Lightspeed: Edited by John Joseph Adams




Saving Your Brain to a Disk

Mind swaps have been a standard science-fictional prop for decades. Just wire up the brain in a tired old body to something new, turn on the magical mind-transfer machine, and your mind gets downloaded into a new body, a computer, or a spaceship. Today, we can do computer mind swaps routinely; a few weeks back, I downloaded the memories of an aging PowerMac G5 into a sparkling-new MacMini through a Firewire cable. It worked like a charm. So why can’t we download our memories from our aging bodies into a shiny new computer?


What Happens in the Brain?

The short answer is that we don’t know how. The longer answer is that although we can observe brain activity, and have a general idea of how memory works, we don’t know how to read the format the brain uses to store memory. Imagine you were a computer operator back in 1955, when circuits were made of vacuum tubes, data was stored on punched cards, and it took a room of electronics to do the job of a pocket calculator. How would you feel if a time-traveler handed you a 100-gigabyte flash drive? You might understand it was a chunk of semiconductor that contained tiny transistors, but you wouldn’t have a clue of how to read the stored data. So you’d probably be better prepared than a scientist trying to read memories from a brain today, particularly because that makes the rather brash assumption that the “mind” can be transformed into digital bits at all.

We do have all sorts of ways to observe activity in the brain. Magnetic resonance imaging can show blood flow, positron emission tomography can map neurotransmitter activity, and electroencephalograms can record electrical activity. Scientists can tell what parts of the brain are active, and how active they are. That’s enough to tell what parts of the body are stimulated, sexually and otherwise.

But nobody knows how to read those images well enough to read your memories, or tell your thoughts beyond that a person of your favored gender is hot. (So much for the mind-reading dreams of all spy agencies since the darkest days of the cold war.) The brain’s code remains unintelligible, and we’re not even sure if we’re looking in the right place for the signals.


Record-Keeping Instead of Downloading

So if true mind reading is out, how can we record our thoughts? One option is an old-fashioned thing called an autobiography, which dates back over 1600 years to the Confessions of Saint Augustine. Writing an autobiography tells the story you want to tell, but it doesn’t re-create your memories.

Technology can help. Famed computer engineer G. Gordon Bell is taking the quest for memory by using computers to store not only his memories, but also things he sees and hears. “I believe this is the quest for what a personal computer really is,” Bell told Computerworld. “It’s to capture one’s entire life. A personal computer wouldn’t be a machine that just sits on my desk. It’s a repository. I think of the system as a personal memory. I feel immensely free by having all the information there.”

Bell calls his project “My Life Bits.” He has written about it in books called Total Recall and Your Life, Uploaded. He also blogs about it, describing gadgets that help him record his life, like a Bluetooth video camera called Looksie, which records everything you see, and stores it for ten hours. If you see something interesting, you can push a button to save 30 seconds of video posterity. If you’re bored, you just let that time scroll out of the camera’s memory into digital oblivion.

Inevitably, it all builds up. Bell suggests making digital copies of artifacts you want to pass along to your heirs, so everyone can get copies of everything. Perhaps that would prevent a few family fights, but it doesn’t explain who would get the original objects, especially the expensive fine china from your great-grandmother.

It’s a fascinating idea. On one hand, you have to wonder if the digital files will just accumulate on servers somewhere, the electronic counterpart of the mountains of old family photos stashed on closet shelves, storing memories that nobody takes time to look at. Yet on the other, could we use that treasure trove of data to recreate a person by simulating them on a computer?


Weaving Together a “Person” From Data

It might be possible. The trick is to program a computer to sift through the accumulated life data to look for patterns of speech or behavior. The computer would not have to be programmed to look for anything in particular. Instead, it would look for speech or behavior that distinguished that person from others.

That futuristic-sounding software already plays a role in our daily lives. Spam filters use it to identify electronic junk mail, but comparing messages you identify as spam with those you consider legitimate (called “ham” in computer jargon). You may never have heard from Mrs. James in Lagos before, and her letter may be original, but a good spam filter knows that “million,” “business partner,” “next of kin,” and “transfer” are clear fingerprints of Nigerian spam.

Google takes the idea further by using similar software to look for patterns that distinguish well-written and authoritative web pages from computer-generated junk laden with keywords linked to hot topics such as celebrity scandals. Search engines use such software tools to rank pages by their quality without having people laboriously read everything on the web. The more data, the better the technique works, so Google scans everything on a site and analyzes how words are used. Computers can even identify writing traits of individual authors. Scan enough writing of a prolific author, and someday a computer might be able to imitate the author’s style—although it would be much easier to match a pulp-fiction hack than a subtle literary stylist.

Spoken words might be used in the same way. It would take a lot of recordings, but Richard Nixon recorded some 3700 hours of conversation while he was President. Transcribe those tapes, leaving in the original expletives, and a powerful enough computer might be able to simulate Nixon. The computer couldn’t solve mysteries like what happened to the missing eighteen minutes of erased tape, but it could give insight into how Nixon thought, and might be able to simulate how he talked with other people.


Can We Make a Transfer, Instead of a Copy? 

Transferring knowledge is a different matter, but engineers have already developed software for that, called expert systems. They codify specific knowledge accumulated by experts in the field, and assemble into an organized form that non-experts can use. For example, the expert system could collect techniques that retiring machinists used to fabricate hard-to-make components. When new workers face similar problems, they can query the system to learn details such as what lubricant to use while cutting a certain alloy.

Similarly, someday veteran space pilots could tell the tricks of their trade to a future expert system, which could codify their knowledge for the next generation of pilots. Of course, such an expert system would not preserve either your personality or your sense of being, so it couldn’t transfer your “self” from a worn-out body into a spaceship where it could survive indefinitely. We have less idea of how to do that than we do of how to record your thoughts or how to translate them into words. But that might not be a bad thing. Would you want to spend eternity as an expert flying a slow freighter to Neptune? That’s the sort of boring place where Gordon Bell would turn his life recorder off.

Enjoyed this article? Consider supporting us via one of the following methods:

Jeff Hecht

Hecht, JeffJeff Hecht is a science-fact writer who contributes regularly to New Scientist magazine and Laser Focus World. His specialties range from lasers and fiber optics to paleontology. His books include Understanding Lasers, Beam: The Race to Make the Laser, Understanding Fiber Optics, and Beam Weapons: The Next Arms Race. He has had short fiction published in venues ranging from Analog and Asimov’s to Nature. More at his web site: