frederic danion, 1 * jonathan s. diamond, 2 * and j. randall flanagan 2,3

Separate Contributions of Kinematic and Kinetic Errors to Trajectory and Grip Force Adaptation

When Transporting Novel Hand-Held Loads

Frederic Danion,1* Jonathan S. Diamond,2* and J. Randall Flanagan2,31 CNRS and Aix-Marseille University, 2 Centre for Neuroscience Studies

and 3 Department of Psychology, Queen’s University

The Journal of Neuroscience, January 30, 2013 • 33(5):2229 –2236

Motor Control Journal ClubMay 20, 2013

Adaptation to novel dynamics

Initially, unknown mapping between applied force and motion when interacting with new object• Kinematic errors: predicted ≠ actual trajectory

Perturbed trajectory• Kinetic errors: predicted ≠ actual load force

Poor modulation of grip force

With practice, people adapt to newdynamics (update internal model)

Adaptation and errorTrajectory adaptation Grip force adaptation

Kinematic error √ ?

Kinetic error ? √

Trajectory adaptationHand path straightens out

Grip force adaptationBetter grip-load modulation

Common or separate internal models?

Do kinematic and kinetic errors update the same model, or separate models?

Modified from Wolpert and Flanagan (2001)

Prediction error used to updated internal model (predictor)

Main ideas of experiment

• Compare grip force adaptation with and without kinematic errors

• Compare trajectory adaptation with and without prior exposure to kinetic errors

Experimental setupPoint-to-point movements

Object in hand• Dynamics rendered by Phantom• Object cursor visible (contextual clue)

Force channel• WristBOT remove kinematic error resulting from moving

object• Wrist cursor and slider visible (contextual clue)• Explore channel (distinguish from object dynamics)

Movement speed• Standard: 400 ms• Fast: 200 ms

Experimental protocols

Group A• Block 1: practice• Block 2: practice

(control)• Block 3: grip force

and trajectory adaptation with kinetic + kinematic errors

Experimental protocols

Group B• Block 1: practice• Block 2: grip force

adaptation with kinetic errors only

• Block 3: trajectory adaptation (facilitated by previous experience of kinetic errors?)

Experimental protocols

Group C• Block 1: practice• Block 2: grip force

adaptation with kinetic errors only (increase salience of object dynamics)

• Block 3: trajectory adaptation (facilitated by previous experience of kinetic errors?)

Data analysis

• Grip-load force coupling– Cross-correlation coefficient (at zero lag)

• Trajectory perturbation– Peak-to-peak lateral deviation

• Adaptation– Fit exponential curve: y = aebx + c

Results: Grip force adaptation in channel

Poor grip-load correlation in 1st trial

Object path only slightly perturbed

• Increase in correlation over trials

• Asymptote: B > C (higher because of consecutive trials?)

Load force scaled with object velocity

Results: Transfer of grip force adaptation outside channel

Maintain predictive grip force control - good grip-load correlation in 1st trial(Higher magnitude?)

Object and wrist paths greatly perturbed

Corrective movements apparent in velocity profile

• Benefit of prior exposure to object dynamics in the channel

• No benefit of additional kinematic errors – similar learning rates for 3 groups

• Higher R without channel than with channel for groups A and B?

Results: Trajectory adaptation outside channel

No benefit of prior exposure to object dynamics (kinetic errors) in channel (groups B and C)

Experimental protocols

Group D• Block 1: practice• Block 2: grip force

adaptation with kinetic errors only (generalize over different speeds?)

• Block 3: trajectory adaptation (facilitated by previous experience of kinetic errors?)

2x speed 2x load force

Results: Generalization across speeds within channel

Predictive grip force generalized across movement speedsExposure to object dynamics at different movements speeds in the channel did not facilitate trajectory adaptation after the channel was removed.

Summary of grip force adaptation

• Kinetic errors are sufficient to drive grip force adaptation– Equivalent grip force adaptation with channel (kinetic

errors) and without channel (kinetic + kinematic errors)

– Importance of tactile vs. proprioceptive information

• Involves updating of internal model of object dynamics– Generalized over different movement speeds

Summary of trajectory adaptation

• Kinematic errors are necessary for arm movement adaptation– Prior experience of moving the object in the

channel (kinetic errors) did not benefit trajectory adaptation when moving without the channel

• Suggests distinct internal models of object dynamics for grip force and trajectory control