Neural implant recovers decision-making ability

Researchers have taken a key step towards recovering specific brain functions in sufferers of brain disease and injuries by successfully restoring the decision-making processes in monkeys. By placing a neural device onto the front part of the monkeys' brains, the researchers, from Wake Forest Baptist Medical Centre, University of Kentucky and University of Southern California, were able to recover, and even improve, the monkeys' ability to make decisions when their normal cognitive functioning was disrupted. The study, which has been published today (Sept. 14) in IOP Publishing's Journal of Neural Engineering, involved the use of a neural prosthesis, which consisted of an array of electrodes measuring the signals from neurons in the brain to calculate how the monkeys' ability to perform a memory task could be restored. In the delayed match-to-sample task an image was flashed onto a screen and, after a delay, the monkeys were prompted to select the same image on the screen from a sampling which included 1-7 other images. Five monkeys (all rhesus, Macaca mulatta) were involved in the experiment and were trained for two years to perform to a 70-75 per cent proficiency in the task. The movement of the monkeys' arms were tracked with a camera and translated to movements of the cursor on the screen; they were awarded with a drop of juice when they correctly matched an image. The prosthesis was placed into two cortical layers -- L2/3 and L5 -- of the brain and recorded brain activity within structures known as minicolumns in the prefrontal cortex area. Once it was confirmed that minicolumn communication between layers L2/3 and L5 was involved in decision making, it was suppressed by administering the dopamine-modifying drug, cocaine, to the monkeys. The task was performed again but this time the researchers deployed a 'multi-input multi-output nonlinear' (MIMO) model to stimulate the neurons that were used in the task. "The MIMO model is a specific type of calculation which looks for the complex mathematical relationship between an input (L2/3) and an output (L5). In the case of neural activity, the output is typically the pattern of firing of individual neurons during the task. "Inputs to that pattern may be blood flow, temperature, the electrical activity of other neurons, and even the prior electrical activity of the same cell," said lead author of the study Professor Robert Hampson. By recording the inputs from layer L2/3 neurons, the MIMO model could predict the output of layer L5 neurons and thus, through the electrodes, electrically stimulate the same necessary L5 neurons. The results showed that the MIMO model was exceedingly effective in recovering performance of the task and was even able to improve performance under normal conditions. "The reason the MIMO model was effective in improving performance in the task was because we specifically 'tuned' the model to analyze the firing of neurons that occurred when the animals correctly performed the behavioral task; the brain doesn't always produce the full 'correct' pattern on every trial," said senior author Professor Sam Deadwyler. On the utilization of this method in the treatment of human brain conditions, Professor Deadwyler continued: "In the case of brain injury or disease where larger areas are affected, the system would record the inputs to that area from other areas and, when they occur, program the delivery of the appropriate output patterns to brain regions that normally receive signals from the injured area, thereby restoring lost brain function." All animal procedures were reviewed and approved by the Institutional Animal Care and Use Committee of Wake Forest University, in accordance with U.S. Department of Agriculture, International Association for the Assessment and Accreditation of Laboratory Animal Care, and National Institutes of Health guidelines. From Sciencedaily