Until recently, the following cascade of events was thought to underlie skill learning (Monfils et al. 2005). Motor training induces synaptic plasticity through neural signaling, gene transcription and protein synthesis. Synaptic plasticity consists in change of excitability of some neurons in order to make them either more likely (long-term potentiation) or more unlikely (long-term depression) to fire. These changes in synaptic plasticity make some specific networks stronger and lead to some reorganization of the local circuitry (motor map reorganization). As a result of this reorganization, microstimulation of the motor cortex results in different movement patterns than before the learning of a skill. This change in local circuitry also involves an increase in the number of synapses (and spines) per neurons, a phenomenon called synaptogenesis. Synaptogenesis corresponds to the formation of new connections between some pairs of neurons and was thought to occur late during the training, i.e. when the performance is already high and the number of errors low.
This increase of synaptogenesis was specific to learning of task for the very first time as rats did not show any further increase in spine formation when they were retrained on the same task after a few months of non-practice. However, if a second task was introduced after this period of non-practice, an increase in spine formation was again observed. This shows that learning a new skill for the very first time engages different processes than re-learning a skill.
Finally, the authors also demonstrate that the spines formed early during the learning were more stable than the spines formed later during the learning and that spine formation and elimination was more present in adolescent than in adult mice (although they did not comment on that point).
This study sheds some light on some aspects of the skill learning or motor learning literature like the phenomenon of savings (Krakauer et al.,1999). Savings is observed in visuomotor rotation tasks where subjects are required to reach to targets while the visual feedback is rotated by a given angle. For instance, if this angle is 90deg, the cursor displayed on the screen will move upward when the subject moves his hand rightward. In this task, the subjects will always reduce their errors faster during the second session than during the first one, even if months have passed. This phenomenon is referred to as saving. The observation that the network will never be the same once you've learned a new task might explain the saving effect in visuomotor rotation tasks.
Offline processes take place between two practice sessions (Robertson et al, 2004). The imaging after the first training session allows us to assess what happened during the first session. Unfortunately, the second imaging does not allow us to assess whether the changes observed after the second training session were due to the training session itself or to the offline processes that took place since the end of the first session.
Krakauer, J. W., Ghilardi, M. F., & Ghez, C. (1999). Independent learning of internal models for kinematic and dynamic control of reaching. Nature neuroscience, 2(11), 1026-31. doi: 10.1038/14826.
Monfils, M., Plautz, E. J., & Kleim, J. a. (2005). In search of the motor engram: motor map plasticity as a mechanism for encoding motor experience. The Neuroscientist : a review journal bringing neurobiology, neurology and psychiatry, 11(5), 471-83. doi: 10.1177/1073858405278015.
Robertson, E. M., Pascual-Leone, A., & Miall, R. C. (2004). Current concepts in procedural consolidation. Nature Reviews Neuroscience, 5(7), 576-82. doi: 10.1038/nrn1426.
Xu, T., Yu, X., Perlik, A. J., Tobin, W. F., Zweig, J. a., Tennant, K., et al. (2009). Rapid formation and selective stabilization of synapses for enduring motor memories. Nature, 462(7275), 915-9. Nature Publishing Group. doi: 10.1038/nature08389.