Performing sequences of movements from memory and adapting them to changing task demands is a hallmark of skilled human behaviour, from handwriting to playing a musical instrument. Prior studies showed a fine-grained tuning of cortical primary motor, premotor, and parietal regions to motor sequences – from the low-level specification of individual movements to high-level sequence features like sequence order and timing. However, it is not known how tuning in these regions unfolds dynamically across planning and execution. To address this, we trained 24 healthy right-handed human participants (14 females, 10 males) to produce four five-element finger press sequences with a particular finger order and timing structure in a delayed sequence production paradigm entirely from memory. Local cortical fMRI patterns during preparation and production phases were extracted from separate ‘No-Go’ and ‘Go’ trials, respectively, to tease out activity related to these peri-movement phases. During sequence planning, premotor and parietal areas increased tuning to movement order or timing, irrespective of their combinations. In contrast, patterns reflecting the unique integration of sequence features emerged in these regions during execution only, alongside timing-specific tuning in the ventral premotor, supplementary motor, and superior parietal areas. This was in line with the participants’ behavioural transfer of trained timing, but not of order to new sequence feature combinations. Our findings suggest a general informational state shift from high-level feature separation to low-level feature integration within cortical regions for movement execution. Recompiling sequence features trial-by-trial during planning may enable flexible last-minute adjustment before movement initiation.SIGNIFICANCE STATEMENT:Musicians and athletes can modify the timing and order of movements in a sequence trial-by-trial, allowing for a vast repertoire of flexible behaviours. How does the brain put together these high-level sequence features into an integrated whole? We found that, trial-by-trial, the control of sequence features undergoes a state shift from separation during planning to integration during execution across a network of motor-related cortical areas. These findings have implications for understanding the hierarchical control of skilled movement sequences, as well as how information in brain areas unfolds across planning and execution for skilled motor control.

SummaryHumans can learn and retrieve novel skilled movement sequences from memory, yet the content and structure of sequence planning are not well understood. Previous computational and neurophysiological work suggests that actions in a sequence are planned as parallel graded activations and selected for output through competition (competitive queuing; CQ). However, the relevance of CQ during planning to sequence fluency and accuracy, as opposed to sequence timing, is unclear. To resolve this question, we assessed the competitive state of constituent actions behaviourally during sequence preparation. In three separate multi-session experiments, 55 healthy participants were trained to retrieve and produce 4-finger sequences with particular timing from long-term memory. In addition to sequence production, we evaluated reaction time (RT) and error rate increase to constituent action probes at several points during the preparation period. Our results demonstrate that longer preparation time produces a steeper CQ activation and selection gradient between adjacent sequence elements, whilst no effect was found for sequence speed or temporal structure. Further, participants with a steeper CQ gradient tended to produce correct sequences faster and with a higher temporal accuracy. In a computational model, we hypothesize that the CQ gradient during planning is driven by the width of acquired positional tuning of each sequential item, independently of timing. Our results suggest that competitive activation during sequence planning is established gradually during sequence planning and predicts sequence fluency and accuracy, rather than the speed or temporal structure of the motor sequence.HighlightsPre-ordering of actions during sequence planning can be assessed behaviourallyCompetitive gradient reflects sequence preparedness and skill, but not speed or timingGradient is retrieved rapidly and revealed during automatic action selectionPositional tuning of actions boosts the competitive gradient during planning

Humans can learn and produce skilled movement sequences from memory, yet the nature of sequence planning is not well understood. Previous computational and neurophysiological work suggests that movements in a sequence are planned as parallel graded activations and selected for output through competition. However, the relevance of this planning patterns to sequence production fluency and accuracy, as opposed to the temporal structure of sequences, is unclear. To resolve this question, we assessed the relative availability of constituent movements behaviourally during the preparation of motor sequences from memory. In three separate multi-session experiments, healthy participants were trained to retrieve and produce 4-element finger press sequences with particular timing according to an abstract sequence cue. We evaluated reaction time (RT) and error rate as markers of movement availability to constituent movement probes. Our results demonstrate that longer preparation time produces more pronounced differences in availability between adjacent sequence elements, whilst no effect was found for sequence speed or temporal grouping. Further, participants with larger position-dependent differences in movement availability tended to initiate correct sequences faster and with a higher temporal accuracy. Our results suggest that competitive pre-activation during sequence planning is established gradually during sequence planning and predicts sequence skill, rather than the temporal structure of the motor sequence.

Performing sequences of movements from memory and adapting them to changing task demands is a hallmark of skilled human behaviour, from handwriting to playing a musical instrument. Prior studies showed a fine-grained tuning of cortical primary motor, premotor, and parietal regions to motor sequences - from the low-level specification of individual movements to high-level sequence features like sequence order and timing. However, it is not known how tuning in these regions unfolds dynamically across planning and execution. To address this, we trained 24 healthy right-handed participants to produce four five-element finger press sequences with a particular finger order and timing structure in a delayed sequence production paradigm entirely from memory. Local cortical fMRI patterns during preparation and production phases were extracted from separate 'No-Go' and 'Go' trials, respectively, to tease out activity related to these peri-movement phases. During sequence planning, premotor and parietal areas increased tuning to movement order and timing, irrespective of their combinations. In contrast, patterns reflecting the unique integration of sequence features emerged in these regions during execution only, alongside timing-specific tuning in the ventral premotor, supplementary motor, and superior parietal areas. This was in line with the participants' behavioural transfer of trained timing, but not of order to new sequence feature combinations. Our findings suggest a general neural state shift from high-level feature separation to low-level feature integration within cortical regions for movement execution. Recompiling sequence features trial-by-trial during planning may enable flexible last-minute adjustment before movement initiation.