ENP alumnus Pierre-Jean Arduin produced a fascinating finding during his PhD. He recorded the activity of several neurons simultaneously and rewarded the animal by moving a water bottle closer to the animal when the firing of a neuron chosen for conditioning passed a specific threshold. During learning and in between trials, this conditioning decreased the variability in the firing rate of the conditioned neuron specifically, showing localized learning. After a learning period, the neuron triggered at trial onset an increase in firing rate, significantly faster and greater than unconditioned neurons around it. This suggests that the neuron chosen to control the brain-machine interface alters its firing and the local cortical network in order to adapt to the interface.
For more details, see the article: Arduin PJ et al. “Master” neurons induced by operant conditioning in rat motor cortex during a brain-machine interface task. J Neurosci. 8 May 2013. 33(19): 8308-8320.
Figure 1B from Arduin et al., J Neurosci, 2013. Experimental setup for training the control neuron. The smoothed firing rate indicates the neuron controlling movement of the water bottle towards the animal.