Hwang and colleagues provide evidence in favor of the structure hypothesis. In a first experiment, the activity of a single neuron (trained neuron) from PRR is used to control cursor motion. In a second experiment, cursor motion is controlled by the firing rate of two neurons (trained neurons) at the same time. In both experiments, both trained and untrained neurons adapt their firing rate when the relationship between the neural activity of trained neurons and cursor motion is perturbed. In addition, the change in firing rate of the trained neurons correlates with the change in firing rates of the untrained neurons.
Overall, this study suggests that modulation of neural activity within the PRR for learning the control of cursor on a screen harvests the pre-existing structure of the PRR neural network. It suggests that these pre-existing structures bias subsequent learning. Therefore, these pre-existing structures act as priors for the learning. This harvesting of existing structure has also been found behaviorally, during force-field learning (Sing et al., 2009).
1. Hwang, E. J., Bailey, P. M. & Andersen, R. a Volitional Control of Neural Activity Relies on the Natural Motor Repertoire. Current biology 23, 353–361 (2013).
2. Sing, G. C., Joiner, W. M., Nanayakkara, T., Brayanov, J. B. & Smith, M. A. Primitives for motor adaptation reflect correlated neural tuning to position and velocity. Neuron 64, 575–89 (2009).