Computational and Behavioral Insights into Task-Dependent Coarticulation of Movement Sequences

Core Concepts

Separate or coarticulated sequences can result from the same task-dependent controller, without implying different representations in the brain.

Abstract

The study presents a computational model and behavioral experiments to investigate how task instructions can influence the coarticulation or separation of movement sequences. The key findings are:

Simulations of an optimal feedback control model show that the requirement to slow down at the first target in a two-reach sequence can reduce the influence of the second target on the first reach, leading to an apparent separation of the sequence elements. In contrast, when there is no penalty on velocity at the first target, the second target can influence the first reach, resulting in coarticulation.

Behavioral experiments with human participants validated the model predictions. In the 'GO' task where there was no constraint on velocity at the first target, the location of the second target influenced the hand trajectory and long-latency feedback responses during the first reach. In contrast, in the 'STOP' task where participants had to slow down at the first target, the second target did not influence the first reach.

The results suggest that coarticulation and separation of sequence elements can arise from the same underlying control mechanism that flexibly integrates multiple goals based on task demands, rather than implying different neural representations of the sequence.

Stats

The hand deviation during the first reach was significantly different between the two second target conditions in the GO task (paired t-test: t(14) = -5.9, p < 10^-4).
The dwell velocity at the first target was significantly different between the two second target conditions in the GO task (paired t-test: t(14) = -3.8, p = 0.002).
The long-latency stretch and shortening responses in the shoulder muscles (PEC and PD) were significantly modulated by the location of the second target in the GO task.

Quotes

"Simulations show that planning for multiple reaches simultaneously allows separate or coarticulated sequences depending on instructions about intermediate goals."
"Human experiments in a two-reach sequence task validated this model."
"The results suggest that coarticulation and separation of sequence elements can arise from the same underlying control mechanism that flexibly integrates multiple goals based on task demands, rather than implying different neural representations of the sequence."

Key Insights Distilled From

Task dependent coarticulation of movement sequences

by Kalidindi,H.... at www.biorxiv.org 12-15-2023

https://www.biorxiv.org/content/10.1101/2023.12.15.571847v2

Deeper Inquiries

How do the computational principles underlying sequence production generalize to longer sequences with more than two elements

The computational principles underlying sequence production can be generalized to longer sequences with more than two elements by extending the same control policy and cost function framework used for two-element sequences. In the model described, the control gains are optimized to minimize a cost function that captures the behavior of the entire sequence. This optimization process can be extended to longer sequences by incorporating additional targets and their associated costs into the control policy. By formulating the cost function to penalize errors and velocities at each target transition, the model can efficiently plan and execute longer sequences while considering the holistic nature of the task.

What are the neural mechanisms that implement the flexible integration of multiple goals in the sensorimotor control system

The flexible integration of multiple goals in the sensorimotor control system is likely implemented through a combination of neural mechanisms that involve feedback control, sensorimotor integration, and motor planning. In the context of the study, the optimal feedback control model suggests that the sensorimotor system can simultaneously process information about multiple targets and adjust control gains based on task requirements. This flexible integration allows for the modulation of movement sequences based on instructions at intermediate goals, leading to either coarticulated or separated movements. Neural circuits involved in this process may include regions responsible for motor planning, sensorimotor integration, and feedback control, such as the primary motor cortex and other areas involved in motor execution and coordination.

How do factors like biomechanics, learning, and individual differences influence the coarticulation of movement sequences

Factors like biomechanics, learning, and individual differences can influence the coarticulation of movement sequences in various ways. Biomechanical constraints can impact the execution of movement sequences by affecting the trajectory and speed of movements, potentially leading to variations in coarticulation patterns. Learning plays a crucial role in shaping movement sequences over time, with practice and habituation leading to the refinement of motor skills and the optimization of control policies. Individual differences in motor control abilities, cognitive strategies, and sensorimotor integration can also influence how coarticulation is manifested in movement sequences, with some individuals showing more efficient and adaptive control of sequences compared to others. Overall, these factors interact to shape the coarticulation of movement sequences and contribute to the variability observed in motor behavior across individuals.