Consecutive training on two movement sequences often leads to retroactive interference—obstructing memory for the initially trained sequence but not for the second. However, in the context of hippocampal-system dependent memories, a poor learning experience, memory for which would soon decay, can be enhanced if temporally paired with a “strong,” memory triggering experience. The synaptic tagging and capture hypothesis explains this paradoxical enhancement by suggesting that only strong experiences generate cellular resources necessary for synaptic remodeling. However, synapses engaged in a “weak” learning experience can capture and utilize plasticity-related resources generated for a subsequent strong learning experience. Here, we tested whether such “paradoxical” outcome would result in the context of motor (procedural) memory, if two movement sequences are unequally trained, consecutively. We show, in young adults ( n = 100), that limited practice on a novel sequence of finger-to-thumb opposition movements led to different long-term outcomes, depending on whether and when (5 min, 5 hr) it was followed by extensive training on a different sequence. Five-minute pairing, only, resulted in overnight gains for the limited-trained sequence that were well-retained a week later; the overnight gains for the extensively trained sequence were compromised. Thus, consecutive training on different motor tasks can result in mnemonic interactions other than interference. We propose that the newly discovered mnemonic interaction provides the first-tier behavioral evidence in support of the possible applicability of notions stemming from the synaptic tagging and capture hypothesis in relation to human motor memory generation, specifically in relation to the practice-dependent consolidation of novel explicitly instructed movement sequences.

It is not clear how the engagement of motor mnemonic processes is expressed in online brain activity. We scanned participants, using fMRI, during the paced performance of a finger-to-thumb opposition sequence (FOS), intensively trained a day earlier (T-FOS), and a similarly constructed, but novel, untrained FOS (U-FOS). Both movement sequences were performed in pairs of blocks separated by a brief rest interval (30 sec). We have recently shown that in the primary motor cortex (M1) motor memory was not expressed in the average signal intensity but rather in the across-block signal modulations, that is, when comparing the first to the second performance block across the brief rest interval. Here, using an M1 seed, we show that for the T-FOS, the M1–striatum functional connectivity decreased across blocks; however, for the U-FOS, connectivity within the M1 and between M1 and striatum increased. In addition, in M1, the pattern of within-block signal change, but not signal variability per se, reliably differentiated the two sequences. Only for the U-FOS and only within the first blocks in each pair, the signal significantly decreased. No such modulation was found within the second corresponding blocks following the brief rest interval in either FOS. We propose that a network including M1 and striatum underlies online motor working memory. This network may promote a transient integrated representation of a new movement sequence and readily retrieves a previously established movement sequence representation. Averaging over single events or blocks may not capture the dynamics of motor representations that occur over multiple timescales.

An almost universally accepted tacit expectation is that learning and memory consolidation processes must be reflected in the average brain activity in brain areas relevant to task performance. Motor cortex (M1) plasticity has been implicated in motor skill acquisition and its consolidation. Nevertheless, no consistent pattern of changes in the average signal, related to motor learning or motor memory consolidation following a single session of training, has emerged from imaging studies. Here we show that the pattern and magnitude of short-term brain activity modulations in response to task repetition, in M1, may provide a robust signature for effective motor memory consolidation processes. We studied participants during the paced performance of a finger-to-thumb opposition sequence (FOS), intensively trained a day earlier, and a similarly constructed untrained FOS. In addition to within-session “on-line” gains, most participants expressed delayed, consolidation-phase gains in the performance of the trained FOS. The execution of the trained FOS induced repetition enhancements in the contralateral M1 and bilaterally in the medial-temporal lobes, offsetting novelty-related repetition suppression effects. Moreover, the M1 modulations were positively correlated with the magnitude of each participant's overnight delayed gains but not with absolute performance levels. Our results suggest that short-term enhancements of brain signals upon task repetition reflect the effectiveness of overnight motor memory consolidation. We propose that procedural memory consolidation processes may affect the excitation–inhibition balance within cortical representations of the trained movements; this new balance is better reflected in repetition effects than in the average level of evoked neural activity.