Scientific Understanding of Consciousness
Consciousness as an Emergent Property of Thalamocortical Activity

Motor Learning Spatiotemporal Neural Activity Patterns


Nature  510, 263–267 (12 June 2014)

Emergence of reproducible spatiotemporal activity during motor learning

Andrew J. Peters,

Neurobiology Section, Center for Neural Circuits and Behavior, and Department of Neurosciences, University of California, San Diego, La Jolla, California 92093, USA

JST, PRESTO, University of California, San Diego, La Jolla, California 92093, USA


The motor cortex is capable of reliably driving complex movements, yet exhibits considerable plasticity during motor learning. These observations suggest that the fundamental relationship between motor cortex activity and movement may not be fixed but is instead shaped by learning; however, to what extent and how motor learning shapes this relationship are not fully understood. Here we addressed this issue by using in vivo two-photon calcium imaging to monitor the activity of the same population of hundreds of layer 2/3 neurons while mice learned a forelimb lever-press task over two weeks. Excitatory and inhibitory neurons were identified by transgenic labelling. Inhibitory neuron activity was relatively stable and balanced local excitatory neuron activity on a movement-by-movement basis, whereas excitatory neuron activity showed higher dynamism during the initial phase of learning. The dynamics of excitatory neurons during the initial phase involved the expansion of the movement-related population which explored various activity patterns even during similar movements. This was followed by a refinement into a smaller population exhibiting reproducible spatiotemporal sequences of activity. This pattern of activity associated with the learned movement was unique to expert animals and not observed during similar movements made during the naive phase, and the relationship between neuronal activity and individual movements became more consistent with learning. These changes in population activity coincided with a transient increase in dendritic spine turnover in these neurons. Our results indicate that a novel and reproducible activity–movement relationship develops as a result of motor learning, and we speculate that synaptic plasticity within the motor cortex underlies the emergence of reproducible spatiotemporal activity patterns for learned movements. These results underscore the profound influence of learning on the way that the cortex produces movements.

To identify how the activity of motor cortex neuronal ensembles is modified during this learning, we combined the lever-press task with chronic two-photon calcium imaging. In this study we focused on neurons in layer 2/3, the major input layer capable of driving deeper layer neurons to produce motor cortex outputs.

We imaged the activity of the same population of excitatory and inhibitory neurons over the course of 2weeks while mice simultaneously learned and performed the lever-press task (n = 7 mice). High correlations of neural activity and lever-press movements were evident both at the level of single neurons and population average in both excitatory and inhibitory neurons. Movement-related neurons did not show obvious spatial clustering.

Individual inhibitory neurons were particularly correlated with nearby excitatory neurons within 150μm, consistent with their connectivity. This local matching of excitatory and inhibitory activity probably provides a basis for the balance between excitatory and inhibitory inputs to individual neurons observed in the cortex. The identity of movement-related neurons was dynamic. In particular, excitatory neurons on average had a higher degree of turnover than the inhibitory population, indicating that the excitatory population is more dynamic during learning.

Many excitatory neurons were transiently movement-related. Various combinations of excitatory neurons were used within the motor cortex during the initial phase of learning, followed by a refinement of the population to form a stable network associated with the learned movement.

The temporal activity pattern became progressively more stable during learning. The population activity shifted towards the beginning of movements over the course of the experiment.

The learned activity pattern was reproducibly observed only when the expert animals made the learned movement, whereas similar movements made in naive sessions were accompanied by very different activity patterns. Furthermore, the general relationship between activity and movement in pairs of trials became more consistent after learning, whereas activity in naive animals was more variable regardless of movement similarity. These analyses of learning-related changes in population activity were performed using the entire movement periods.

The plasticity of population activity could simply reflect changes in other brain areas providing inputs to these neurons. However, synaptic plasticity within the motor cortex could also contribute to the changes in population activity. To test whether learning of this lever-press task induces synaptic plasticity in layer 2/3 of the motor cortex, we labelled sparse subsets of layer 2/3 neurons and chronically imaged spines on the same dendritic branches of excitatory neurons over the course of learning (n = 191 spines in 3 mice). Imaging was performed immediately before each training session in awake animals. We observed the formation of a number of dendritic spines during the initial sessions of training, followed by the elimination of some of the spines that were present at the beginning of the experiment. Most (95%, 19 out of 20) of the spines that formed during training persisted for the entire 2weeks. At the population level, these changes resulted in a transient 10% increase in the density of spines followed by return to the baseline. These results are analogous to previous reports in motor cortex layer 5 neurons during different motor learning tasks. Spines were largely stable in a separate group of animals that did not undergo training but otherwise were treated identically including water restriction and head fixation. The spine density was also stable in the hindlimb area in the motor cortex during learning. These results indicate that our lever-press task induces area-specific reorganizations of excitatory synapses onto layer 2/3 neurons during learning.

Our results indicate that the relationship between movements and activity is initially degenerate, and the early days of learning involve the expansion and exploration of various movement-related activity in the motor cortex. An increased variability of single-neuron activity in the motor cortex has been observed during learning of visuomotor adaptation and a brain–machine interface task. Such trial-to-trial variability has been proposed to provide the basis for exploration of possible network states and facilitate learning. Our results directly demonstrate such an exploration during initial learning at the population level. After this period of high variability, the activity–movement degeneracy is reduced and a reproducible temporal sequence of activity emerges in a stable population of excitatory neurons. Such spatiotemporal activity may orchestrate the temporal dynamics of the learned movement. Reproducible temporal patterns of population activity during learned movements are proposed to be generated by internal connections within the motor cortex. We note that our results do not provide a causal link between local synaptic plasticity and changes in population activity. Nevertheless, we show that these processes occur during motor learning on similar timescales, which supports the notion that local synaptic plasticity may generate a circuit to reproduce a particular spatiotemporal activity pattern. These new circuits may be more efficient in driving movements, which could underlie the lower metabolic activity in the motor cortex observed during execution of well-practiced movements. Our study provides a glimpse of the emergence of population activity patterns for learned movements.

[end of paraphrase]


Return to — Movement Control

Return to — Fixed Action Patterns