol. 62, No. 5, 7004021, 01.05.2015, p. 1363-1375.

Research output: Contribution to journalArticle

@article{dac5162b9e8a432e9871ab7a8ebd2a14,
title = "Extraction of time and frequency features from grip force rates during dexterous manipulation",
abstract = "The time course of grip force from object contact to onset of manipulation has been extensively studied to gain insight into the underlying control mechanisms. Of particular interest to the motor neuroscience and clinical communities is the phenomenon of bell-shaped grip force rate (GFR) that has been interpreted as indicative of feedforward force control. However, this feature has not been assessed quantitatively. Furthermore, the time course of grip force may contain additional features that could provide insight into sensorimotor control processes. In this study, we addressed these questions by validating and applying two computational approaches to extract features from GFR in humans: 1) fitting a Gaussian function to GFR and quantifying the goodness of the fit [root-mean-square error, (RMSE)]; and 2) continuous wavelet transform (CWT), where we assessed the correlation of the GFR signal with a Mexican Hat function. Experiment 1 consisted of a classic pseudorandomized presentation of object mass (light or heavy), where grip forces developed to lift a mass heavier than expected are known to exhibit corrective responses. For Experiment 2, we applied our two techniques to analyze grip force exerted for manipulating an inverted T-shaped object whose center of mass was changed across blocks of consecutive trials. For both experiments, subjects were asked to grasp the object at either predetermined or self-selected grasp locations ('constrained' and 'unconstrained' task, respectively). Experiment 1 successfully validated the use of RMSE and CWT as they correctly distinguished trials with versus without force corrective responses. RMSE and CWT also revealed that grip force is characterized by more feedback-driven corrections when grasping at self-selected contact points. Future work will examine the application of our analytical approaches to a broader range of tasks, e.g., assessment of recovery of sensorimotor function following clinical intervention, interlimb differences in force control, and force coordination in human-machine interactions.",
keywords = "Continuous wavelet transform (CWT), feedback, feedforward, hand, sensorimotor memory.",
author = "Keivan Mojtahedi and Qiushi Fu and Marco Santello",
year = "2015",
month = "5",
day = "1",
doi = "10.1109/TBME.2015.2388592",
language = "English (US)",
volume = "62",
pages = "1363--1375",
journal = "IEEE Transactions on Biomedical Engineering",
issn = "0018-9294",
publisher = "IEEE Computer Society",
number = "5",

}

TY - JOUR

T1 - Extraction of time and frequency features from grip force rates during dexterous manipulation

AU - Mojtahedi, Keivan

AU - Fu, Qiushi

AU - Santello, Marco

PY - 2015/5/1

Y1 - 2015/5/1

N2 - The time course of grip force from object contact to onset of manipulation has been extensively studied to gain insight into the underlying control mechanisms. Of particular interest to the motor neuroscience and clinical communities is the phenomenon of bell-shaped grip force rate (GFR) that has been interpreted as indicative of feedforward force control. However, this feature has not been assessed quantitatively. Furthermore, the time course of grip force may contain additional features that could provide insight into sensorimotor control processes. In this study, we addressed these questions by validating and applying two computational approaches to extract features from GFR in humans: 1) fitting a Gaussian function to GFR and quantifying the goodness of the fit [root-mean-square error, (RMSE)]; and 2) continuous wavelet transform (CWT), where we assessed the correlation of the GFR signal with a Mexican Hat function. Experiment 1 consisted of a classic pseudorandomized presentation of object mass (light or heavy), where grip forces developed to lift a mass heavier than expected are known to exhibit corrective responses. For Experiment 2, we applied our two techniques to analyze grip force exerted for manipulating an inverted T-shaped object whose center of mass was changed across blocks of consecutive trials. For both experiments, subjects were asked to grasp the object at either predetermined or self-selected grasp locations ('constrained' and 'unconstrained' task, respectively). Experiment 1 successfully validated the use of RMSE and CWT as they correctly distinguished trials with versus without force corrective responses. RMSE and CWT also revealed that grip force is characterized by more feedback-driven corrections when grasping at self-selected contact points. Future work will examine the application of our analytical approaches to a broader range of tasks, e.g., assessment of recovery of sensorimotor function following clinical intervention, interlimb differences in force control, and force coordination in human-machine interactions.

AB - The time course of grip force from object contact to onset of manipulation has been extensively studied to gain insight into the underlying control mechanisms. Of particular interest to the motor neuroscience and clinical communities is the phenomenon of bell-shaped grip force rate (GFR) that has been interpreted as indicative of feedforward force control. However, this feature has not been assessed quantitatively. Furthermore, the time course of grip force may contain additional features that could provide insight into sensorimotor control processes. In this study, we addressed these questions by validating and applying two computational approaches to extract features from GFR in humans: 1) fitting a Gaussian function to GFR and quantifying the goodness of the fit [root-mean-square error, (RMSE)]; and 2) continuous wavelet transform (CWT), where we assessed the correlation of the GFR signal with a Mexican Hat function. Experiment 1 consisted of a classic pseudorandomized presentation of object mass (light or heavy), where grip forces developed to lift a mass heavier than expected are known to exhibit corrective responses. For Experiment 2, we applied our two techniques to analyze grip force exerted for manipulating an inverted T-shaped object whose center of mass was changed across blocks of consecutive trials. For both experiments, subjects were asked to grasp the object at either predetermined or self-selected grasp locations ('constrained' and 'unconstrained' task, respectively). Experiment 1 successfully validated the use of RMSE and CWT as they correctly distinguished trials with versus without force corrective responses. RMSE and CWT also revealed that grip force is characterized by more feedback-driven corrections when grasping at self-selected contact points. Future work will examine the application of our analytical approaches to a broader range of tasks, e.g., assessment of recovery of sensorimotor function following clinical intervention, interlimb differences in force control, and force coordination in human-machine interactions.

KW - Continuous wavelet transform (CWT)

KW - feedback

KW - feedforward

KW - hand

KW - sensorimotor memory.

UR - http://www.scopus.com/inward/record.url?scp=84929075005&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84929075005&partnerID=8YFLogxK

U2 - 10.1109/TBME.2015.2388592

DO - 10.1109/TBME.2015.2388592

M3 - Article

C2 - 25576561

AN - SCOPUS:84929075005

VL - 62

SP - 1363

EP - 1375

JO - IEEE Transactions on Biomedical Engineering

JF - IEEE Transactions on Biomedical Engineering

SN - 0018-9294

IS - 5

M1 - 7004021

ER -