T124. Effect of subthreshold paired associative stimulation during voluntary contraction on different forms of motor learning

Abstract

Introduction Paired associative stimulation can induce long-term potentiation (LTP)-like or long-term depression (LTD)-like synaptic plasticity changes in the human primary motor cortex (M1) by repeated pairing of an electrical stimulus to the median nerve with a suprathreshold transcranial magnetic stimulation (TMS) pulse to M1. The effective interstimulus intervals (ISIs) between electric stimulation and TMS for inducing LTP-like plasticity are approximately 21–25 ms, while ISI of about 10 ms is suitable for inducing LTD-like changes. We previously reported that standard PAS with ISI of 21.5 ms (PAS21.5) and 25 ms (PAS25) differentially modulate two types of motor learning, that is, model-free motor learning was facilitated by PAS21.5 while model-based motor learning was inhibited by PAS25. We argued that standard PAS at different ISIs might involve different inputs to corticospinal neurons, which are sensitive to current direction of TMS applied in PAS. This is because the facilitatory effects of subthreshold PAS during target muscle voluntary contraction (sub-PAS) on motor evoked potentials (MEP) depended on current direction by TMS (i.e. sub-PAS21.5 required posterior-to-anterior (PA) currents, while sub-PAS25 was only effective when applied with anterior-to-posterior (AP) currents). In this study, we examined the effects of sub-PAS on model-free and model-based motor learning. We hypothesized that sub-PAS might be more effective to modulate motor learning than standard PAS, since selective stimulation by sub-PAS would reduce the interindividual variability of PAS effects. In addition to LTP-inducing protocols, we examined whether and how LTD-inducing sub-PAS protocol affects motor learning in this study. Methods Forty healthy volunteers (11 females) participated. We examined changes of performance of model-free learning (i.e. acceleration of repeated rapid thumb abduction) and model-based learning (i.e. visuomotor gain adaptation of pincer grip) after different PAS interventions. ISIs in PAS were adjusted in each subject using the latency of the individual N20 of the right median nerve somatosensory evoked potentials. Results Model-free learning was specifically facilitated by sub-PAS with ISI equaling N20 (sub-PASN20) when applied with PA currents, while this learning was exclusively inhibited by sub-PAS with ISI of N20-8.5 (sub-PASN20−8.5) when applied with AP currents. In contrast, model-based learning was interfered by sub-PAS only when ISI equals N20 + 5 (sub-PASN20+5), irrespective of current direction of TMS. Conclusion These findings showed that model-free and model-based motor learning are mediated in M1 by considerably different procedures, since modulation of model-free learning is likely to depend on specific combination between current direction and ISI adopted in sub-PAS, while model-based motor learning appears to be associated with interneuron circuits dependent only on ISI, but not on current direction applied in sub-PAS.