blank chalkboard carries no information, but as we begin to scribble facts and figures on it it can serve as the focal point for lectures and lessons beyond number. But imagine that we have no eraser. Each new lesson has to be scrawled in around the margins of the old, making the board harder and harder to read, until finally it's completely white. At this point, as in the beginning, no information can be extracted from it. For similar reasons, Mnemosyne, the Greek goddess of memory, has a delicate job to perform inside our brains, keeping our mental chalkboards from getting too black or too white, too patchy or sparse. Somewhere in the happy middle, our useful memories are maximized.
In John Woo's new movie, Paycheck (based on a short story by hot property Philip K. Dick), a man named Michael Jennings (Ben Affleck) undergoes a neurochemical process to erase selected memories from his brain. This idea has cropped up many times in science fiction, from The Sirens of Titan to The Simpsons, and forms the basis of many an exciting mystery plot. But just how realistic is it? Can memories really be cut and pasted and spliced like so much celluloid film?
Certainly, forgetting is a part of life. The ability to forget traumatic experiences may be an adaptation to avoid unnecessary anxiety, while the forgetting of ordinary experiences keeps us grounded in the present, and focuses our attention on what's important here and now. Like the blue haze obscuring a distant mountain, the very murkiness of the past keeps us from mistaking it for the present. As James Cameron warns us in Strange Days, "Memories are supposed to fade. They were made that way for a reason."
Thanks for the short-term memories
But what's going on there? Just what is memory? The answer appears to be twofold: our brains employ separate systems—long-term and short-term memory—which are roughly equivalent to the RAM and hard disk on your computer. Very little is known about short-term memories, but they're probably electrochemical in nature, since they occur on a time scale too short for any other brain process to operate.
Short-term memory can hold seven to 10 "things" (words, digits, objects in a visual scene, etc.) for around 30 seconds. This buffer can be continually refreshed, as when you recite a phone number over and over while you hunt for a phone, but it can also be destroyed by interruption. When this happens, the information is lost beyond any power of retrieval or recall. It simply isn't stored.
By contrast, long-term memory involves physical rewiring of the axons, or connections between brain cells, and can persist for a lifetime. These memories are stored in various regions throughout the brain, but are transcribed from the short-term buffer by a bit of tissue called the hippocampus. Damage to this region—whether from trauma, alcohol abuse, tumors, infections or vitamin deficiency—results in a disorder called Korsakoff's syndrome, which leaves older memories intact in the brain but prevents any new ones from being laid down. Guy Pearce, the protagonist in Christopher Nolan's chilling film Memento, suffers from an extreme case of Korsakoff's resulting from a blow to the head. You don't have to go that far to create a memory hole, though. The hippocampus is particularly active during sleep, and the encoding of long-term memory can be temporarily suppressed through drugs or alcohol (the proverbial "lost weekend") or even simple sleep deprivation.
Long-term memories are stored not as facts or sensory recordings but as associations between these things. As such, they're actually rather sparse, and our recollection of them is as much a process of imagination as it is a playback of stored data. Recall is triggered when our attention is drawn to the impressions or experiences we've related to a particular memory. Practice aids recall by beefing up these associations, or adding new ones. The simple act of remembering something—dredging it up from the murk of forgetfulness—is sometimes all the reinforcement we need. This is why studying works. Without reinforcement, though, the associations weaken, and memories become difficult to recall. They're still in there somewhere, but we've lost track of the keys to unlock them. This is why cramming for final exams may get you a good grade, but little enduring knowledge. It's also why the details of childhood can be so difficult to recall.
Still, recognition is memory's easiest chore. Even patients without a functioning hippocampus seem, over time, to develop a familiarity with people and places and tasks. Possibly, these associations are not "memories" at all in the conscious sense. We find the cerebral task of recall—arguably one of the things that set human beings apart from other animals—far more taxing. This is why, for example, true/false or multiple choice tests are much easier than essay questions, and picking a face out of a police lineup is easier than describing it to a sketch artist. Recognition can also be a major aid to recall, as when a smell triggers long-buried childhood memories.
