Neural implants could spark better decisions

18:00 19 September 2012 by Douglas Heaven Magazine issue 2883.

Ever wish you could make better choices? That could one day be possible thanks to an electronic brain implant that can enhance short-term memory and decision-making in primates. The implant can also restore these functions in an animal model of Alzheimer’s disease and other types of brain damage, paving the way for the development of new treatments for people with these conditions.

Sam Deadwyler at Wake Forest University School of Medicine in Winston-Salem, North Carolina, and colleagues have previously shown that a neural implant can restore some motor and sensory functions in rats. Now they have used a similar implant to stimulate higher-level thinking in monkeys.

During normal brain function, neurons “fire” when they receive an input from another neuron via the connection between them, called a synapse. The spatial and temporal pattern of this activity  – where and when the neurons fire   – can be detected and recorded.

To find out if it is possible to hijack and then retune these patterns of activity, Deadwyler’s team first trained five rhesus macaques to perform a task that tests their attention, short-term memory and decision-making skills.

First, the monkeys were shown a random image from a pool of 5000. The image was then blanked out for an interval of 1 to 90 seconds, before reappearing in a different position, alongside up to seven other images. If the monkey selected the original image once it reappeared it was rewarded with juice.

After two years of training, the monkeys correctly matched the images 75 per cent of the time at the easier levels  – when the image was blanked out for a short period of time, and fewer images were displayed when it reappeared. Their success rate declined to 40 per cent in tasks with the longest interval and maximum amount of images to choose from.

The animals then had a device capable of recording and stimulating neuronal activity implanted into their prefrontal cortex, an area of the brain responsible for many facets of intelligence. Specifically, the implant was able to record neuronal communication in the so-called supra-granular layers (layers 2/3), part of the six layers that make up this part of the cortex, and both record from and stimulate neurons in layer 5 (see diagram). Layers 2/3 and 5 are around 1350 micrometres apart and differ in their predominant cell type and connectivity.

The challenge lay in working out which patterns of activity should be induced and when in order to enhance the monkeys’ performance in the task. To do so, the team recorded neuronal activity while the animals performed the task again. They then analysed the activity going into and out of the different layers and extracted patterns of neuronal firing that correlated with correct and incorrect decisions.

The team was then able to enhance the animals’ decision-making process by ensuring that the implant kicked in whenever the neuronal activity in layers 2/3 resembled that seen when an incorrect decision was being made. When it did, the implant stimulated a pattern of activity in layer 5 that replicated the activity recorded when a correct decision was made.

The implant was able to improve average performance in the task by 10 to 20 per cent. In the hardest versions of the task – such as when the original image reappeared alongside several other images  – the monkeys also significantly increased their speed, taking 2 rather than 3 seconds to correctly identify the image.

The team next tested the implant’s ability to restore cognitive function in monkeys that had been given cocaine. The drug is commonly used in animals to model the loss of synaptic connections seen in Alzheimer’s, dementia and other types of brain injury.

Without stimulation from the implant, the monkeys’ performance dropped by 10 per cent across all difficulty levels. When the implant was switched on, however, their performance was boosted to just below levels seen in monkeys who hadn’t been given cocaine or an implant.

“It’s a wonderful piece of work,” says Daniel Chew at the University of Cambridge, who was not involved in the study. He suggests that since the implant reduces the number of incorrect responses there may be an even greater degree of improvement on a more difficult task.

Simon Schultz at Imperial College London agrees that it is an impressive study, but says that it is not clear what exactly is going on. We understand the motor and sensory domains quite well, he says, but we still don’t know how the cortex works. These guys step around that, he says, by effectively recognising what a correct decision looks like, recording that pattern and playing it back.

In the study, the patterns of activity used to stimulate a correct answer were specific to each monkey. Would it be possible to use a pattern taken from one high-performing individual and use it to raise the game of others? According to team member Robert Hampson, giving one monkey a “correct” pattern of activity from another didn’t work, and in fact reduced performance, just as scrambled patterns did.

To apply this technology to people with conditions such as Alzheimer’s, this limitation would have to be overcome, either by learning much more about how a correct pattern is shaped from inputs into each area, or recording healthy brain activity before a person developed symptoms of brain damage.

A further problem is the invasive nature of implants. “It causes very acute inflammation and scarring,” says Chew. This can kill the neurons around the implant, insulating the electrodes and diminishing their effectiveness. Chew and his colleagues are trying to create biologically compatible electrodes to get around this problem.

Another possibility is bypassing the need for an implant entirely. Deadwyler and Hampson both point out that transcranial imaging and stimulation – the ability to visualise and stimulate activity in the brain non-invasively – is advancing quickly and that it may one day be possible to adapt their approach to work from outside the skull.

Regardless, the pair believe that human trials of the implant are in sight since similar hardware has already been approved by the US Food and Drug Administration for use in people with Parkinson’s. “[Human trials] may not be too far down the line,” says Deadwyler.

This is great news for people with brain deterioration  – the likely first participants of such a trial. But besides therapeutic treatment, the possibilities are endless. Imagine an implant in your visual cortex, says Schultz. This, he suggests, would make it possible to identify the pattern of activity that occurs when you imagine a number. When this pattern was recognised by the implant it could stimulate the pattern for a new number for you to picture. In theory, such an implant could be used to create a secure code for a cryptographic key based on values only you can access.

The problem with developing a prosthesis for cognitive enhancement rather than restoration is that it is harder to justify the trials. But, as Schultz jokes, “why stop at repair when you can enhance as well?” It’s a nice idea, but the ethical hurdles mean that developing a prosthesis for cognitive enhancement rather than restoration is not currently justifiable, he says. For the time being, at least, the focus will rightly be on ways to restore lost function in people with brain damage. And that’s surely a good decision.

Journal reference: Journal of Neural Engineering, doi.org/jcx

http://www.newscientist.com/article/dn22282-neural-implants-could-spark-better-decisions.html?full=true&print=true