Military-service members can suffer brain injury and memory loss when exposed to explosions in enclosed spaces, even if they do not sustain overt physical injury. Credit: Eros Hoagland/Redux/eyevine

A strategy designed to improve memory by delivering brain stimulation through implanted electrodes is undergoing trials in humans. The US military, which is funding the research, hopes that the approach might help many of the thousands of soldiers who have developed deficits to their long-term memory as a result of head trauma. At the Society for Neuroscience meeting in Chicago, Illinois, on 17–21 October, two teams funded by the Defense Advanced Research Projects Agency presented evidence that such implanted devices can improve a person’s ability to retain memories.

By mimicking the electrical patterns that create and store memories, the researchers found that gaps caused by brain injury can be bridged. The findings raise hopes that a ‘neuro­prosthetic’ that automatically enhances flagging memory could aid not only brain-injured soldiers, but also people who have had strokes — or even those who have lost some power of recall through normal ageing.

Because of the risks associated with surgically placing devices in the brain, both groups are studying people with epilepsy who already have implanted electrodes. The researchers can use these electrodes both to record brain activity and to stimulate specific groups of neurons. Although the ultimate goal is to treat traumatic brain injury, these people might benefit as well, says biological engineer Theodore Berger at the University of Southern California (USC) in Los Angeles. That is because repeated seizures can destroy the brain tissue needed for long-term-memory formation.

Short-term memories are thought to be created when a part of the brain called the hippo­campus aggregates sensory information, as well as the perception of space and time, and holds it readily accessible for a short while. Accessing the memory during that time will solidify it into a long-term memory.

Key to this process is a signal that travels from one part of the hippocampus called CA3 to another, called CA1. Berger and his colleagues hypothesize that recreating that signal might restore the ability to solidify memories in people with damage to the hippo­campus.

In one of the studies presented at the Chicago meeting, researchers asked 12 people with epilepsy to look at pictures and then recall up to 90 seconds later which ones they had seen. While the participants did this, the researchers recorded the firing patterns in both CA3 and CA1.

They then developed an algorithm that could use the activity of the CA1 cells to predict the pattern that was coming from CA3. Compared with the actual patterns, their predictions were accurate about 80% of the time.

By using this algorithm, the researchers should be able to stimulate the CA1 cells with a pattern that mimics an appropriate CA3 signal even if a person’s CA3 cells are damaged, Berger says. In previous studies on monkeys trained to do the picture-recall task, receiving a juice reward when correct, his group has shown that stimulating CA1 with an appropriate pattern significantly improved the animals’ performance (R. E. Hampson et al. J. Neural Eng. 10, 066013; 2013).

USC biomedical engineer Dong Song, a member of the team, says that the group has tried the stimulation on a woman with epilepsy, but that it is too early to know whether it has improved her memory. He says that the researchers plan to apply it to more people in the coming months. Eventually, a device might be developed that would detect when the hippo­campus is not efficiently encoding short-term into long-term memory and provide stimulation to support the process.

It is amazing that the memory-formation code can be so accurately predicted, says neuro­biologist Howard Eichenbaum at Boston University in Massachusetts. But he cautions that mimicking it could be difficult if the CA1 cells are so badly damaged that they will not respond properly to stimulation. And he adds that because the hippocampus is so complex and receives inputs from many connections in the brain, stimulating it with the CA3 signal alone may not be enough.

It is amazing that the memory-formation code can be so accurately predicted.

Thomas McHugh, a neuroscientist at the RIKEN Brain Science Institute in Tokyo, says that he has been following the team’s work for years and has been consistently surprised at how well the approach has worked in animal models. “The data is convincing, but I’m still at a loss for understanding,” he says. Many parts of the brain are organized in obvious ways: in the motor cortex, for example, stimulating a particular spot causes motion in a specific part of the body. But there is no such obvious organization in the hippocampus, so it is unclear why stimulating certain locations leads to predictable results.

A team at the University of Pennsylvania (Penn) in Philadelphia is taking a different approach to enhancing memory that requires an even less detailed understanding of how the process works.

The team exploits the fact that people’s memory skills fluctuate over time depending on variables such as how much caffeine they have consumed or whether they are under stress. The team has found, again by working with people with epilepsy, that stimulating a region called the medial temporal lobe, which houses the hippocampus, improves memory that is functioning poorly. But when memory is functioning well, stimulation impedes it.

In a study that they presented at the Chicago meeting, Penn neuroscientist Daniel Rizzuto and his colleagues recorded brain activity in 28 people as they recalled a list of words. Using these patterns, the researchers developed an algorithm that predicted with high accuracy whether a person would remember a given word. By stimulating the brain only when a person read words that were likely to be forgotten, the researchers could boost performance by up to 140%.

Penn psychologist Michael Kahana says that the team has recorded from the brains of about 80 people in total and is seeking regulatory approval to use a more precise electrode array.

Although it would be useful from a basic-science viewpoint to discover why stimulation works so well, McHugh says, it may be worth developing therapies based on it even if it is not fully understood — as long as it can be proved to be safe and effective.