Can neural implants hotwire damaged brain circuits?

Scientists from North Carolina have developed and implemented a neural implant designed to improve higher-order brain processing- providing hope that one day such implants may be developed to alleviate symptoms of cognitive impairments such as Alzheimer’s.

The brain: the final frontier. Neuroscientists around the world are working painstakingly to uncover the mysteries of the brain and ultimately find ways to reinstate functionality lost through damage or disease. One area of research offering such promise is the field of neural prostheses. At a basic level this work endeavors to repair faulty neural connections by interfacing the brain with computer technology. Significant advances have been made in connecting the human brain with the outside world through sensory and motor prostheses (for example the artificial retina). However, the task of restoring or improving cognitive function has yielded very different problems to those addressed in the development of sensory and motor interfaces.

To restore cognitive function, a neural implant must gather data from one area of the brain, process this information correctly and then deliver the resulting signal to another brain region, bypassing any damaged tissue. This process necessitates an understanding of how different brain regions communicate with each other and how this communication is modified as it travels through the brain in the form of electrochemical impulses. Although this is certainly not an easy task, if successful, this research will have a profound impact on the quality of life throughout society, offering promise for people suffering from a range of disorders including Alzheimer’s, stroke and various other forms of brain trauma and degeneration.

Sam Deadwyler and his team at Wake Forest University School of Medicine in Winston-Salem, North Carolina have recently taken an important step toward making cognitive enhancement via neural prostheses a reality. This group is the first to study the effect of brain implants on decision-making in behaving primates. Their work focused on an area of the brain known to be involved in decision-making; the prefrontal cortex. The structure of this region is well-known and can be separated into a number of layers, each containing different types of brain cell. These cells form connections between layers, passing information through the structure. The team developed an implant which could span a number of these layers. This implant was positioned to record activity simultaneously from both layer2/3 and layer 5 and to deliver custom-designed stimulation to layer 5 cells.

The first task faced by the team was to understand how cellular activity changed when the monkey made either a correct or incorrect decision. To achieve this, the group monitored activity across the implant as the monkeys performed a memory task (choosing a familiar image from a group of unfamiliar images). The team focused on cellular recordings taken just prior to the point when the animals made their choice: the decision period. After observing a number of trials, they found that they could predict how cells would respond to both correct and incorrect decisions. This meant that the scientists were able to ‘decode’ the language of the cells and predict what choice would be made before the animal actually made it!

Now that the group knew what a correct response looked like they were able to write a pretty complicated algorithm to mimic this activity and replay it to cells in layer 5. Amazingly they found that animals stimulated with this artificial activity pattern performed significantly better on the task than animals receiving no stimulation. Indeed, in some of the harder tasks (ones using more images) the animals improved both their speed and accuracy, in some cases improving their average performance by 10-20%. This improvement was also dependent on the stimulation provided. For example an improvement would only be seen if a monkey was provided with a ‘correct’ stimulation pattern calculated from its own data, but not when the experimenters used patterns taken from other animals or arbitrary patterns.

The next question was: if this stimulation could improve performance in normal animals, could it also recover the ability to make correct decisions in animals with specific impairments? This is an important question, since the ultimate medical goal for these implants would be to restore lost functionality. To answer this question the team used a drug that they knew reduces connectivity in the prefrontal cortex and impairs decision-making: cocaine. Monkeys given cocaine performed poorly on the task, being on average 10% worse than their sober counterparts. The team was able to monitor layer 2/3 activity in these animals and judge when an incorrect decision was about to be made, then replace the incorrect activity with their own simulated correct firing pattern. This intervention not only restored normal function but actually raised the cocaine-treated animals performance scores even higher than non-treated animals.

It can’t be denied that from both a technical and medical standpoint, these findings are amazing. However, I believe there are still a number of hurdles to be cleared before we see this technology implemented in patients suffering from cognitive deficits. Many cognitive disorders, such as Alzheimer’s, involve widespread damage incorporating a large number of higher-level processing areas. It is therefore an absolute requirement that we first understand exactly how information is processed in these systems, before we attempt to bypass or repair them. Although this implant certainly improved performance, it relies significantly upon mimicking what is known to be a correct response rather than understanding how the system works. This reminds me of my primary school recorder lessons, where I learned to play by watching where other pupils put their fingers instead of learning to read the sheet music. The end result may be similar, but you can go much further and make fewer mistakes if you fully understand the system! Therefore I think the ‘take home’ message from this study is; we have made some promising progress towards improving cognition through neural implantation, however if we ultimately want to treat widespread neural damage we still need to get a better grip on how these systems function before we move toward treatment.

Post by Sarah Fox