Figure 1

Experimental setup for artificial sensory discrimination using DCS and brain-to-spine interface. (a) Rats were implanted with recording electrodes in motor cortex (M1), somatosensory cortex (S1) and striatum (STR) and dorsal column stimulating electrodes in the thoracic epidural space. Stimulation consisted of charge-balanced, biphasic pulses of 200–1000 µsec pulse duration delivered through an electrical microstimulator controlled by custom Matlab scripts. (b) Behavioral setup for artificial sensory discrimination using DCS consisted of a modified aperture width tactile discrimination box where rats waited in the outer chamber until the sliding door opened. Then they had to poke into a hole in the center of inner chamber. Inside the inner chamber, depending on the type of trial, stimulation pulses were delivered. As the rats came out from the inner chamber they had to poke in the correct reward port associated with the stimulation pattern to get a water reward. (c) Setup for the brain-to-spine interface consisted of two modified aperture width tactile discrimination boxes. While the encoder received DCS pulses inside the inner chamber on its way to the center poke, its post-stimulus neuronal activity was sent to a neural decoder (an adaptive logistic regression classifier) which predicted the number of DCS pulses (between 101 and 1) to be delivered to the decoder’s spinal cord. For the encoder, ‘virtual narrow’ width was rewarded with a response in the ‘left’ reward port while ‘virtual wide’ width was rewarded in the ‘right’ reward port. If the decoder chose the corresponding left or right reward port, based on the translated DCS pulses, the encoder received an additional feedback reward.