Closed-loop brain computer interfaces
We have developed an electronic device that converts recorded neural or muscular activity to activity-dependent stimulation. This so-called “Neurochip” (NC) is sufficiently small to be carried on a monkey’s head and operates autonomously for many hours of free behavior and sleep . The NC was used to create synaptic plasticity through spike-triggered stimulation. When action potentials recorded at one cortical site triggered stimuli at a second, the connections between them typically strengthened  (Fig. 1). When spikes from CM cells triggered stimuli at spinal target sites, the strength of corticospinal connections was modified according to spike-timing- dependent plasticity rules  (Fig. 2). When cortical stimuli were triggered from cycles of spontaneous beta oscillations the strength of corticocortical connections was transiently modified in a phase-dependent manner  (Fig. 3).
The latest version of this device, Neurochip 3 (NC3, fig. 4), includes 32 recording channels, 6 stimulating channels, and computation provided by FPGA and Atmel ARM processor . Data can be recorded to 64 GB of onboard memory. Current work uses NC3 to investigate mechanisms of cortical plasticity induced by spike-triggered stimulation and by stimuli triggered from beta oscillations. We plan to study the degree to which plasticity depends on behavioral states like sleep and waking. We will also study ways to strengthen and prolong the plasticity by delivering neuromodulators through vagus nerve stimulation and by transcranial direct current stimulation.
Figure 1. Spike-triggered stimulation strengthens corticocortical connections 
Figure 2. Spike-triggered stimulation strengthens corticospinal connections 
Figure 3. Cycle-triggered stimulation changes corticocortical connection 
Figure 4. Neurochip 3