Many hurdles have to be negotiated, of course. Despite all we have learned about the hippocampus since HM, it still remains something of a black box whose inner workings are largely unknown. As a result, it is not possible to construct a memory from scratch, but once a task has been performed and the memory processed, it is possible to record it and play it back.
FUTURE DIRECTIONS
Working with the hippocampus of primates and even humans will be more difficult, since their hippocampi are much larger and more complex. The first step is to create a detailed neural map of the hippocampus. This means placing electrodes at different parts of the hippocampus to record the signals that are constantly being exchanged between different regions. This will establish the flow of information that constantly moves across the hippocampus. The hippocampus has four basic divisions, CA1 to CA4, and hence scientists will record the signals that are exchanged between them.
The second step involves the subject performing certain tasks, after which scientists will record the impulses that flow across the various regions of the hippocampus, thereby recording the memory. For example, the memory of learning a certain task, such as jumping through a hoop, will create electrical activity in the hippocampus that can be recorded and carefully analyzed. Then a dictionary can be created matching the memory with the flow of information across the hippocampus.
Finally, step three involves making a recording of this memory and feeding the electrical signal into the hippocampus of another subject via electrodes, to see if that memory can been uploaded. In this fashion, the subject may learn to jump through a hoop although it has never done so before. If successful, scientists would gradually create a library containing recordings of certain memories.
It may take decades to work all the way up to human memories, but one can envision how it might work. In the future, people may be hired to create certain memories, like a luxury vacation or a fictitious battle. Nanoelectrodes will be placed at various places in their brain to record the memory. These electrodes must be extremely small so that they do not interfere with the formation of the memory.
The information from these electrodes will then be sent wirelessly to a computer and then recorded. Later a subject who wants to experience these memories will have similar electrodes placed in his hippocampus, and the memory will be inserted into the brain.
(There are complications to this idea, of course. If we try to insert the memory of physical activity, such as a martial art, we have the problem of “muscle memory.” For example, when walking, we do not consciously think about putting one leg in front of the other. Walking has become second nature to us because we do it so often, and from an early age. This means that the signals controlling our legs no longer originate entirely in the hippocampus, but also in the motor cortex, the cerebellum, and the basal ganglia. In the future, if we wish to insert memories involving sports, scientists may have to decipher the way in which memories are partially stored in other areas of the brain as well.)
VISION AND HUMAN MEMORIES
The formation of memories is quite complex, but the approach we have been discussing takes a shortcut by eavesdropping on the signals moving through the hippocampus, where the sensory impulses have already been processed. In The Matrix, however, an electrode is placed in the back of the head to upload memories directly into the brain. This assumes that one can decode the raw, unprocessed impulses coming in from the eyes, ears, skin, etc., that are moving up the spinal cord and brain stem and into the thalamus. This is much more elaborate and difficult than analyzing the processed messages circulating in the hippocampus.
To give you a sense of the sheer volume of unprocessed information that comes up the spinal cord into the thalamus, let’s consider just one aspect: vision, since many of our memories are encoded this way. There are roughly 130 million cells in the eye’s retina, called cones and rods; they process and record 100 million bits of information from the landscape at any time.
This vast amount of data is then collected and sent down the optic nerve, which transports 9 million bits of information per second, and on to the thalamus. From there, the information reaches the occipital lobe, at the very back of the brain. This visual cortex, in turn, begins the arduous process of analyzing this mountain of data. The visual cortex consists of several patches at the back of the brain, each of which is designed for a specific task. They are labeled V1 to V8.
Remarkably, the area called V1 is like a screen; it actually creates a pattern on the back of your brain very similar in shape and form to the original image. This image bears a striking resemblance to the original, except that the very center of your eye, the fovea, occupies a much larger area in V1 (since the fovea has the highest concentration of neurons). The image cast on V1 is therefore not a perfect replica of the landscape but is distorted, with the central region of the image taking up most of the space.
Besides V1, other areas of the occipital lobe process different aspects of the image, including:
• Stereo vision. These neurons compare the images coming in from each eye. This is done in area V2.
• Distance. These neurons calculate the distance to an object, using shadows and other information from both eyes. This is done in area V3.
• Colors are processed in area V4.
• Motion. Different circuits can pick out different classes of motion, including straight-line, spiral, and expanding motion. This is done in area V5.
More than thirty different neural circuits involved with vision have been identified, but there are probably many more.
From the occipital lobe, the information is sent to the prefrontal cortex, where you finally “see” the image and form your short-term memory. The information is then sent to the hippocampus, which processes it and stores it for up to twenty-four hours. The memory is then chopped up and scattered among the various cortices.
The point here is that vision, which we think happens effortlessly, requires billions of neurons firing in sequence, transmitting millions of bits of information per second. And remember that we have signals from five sense organs, plus emotions associated with each image. All this information is processed by the hippocampus to create a simple memory of an image. At present, no machine can match the sophistication of this process, so replicating it presents an enormous challenge for scientists who want to create an artificial hippocampus for the human brain.
