Sensorimotor feedback loops are selectively sensitive to reward

  1. Olivier Codol  Is a corresponding author
  2. Mehrdad Kashefi
  3. Christopher J Forgaard
  4. Joseph M Galea
  5. J Andrew Pruszynski
  6. Paul L Gribble
  1. Brain and Mind Institute, University of Western Ontario, Canada
  2. Department of Psychology, University of Western Ontario, Canada
  3. School of Psychology, University of Birmingham, United Kingdom
  4. Department of Physiology & Pharmacology, Schulich School of Medicine & Dentistry, University of Western Ontario, Canada
  5. Robarts Research Institute, University of Western Ontario, Canada
  6. Haskins Laboratories, United States
14 figures, 2 tables and 1 additional file

Figures

Different sensorimotor feedback responses are emphasized in different task designs.

Feedback responses can be classified along three dimensions: the sensory modality on which they rely (vertical axis), their post-perturbation latency (horizontal axis), and the function they perform …

Results for the SLR contrast.

(a) Schematic representation of the apparatus from a top view. Participants could move their arm in a horizontal plane. Background forces were applied to pre-activate the extensor muscles (dashed …

Results for the LLR contrast.

(a) Contrast used to observe the LLR. Background loads are not drawn here for clarity. (b) Example trajectories for one participant for an outward (blue) or inward (brown) target. (c) Schematic …

Results for the Target Selection task.

(a) Schematic representation of the apparatus from a top view. Participants could move their arm in a horizontal plane. Background forces were applied to pre-activate the extensor muscles (dashed …

Results for the Reaction Time tasks.

(a) Schematic of task design for Proprioception-cued Reaction Times. Participants were informed to initiate an elbow extension by a small mechanical perturbation at the shoulder (solid black arrow). …

results for the Cursor Jump task.

(a) Schematic representation of the apparatus from a top view. Participants could move their arm in a horizontal plane. 2 N·m Background forces were applied to pre-activate the extensor muscles …

Results for the Target Jump task.

(a) Schematic representation of the apparatus from a top view. Participants could move their arm in a horizontal plane. 2 N·m Background forces were applied to pre-activate the extensor muscles …

Overview of expected reward impact on sensorimotor feedback responses.

Reward can impact a feedback loop response by increasing feedback gains or reducing latency. The color code indicates function and is identical to the one in Figure 1. Results for the Alternative …

Author response image 1
Position of maximum excursion following the perturbation in the condition with an inward (counterclockwise) push and an outward (clockwise) target.

Trials where reward was provided are colorcoded in green, and trials where no reward was provided are color-coded in red. The triangle indicates the starting position from which the perturbation …

Author response image 2
Author response image 3
Simulated data with gaussian noise (noise standard deviation = 0).

2).

Author response image 4
Estimated latency as noise standard deviation in the simulated data was varied.

The blue line indicates the true latency of signal divergence.

Author response image 5
Extrapolation method from Brenner and Smeets, 2019, on triceps EMG from the cursor jump task.
Author response image 6
ROC method from Weiler et al.

, 2015, on triceps EMG from the cursor jump task. The solid blue lines indicate the segmented linear fits, and the “knee” of this fit is highlighted with vertical dashed lines.

Tables

Table 1
Task to feedback response mapping.

This table indicates the correspondence between tasks and published work used, and the feedback responses assessed in the present study. RT, reaction time.

Feedback responseTaskReference
SLRIn-Out Target task
LLR
Target SelectionTarget Selection
Target JumpTarget Jump
Cursor JumpCursor Jump
Proprioception-cued RTsProprioception-cued RTs
Vision-cued RTsStillings et al., 1968
Alternative TargetCarroll et al., 2019
Choice RTs(Data set re-analysed)Codol et al., 2020a
Table 2
Parameters used to compute the return in each rewarded trial, for each condition in the Proprioception-cued Reaction Time, Cursor Jump, Target Jump, and Target Selection tasks.
TaskConditionScalerShifterτMTmax (ms)
Reaction TimeN/A102.447728
Cursor JumpInward0.996–0.0294.2732781
No jump0.6670.0795.4332335
Outward0.996–0.0413.9582864
Target JumpInward0.999–0.0264.2812697
No jump0.683–0.0403.8932882
Outward0.999–0.0543.8532690
Target SelectionOne target
Inward Pert.
0.676–0.0346.2361673
One target
Outward Pert.
0.6900.0045.5342241
Two targets
Inward Pert.
0.749–0.0214.9042373
Two targets
Outward Pert.
0.7490.0095.3502208

Additional files

Download links