In the absence of these triggering cues, we gradually lose the things we once knew. Abstract information is the first to go, then sounds, sights, and finally smells. New memories may also interfere with old ones (the "you remind me of Joe" effect), and vice versa (the "that's not what I learned in school" effect), although new information usually has the weakest associations. In all of these cases, though, simple things are remembered longer and better than complex ones, while new, exciting, sensually rich experiences are recalled better than boring or familiar ones. But none of these forgetting processes can explain the complete erasure of a specific recent event. For that, we have to invoke two additional mechanisms: amnesia and repression.
The unforgettable orders of disorders
Amnesia is a brain disorder resulting from injury, shock, fatigue, drugs or illness. It's sometimes associated with dementia (as in Alzheimer's disease and other senility), though any soap opera can tell you it occurs in clear minds as well. Retrograde amnesia, or loss of memory for older events, is the most typical form, although global amnesia—the loss of all episodic memories, plus an inability to store new ones—is also common. In both cases, general memories such as language and motor skills are not affected. Patients recovering from concussion will often experience global amnesia which gradually recedes over a period of minutes or hours, leaving a permanent memory gap for the 45 minutes or so preceding the injury.
(This is a clue about how long the hippocampus takes to do its thing, and suggests that there may be an "intermediate-term memory" as well, when the original chemical patterns have dispersed but the nerve pathways are not yet fully burned in.)
There's also a disorder called "fugue state," in which a person suddenly exhibits little or no memory of their prior life, and may wander off or behave in uncharacteristic ways—even to the extent of assuming a new identity—until the fugue passes. Afterward, there are no memories of the fugue state itself. However, based on these symptoms, this condition may have more in common with multiple personality disorder than with amnesia per se, so we'll disregard it as an example of memory erasure.
By contrast, repression (also known as hysterical amnesia, inorganic amnesia or motivated forgetting) is the incomplete but highly selective forgetting of specific events—usually traumatic in nature. Repressed memories are not available to the conscious mind, but continue to inform the subconscious, and thus affect behavior. There is some controversy surrounding this notion after a number of ugly legal battles, but even so it's generally agreed that repression does occur in many people, and that with effort the repressed memories can be consciously recovered.
In a physical sense, I suspect these memories are re-routed through the amygdala—the brain's central switchboard for strong emotion. This almond-shaped region near the stem of the brain can feed feelings and vague impressions up to the conscious mind, and can even hijack the body for fight-or-flight activity over which the neocortex—the most conscious portion of the brain—has no control. We call this sensation "panic," and marvel afterwards at how disconnected we felt from our own bodies. If we understood the organic mechanisms of repression, we might be able to induce them artificially, and so delete a particular experience.
Still, this would merely hide the memories. Their behavioral effects would probably lead, sooner or later, to their rediscovery. If you want to erase a memory irretrievably, you've got to get in and damage the brain. Particular brain regions are apparently specialized for particular kinds of information (language, music, visual impressions, etc.), so in theory, with good enough technology it should be possible to identify and disrupt the specific neurons (perhaps even the specific axons) responsible for a particular memory. An ultra-high-resolution MRI scanner, coupled to a transcranial magnetic stimulator in the hands of a skilled interrogator, could probably do the job without even peeling the skull. In fact, such procedures may be commonplace in the future, as neurological or psychological therapy, for personal or legal reasons, or even, ironically, as a memory aid in other areas.
However, since memories are associational rather than binary, they probably aren't localized in the same way that computer bits are. Rather, a particular memory is a pattern of neurons firing in several different regions of the brain. Each of these neurons is probably involved in countless other memories as well, so that the deletion of an unwanted memory would cause a lot of collateral damage in wanted memories, associations and perhaps even the sense of identity itself. So if you're ever in a position to try this, a la Paycheck, make sure you really, really need to, because the person who comes out the other side may not be entirely you.
But hey, for some people that may be a good thing.
Wil McCarthy is a rocket guidance engineer, robot designer, science-fiction author and occasional aquanaut. He has contributed to three interplanetary spacecraft, five communication and weather satellites, a line of landmine-clearing robots and some other "really cool stuff" he can't tell us about. His short writings have graced the pages of Analog, Asimov's, Wired, Nature and other major publications, and his book-length works include the New York Times notable Bloom, The Collapsium and most recently The Wellstone and a related nonfiction book, Hacking Matter.