REMEMBERING THE FUTURE
If encoding the memory of just one of the senses is such a complex process, then how did we evolve the ability to store such vast amounts of information in our long-term memory? Instinct, for the most part, guides the behavior of animals, which do not appear to have much of a long-term memory. But as neurobiologist Dr. James McGaugh of the University of California at Irvine says, “The purpose of memory is to predict the future,” which raises an interesting possibility. Perhaps long-term memory evolved because it was useful for simulating the future. In other words, the fact that we can remember back into the distant past is due to the demands and advantages of simulating the future.
Indeed, brain scans done by scientists at Washington University in St. Louis indicate that areas used to recall memories are the same as those involved in simulating the future. In particular, the link between the dorsolateral prefrontal cortex and the hippocampus lights up when a person is engaged in planning for the future and remembering the past. In some sense, the brain is trying to “recall the future,” drawing upon memories of the past in order to determine how something will evolve into the future. This may also explain the curious fact that people who suffer from amnesia—such as HM—are often unable to visualize what they will be doing in the future or even the very next day.
“You might look at it as mental time travel—the ability to take thoughts about ourselves and project them either into the past or into the future,” says Dr. Kathleen McDermott of Washington University. She also notes that their study proves a “tentative answer to a longstanding question regarding the evol
utionary usefulness of memory. It may just be that the reason we can recollect the past in vivid detail is that this set of processes is important for being able to envision ourselves in future scenarios. This ability to envision the future has clear and compelling adaptive significance.” For an animal, the past is largely a waste of precious resources, since it gives them little evolutionary advantage. But simulating the future, given the lessons of the past, is an essential reason why humans became intelligent.
AN ARTIFICIAL CORTEX
In 2012 the same scientists from Wake Forest Baptist Medical Center and the University of Southern California who created an artificial hippocampus in mice announced an even more far-reaching experiment. Instead of recording a memory in the mouse hippocampus, they duplicated the much more sophisticated thinking process of the cortex of a primate.
They took five rhesus monkeys and inserted tiny electrodes into two layers of their cortex, called the L2/3 and L5 layers. They then recorded neural signals that went between these two layers as the monkeys learned a task. (This task involved the monkeys seeing a set of pictures, and then being rewarded if they could pick out these same pictures from a much larger set.) With practice, the monkeys could perform the task with 75 percent accuracy. But if the scientists fed the signal back into the cortex as the monkey was performing the test, its performance increased by 10 percent. When certain chemicals were given to the monkey, its performance dropped by 20 percent. But if the recording was fed back into the cortex, its performance exceeded its normal level. Although this was a small sample size and there was only a modest improvement in performance, the study still suggests that the scientists’ recording accurately captured the decision-making process of the cortex.
Because this study was done on primates rather than mice and involved the cortex and not the hippocampus, it could have vast implications when human trials begin. Dr. Sam A. Deadwyler of Wake Forest says, “The whole idea is that the device would generate an output pattern that bypasses the damaged area, proving an alternative connection” in the brain. This experiment has a possible application for patients whose neocortex has been damaged. Like a crutch, this device would perform the thinking operation of the damaged area.
AN ARTIFICIAL CEREBELLUM
It should also be pointed out that the artificial hippocampus and neocortex are but the first steps. Eventually, other parts of the brain will have artificial counterparts. For example, scientists at Tel Aviv University in Israel have already created an artificial cerebellum for a rat. The cerebellum is an essential part of the reptilian brain that controls our balance and other basic bodily functions.
Usually when a puff of air is directed at a rat’s face, it blinks. If a sound is made at the same time, the rat can be conditioned to blink just by hearing the sound. The goal of the Israeli scientists was to create an artificial cerebellum that could duplicate this feat.
First the scientists recorded the signals entering the brain stem when the puff of air hit the rat’s face and the sound was heard. Then the signal was processed and sent back to the brain stem at another location. As expected, the rats blinked upon receiving the signal. Not only is this the first time that an artificial cerebellum functioned correctly, it is the first time that messages were received from one part of the brain, processed, and then uploaded into a different part of the brain.
Commenting on this work, Francesco Sepulveda of the University of Essex says, “This demonstrates how far we have come towards creating circuitry that could one day replace damaged brain areas and even enhance the power of the healthy brain.”
He also sees great potential for artificial brains in the future, adding, “It will likely take us several decades to get there, but my bet is that specific, well-organized brain parts such as the hippocampus or the visual cortex will have synthetic correlates before the end of the century.”
Although progress in creating artificial replacements for the brain is moving remarkably fast given the complexity of the process, it is a race against time when one considers the greatest threat facing our public health system, the declining mental abilities of people with Alzheimer’s.
