Reinforcement feedback improves motor learning: The role of striatal oscillatory activity explored

By Vijay Kumar MalesuReviewed by Lily Ramsey, LLMMay 31 2024

In a recent study published in Nature Human Behaviour, researchers investigated the causal contribution of specific oscillatory activity patterns within the human striatum to reinforcement motor learning using transcranial temporal interference stimulation (tTIS) with concurrent neuroimaging.

Study: Non-invasive stimulation of the human striatum disrupts reinforcement learning of motor skills. Image Credit: New Africa/Shutterstock.com

Background 

Reinforcement feedback can enhance motor learning, yet the underlying brain mechanisms are not fully understood, particularly regarding the role of specific oscillatory activity within the human striatum.

An emerging area of interest is the potential of noninvasive deep brain stimulation techniques, such as tTIS, to modulate these mechanisms.

Understanding how different frequencies of tTIS impact neural activity and influence motor learning could reveal crucial insights into the striatal contribution to these processes. Further research is needed to explore these relationships and develop targeted motor skill enhancement and rehabilitation interventions.

About the study 

The present study involved 48 right-handed healthy volunteers: 24 (15 women, 25.3 ± 0.7 years old) in the main tTIS study and 24 (14 women, 24.2 ± 0.5 years old) in a behavioral control experiment.

Handedness was assessed using the Edinburgh Handedness Inventory. All participants provided informed consent following the Declaration of Helsinki, approved by the Cantonal Ethics Committee Vaud, Switzerland.

Participants had no neurological or psychiatric disorders and completed a delay-discounting monetary choice questionnaire. They were compensated at 20 CHF per hour.

Participants performed a motor learning task with concurrent tTIS of the striatum and fMRI using a randomized, double-blind, sham-controlled design. They practiced six blocks of trials, combining two reinforcement feedback conditions (Reinf_TYPE: Reinf_ON or Reinf_OFF) with three types of striatal stimulation (tTIS_TYPE: tTIS_Sham, tTIS_20Hz or tTIS_80Hz).

The task involved dynamic force control using a Functional Magnetic Resonance Imaging (fMRI)-compatible grip-force sensor. Participants controlled a cursor by squeezing the sensor, adapting to changing sensory information.

During training, the cursor was intermittently displayed to enhance reinforcement impact. Reinf_ON trials provided real-time success feedback, while Reinf_OFF trials had random color changes.

The protocol included familiarization and training blocks, with tTIS targeting the striatum using optimized electrode placement. Data analysis involved robust linear regressions and functional imaging preprocessing. 

Study results 

A total of 24 healthy participants performed a force-tracking task in an MRI scanner with concurrent tTIS of the striatum. The task required participants to modulate the force applied to a hand-grip sensor to track a moving target with a cursor using their dominant hand. Each block introduced a new pattern of motion for the target.

During Reinf_ON blocks, participants received real-time reinforcement feedback indicating success or failure through green or red targets, respectively.

This feedback followed a closed-loop schedule, updating success criteria based on median performance over the previous four trials. In Reinf_OFF blocks, participants practiced with visually matched random feedback (cyan/magenta).

Both types of blocks employed partial visual feedback of the cursor, enhancing reinforcement effects on motor learning. Pre- and post-training assessments involved full visual feedback without reinforcement or tTIS, evaluating motor learning.

To assess tTIS effects on reinforcement-related motor learning benefits and neural changes, participants completed six blocks of 36 trials with concurrent tTIS during training, delivered at either 20 Hz (tTIS_20Hz), 80 Hz (tTIS_80Hz) or as a sham (tTIS_Sham). The condition order was balanced to reduce carry-over effects.

Electrode montage for optimal striatum stimulation was determined via computational modeling using a realistic head model. The selected montage generated a temporal interference electric field significantly stronger in the striatum than in the overlying cortex.

The Error evaluated task performance, defined as the absolute difference between applied and target force.

Post-training Error was lower than pre-training Error, indicating significant motor learning. Reinforcement improved learning, but this effect depended on the stimulation type. Reinforcement significantly enhanced learning with tTIS_Sham and tTIS_20Hz but not with tTIS_80Hz, indicating that high gamma striatal tTIS disrupts reinforcement benefits.

During training, Errors were generally higher due to visual uncertainty. Reinforcement reduced this disruption, demonstrating participants' ability to use real-time feedback for improvement.

However, tTIS affected tracking performance, with Error increasing during tTIS_20Hz and tTIS_80Hz applications. Additional analysis indicated that tTIS impaired the ability to improve performance during training, with no dependency on reinforcement presence. This effect was attributed to a general impact of tTIS on motor performance.

Task-based fMRI acquired during training allowed evaluation of tTIS's neural effects. Whole-brain analysis revealed increased striatal activity with reinforcement but no tTIS effect.

However, behavioral effects of tTIS_80Hz on reinforcement learning were linked to modulation of striatal neural activity. Effective connectivity analysis showed that tTIS_80Hz enhanced striatum-to-frontal-cortex connectivity, dependent on reinforcement presence.

Conclusions 

This study combined striatal tTIS with electric field modeling, behavioral analyses, and fMRI to assess the striatum's role in motor skill reinforcement learning. tTIS_80Hz disrupted learning from reinforcement feedback, linked to striatal neural activity modulation and increased influence on frontal cortical areas. Individual differences in impulsivity explained variability in tTIS_80Hz effects.

These findings show striatal tTIS can non-invasively modulate striatal mechanisms in reinforcement learning, providing a new tool for studying deep brain structures and behavior. Furthermore, the study highlights the frequency-specific effects of tTIS on striatal and cortical connectivity.