Distinct basal ganglia territories are engaged in early and advanced motor sequence learning

Lehéricy et al. 10.1073/pnas.0502762102.

Supporting Information

Files in this Data Supplement:

Supporting Text
Supporting Table 1
Supporting Figure 3
Supporting Figure 4

Supporting Figure 3

Fig. 3. Behavioral results. (Left) Performance during the speed tests outside the magnet (each data point represents the mean ± SD of five tests, 30 s long, averaged over the 14 subjects). (Upper Left) Movement frequency. (Lower Left) Percentage of errors. (Right) Performance during the MRI sessions. (Upper Right) Average reaction time across the 10 epochs of the first three runs of session1 (Begin, 10 min and 50 min on day1), session 2 (day 14), and session 3 (day 28) of the trained sequence. Reaction times decreased rapidly and significantly over the 10 epochs of the first run and remained unchanged afterward (ANOVA for repeated measurements, P = 0.045). (Lower Right) Percentage of errors decreased more slowly and regularly over the 3 MRI sessions (63% fewer errors, ANOVA for repeated measurements, P = 0.009).

Supporting Figure 4

Fig. 4. Activation patterns in the cortex and the cerebellum. (A) Areas activated during early learning. Comparison of run 1 (action - rest, first run = T1) versus run 3 (action - rest after 50 min of practice = T3) of the trained sequence (height threshold P < 0.001, cluster corrected at P < 0.05). Activation maps are superimposed on a 3D T₁-weighted surface render (Lower Right) or axial views of the average echo-planar imaging (EPI) image of the 14 subjects (all other views). There was more activation at the beginning of learning than after 50 min of practice in bilateral pre-SMA, lateral premotor areas, and ventral prefrontal cortex, associative parietal areas (BA 40), dorsal putamen, anterior thalamus, subthalamic nucleus, mesencephalon, and cerebellar hemispheres, the pons and the right dorsolateral prefrontal cortex. (B) Areas more activated during late learning. Comparison of run 3 (action – rest after 50 min of practice = T3) versus run 1 (action – rest, first run = T1) of the trained sequence. Activation maps are superimposed on sagittal (Lower Left) and axial views (all other views) of the average EPI image of the 14 subjects. There was more activation after 50 min of practice than at the beginning of learning in the orbitomedial frontal cortex, the left inferior parietal cortex, the left amygdala, and posterior cerebellum (P < 0.001, not corrected).

Table 1. Performances during speed tests (mean ± SD)

		Trained			Untrained
		Frequency, Hz	Inter-tap interval, ms	Errors, %	Frequency, Hz	Inter-tap interval, ms	Errors, %
		Frequency, Hz	Inter-tap interval, ms	Errors, %	Frequency, Hz	Inter-tap interval, ms	Errors, %
Day 1 pre		2.51 ± 0.11	398.2 ± 43.3	0.88 ± 0.94	2.64 ± 0.10	379.0 ± 38.3	1.23 ± 0.82
Day 1 post		3.50 ± 0.09	286.0 ± 26.2	0.49 ± 0.36
Day 7		4.47 ± 0.04	223.7 ± 9.6	0.57 ± 0.90
Day 14		4.76 ± 0.05	210.2 ± 10.5	0.63 ± 0.78	3.15 ± 0.09	317.9 ± 28.7	1.48 ± 0.94
Day 21		5.53 ± 0.05	180.9 ± 8.7	0.42 ± 0.28
Day 28	Single	5.59 ± 0.05	179.0 ± 9.1	0.25 ± 0.21	3.33 ± 0.08	300.6 ± 24.0	1.40 ± 0.93
		(89.0 ± 9.9)
	Dual	4.55 ± 0.10	220.5 ± 15.0	1.22 ± 1.21	1.84 ± 0.35	544.9 ± 102.0	9.38 ± 6.66
		(80.3 ± 4.8)			(62.0 ± 3.92)

Numbers in parenthesis indicate the number of words read during 30 s.

Supporting Text

Goniometer Measurements. The amplitude of subjects’ movements was measured outside the MR unit in the 14 subjects using a single-axis goniometer (F35) and angle display unit (ADU301), both manufactured by Biometrics Ltd. (Cwmfelinfach, Gwent, U.K.). The goniometer was placed on the proximal joint of the index finger. Data were collected while subjects performed the T-sequence, audio-paced sequence at 2 Hz during 1 min. Measurements were obtained before and after 4 weeks of training. The data of one subject were discarded due to large artifacts. Data were electronically recorded and preprocessed using a customized labview program by National Instruments (Austin, TX) at a sampling rate of 100 Hz. They were smoothed with a low-pass Butterwoth filter (cutoff 5 Hz). Statistical analysis was performed with spss 11.5 for windows (SPSS, Chicago, IL).

Cortical Activation Changes with Learning. In the cortex, activation decreased with practice (T1 - T3 at P < 0.001, T3 - T4 and T4 - T5 at P < 0.01, Montreal Neurological Institute coordinates (MC) are provided for T1 - T3) in bilateral pre-supplementary motor area (SMA) (MC, -5, +8, +47 and +9, +12, +47), lateral premotor cortex [Brodmann’s area (BA) 6; MC, -21, -6, +54 and +27, -11, +50], inferior prefrontal (BA 44; MC, -51, +6, 0 and +53, +6, +11), parietal cortex (BA 40; MC, -36, -41, +50 and +39, -33, +45) and insula (MC, -39, +12, +3 and +33, +17, +3), the left SMA (MC, -11, -2, +51) and right anterior cingulate cortex (ACC; MC, +11, +24, +30) and dorsal prefrontal cortex (BA 9; MC, +36, +50, +27) (Fig. 4A).

By contrast, activation increased during the first session in several cortical areas (T3 - T1, P < 0.001, Fig. 4B), including reward-related areas (bilateral orbitomedial frontal cortex; MC, +3, +35, -18) and the right posterior cerebellar hemisphere (+24, -86, -32). At lower thresholds (P < 0.001, uncorrected), there were additional activation trends in the left amygdala (MC, -30, 0, +21), the left inferior parietal cortex (BA 40; MC, -48, -69, +41) and the left posterior cerebellar hemisphere (MC, -26, -81, -39). Finally, there was also a trend for an increase in activation in the hippocampus/parahippocampal gyrus bilaterally on day 14 (MC, -23, -17, -21 and +26, -14, -18; T4 - T1, P < 0.001) and day 28 (MC, -24, -33, -12 and +32, -14, -18; T5 - T1, P < 0.001, uncorrected).