Neurophysiological mechanisms involved in transfer of procedural knowledge

Abstract

Learning to perform a motor task with one hand results in performance improvements in the other hand, a process called intermanual transfer. To gain information on its neural mechanisms, we studied this phenomenon using the serial reaction-time task (SRTT). Sixteen, right-handed volunteers trained a 12-item sequence of key presses repeated without the subjects' knowledge. Blocks with no repeating sequence, called random blocks, were interspersed with sequence-training blocks. Response times improved in random and training blocks in both hands. The former result reflects nonspecific improvement in performance, and the latter represents a sequence-specific improvement. To evaluate changes in the primary motor cortex (M1), we tested resting motor thresholds (RMT), recruitments curves to transcranial magnetic stimulation (RC), short intracortical inhibition (SICI), and interhemispheric inhibition (IHI) from the dominant left (learning) to the nondominant right (transfer) hemisphere, before and after SRTT training. Training resulted in (1) increased RC and decreased SICI but no changes in RMT in the learning hemisphere, (2) decreased SICI and no changes in RC or RMT in the transfer hemisphere, and (3) decreased IHI. The amount in IHI after training correlated with nonspecific performance improvements in the transfer hand but not with sequence-specific performance improvements. Our results indicate that modulation of interhemispheric inhibition between the M1 areas may, as a result of the learning that has occurred in one hemisphere after practice with one hand, contribute to faster, more skilled performance of the opposite hand.

Figure 1.

Experimental design. A, Diagram showing the order of the sequence and random (R) blocks practiced with the right and left hand. For the right hand, procedural learning that was sequence specific was calculated as the difference in RT between the last sequence block (A10) and the last random block (R4) (Willingham et al., 2000), and learning that was nonspecific to the sequence was calculated as the difference in RT between the right-hand random blocks R2 and R4 (dotted lines). For the left hand, sequence-specific transfer was calculated as the difference in RT between the sequence (A) and random block (R2) and nonspecific performance as the difference in RT between the R2 random blocks in the right and left hands (dashed lines). B, Diagram showing measures of motor cortical excitability tested before and after the SRTT.

Figure 2.

Response time (A) and mean errors (B) in random and sequence blocks during the SRTT. A, The abscissa shows block type and performing hand in temporal order, and the ordinate shows response times. Note the progressive shortening of response times in random (R1–R4, right hand) and sequence A (A1–A10, right hand) blocks during performance of the SRTT with the right hand, maintained at the end of the experiment (A11, A12, right hand). Performance of the same sequence A with the left hand showed shorter response times compared with both unpracticed sequence B (gray bar) and random blocks. Sequence-specific transfer and nonspecific transfer-hand performances are shown in dotted lines for the right-hand and left transfer-hand performance in dashed lines. B, Mean number of accuracy errors (wrong key presses) with the right and left hand. The abscissa shows the order of all random and sequence blocks practiced with the right and left hand. Note the stable overall accuracy during the course of the experiment. Variance is expressed as SE; *p < 0.05.

Figure 3.

MEP recruitment curves. A, B, MEP amplitudes from FDI muscles (A, B) before and after (pre and post) training on the SRRT task with stimulation of the left (A) and right (B) motor cortices in a single subject. Note the increase in MEP amplitudes in the learning hemisphere after the SRTT. C and D show the recruitment curves group data indicating the effect of SRTT training with the right hand on left (C) and right (D) FDI excitability recruitment curves. The abscissa shows TMS stimulus intensity relative to the RMT in each subject, and the ordinate shows MEP amplitudes (as a percentage of the FDI M-max). Measurements were taken before (filled circles) and after (open circles) the SRTT training. Note the increase in recruitment curves in the learning hemisphere in the absence of changes in the transfer hemisphere. Error bars indicate SEs; *p < 0.05.

Figure 4.

Short intracortical inhibition. SICI (2.5 ms) in FDI (A, B) measured as the ratio between [(conditioned MEP × 100)/test MEP] before (Pre) and after (Post) performance of the SRTT task with stimulation of the learning (A) and transfer (B) hemisphere in a single subject. Note the presence of disinhibition in both learning and transfer hemispheres.

Figure 5.

Interhemispheric inhibition. A, IHI at 10 ms from a single subject recorded before (Pre) and after (Post) the SRTT task. Test MEPs are shown in solid lines and conditioned MEPs in dotted lines. Note the well defined IHI before training (Pre) at the three test MEP amplitudes (test MEP1, test MEP2, and test MEP3; see Materials and Methods) and the relative disinhibition shown after training (Post). B, Group data (n = 10). The abscissa shows the three test MEP amplitudes evaluated (0.34 ± 0.05, 0.86 ± 0.06, and 1.41 ± 0.04 mV) with a CS of 120% RMT. The ordinate indicates the magnitude of IHI, in which the size of the conditioned MEP is expressed as a percentage of the size of test MEP amplitude. Note the attenuation in IHI (from filled circles to open circles) after practice of the SRTT, identifiable at all three test MEP amplitudes. C, Influence of changing conditioning stimulus intensities (expressed as percentage of RMT) on the magnitude of IHI. Note the existence of attenuation in IHI (from filled circles to open circles) at all intensities above RMT. Error bars indicate SEs; *p < 0.05.

Figure 6.

Correlations. A, The abscissa indicates the magnitude of IHI_, in which the size of the conditioned MEP is expressed as a percentage of the size of test MEP amplitude. The ordinate shows the median RT in random block (R2) practiced with the left (transfer) hand (nonspecific transfer-hand performance). Note that the decrease in RT is associated with a decrease in IHI. B, The abscissa is as in A. The ordinate shows the median RT of the difference between blocks R2 and A practiced with the left hand (sequence-specific transfer). Note that there was no relationship between RT and IHI.