ALZHEIMER’S—DESTROYER OF MEMORY
Alzheimer’s disease, some people claim, might be the disease of the century. There are 5.3 million Americans who currently have Alzheimer’s, and the number is expected to quadruple by 2050. Five percent of people from age sixty-five to seventy-four have Alzheimer’s, but more than 50 percent of those over eighty-five have it, even if they have no obvious risk factors. (Back in 1900, life expectancy in the United States was forty-nine, so Alzheimer’s was not a significant problem. But now, people over eighty are one of the fastest-growing demographic groups in the country.)
In the early stages of Alzheimer’s, the hippocampus, the part of the brain through which memories are processed, begins to deteriorate. Indeed, brain scans clearly show that the hippocampus shrinks in Alzheimer’s patients, but the wiring linking the prefrontal cortex to the hippocampus also thins, leaving the brain unable to properly process short-term memories. Long-term memories already stored throughout the cortices of the brain remain relatively intact, at least at first. This creates a situation where you may not remember what you just did a few minutes ago but can clearly recall events that took place decades ago.
Eventually, the disease progresses to the point where even basic long-term memories are destroyed. The person is unable to recognize their children or spouse and to remember who they are, and can even fall into a comalike vegetative state.
Sadly, the basic mechanisms for Alzheimer’s have only recently begun to be understood. One major breakthrough came in 2012, when it was revealed that Alzheimer’s begins with the formation of tau amyloid proteins, which in turn accelerates the formation of beta amyloid, a gummy, gluelike substance that clogs up the brain. (Before, it was not clear if Alzheimer’s was caused by these plaques or whether perhaps these plaques were by-products of a more fundamental disorder.)
What makes these amyloid plaques so difficult to target with drugs is that they are most likely made of “prions,” which are misshapen protein molecules. They are not bacteria or viruses, but nevertheless they can reproduce. When viewed atomically, a protein molecule resembles a jungle of ribbons of atoms tied together. This tangle of atoms must fold onto itself correctly for the protein to assume the proper shape and function. But prions are misshapen proteins that have folded incorrectly. Worse, when they bump into healthy proteins, they cause them to fold incorrectly as well. Hence one prion can cause a cascade of misshapen proteins, creating a chain reaction that contaminates billions more.
At present, there is no known way to stop the inexorable progression of Alzheimer’s. Now that the basic mechanics behind Alzheimer’s are being unraveled, however, one promising method is to create antibodies or a vaccine that might specifically target these misshapen protein molecules. Another way might be to create an artificial hippocampus for these individuals so that their short-term memory can be restored.
Yet another approach is to see if we can directly increase the brain’s ability to create memories using genetics. Perhaps there are genes that can improve our memory. The future of memory research may lie in the “smart mouse.”
THE SMART MOUSE
In 1999, Dr. Joseph Tsien and colleagues at Princeton, MIT, and Washington University found that adding a single extra gene dramatically boosted a mouse’s memory and ability. These “smart mice” could navigate mazes faster, remember events better, and outperform other mice in a wide variety of tests. They were dubbed “Doogie mice,” after the precocious character on the TV show Doogie Howser, M.D.
Dr. Tsien began by analyzing the gene NR2B, which acts like a switch controlling the brain’s ability to associate one event with another. (Scientists know this because when the gene is silenced or rendered inactive, mice lose this ability.) All learning depends on NR2B, because it controls the communication between memory cells of the hippocampus. First Dr. Tsien created a
strain of mice that lacked NR2B, and they showed impaired memory and learning disabilities. Then he created a strain of mice that had more copies of NR2B than normal, and found that the new mice had superior mental capabilities. Placed in a shallow pan of water and forced to swim, normal mice would swim randomly about. They had forgotten from just a few days before that there was a hidden underwater platform. The smart mice, however, went straight to the hidden platform on the first try.
Since then, researchers have been able to confirm these results in other labs and create even smarter strains of mice. In 2009, Dr. Tsien published a paper announcing yet another strain of smart mice, dubbed “Hobbie-J” (named after a character in Chinese cartoons). Hobbie-J was able to remember novel facts (such as the location of toys) three times longer than the genetically modified strain of mouse previously thought to be the smartest. “This adds to the notion that NR2B is a universal switch for memory formation,” remarked Dr. Tsien. “It’s like taking Michael Jordon and making him a super Michael Jordan,” said graduate student Deheng Wang.
There are limits, however, even to this new mice strain. When these mice were given a choice to take a left or right turn to get a chocolate reward, Hobbie-J was able to remember the correct path for much longer than the normal mice, but after five minutes he, too, forgot. “We can never turn it into a mathematician. They are rats, after all,” says Dr. Tsien.
It should also be pointed out that some of the strains of smart mice were exceptionally timid compared to normal mice. Some suspect that, if your memory becomes too great, you also remember all the failures and hurts as well, perhaps making you hesitant. So there is also a potential downside to remembering too much.