April 11, 2009

Memory modifiction

DisputationThe Messy Future of Memory-Editing Drugs - I was interviewed by Brandon Keim about memory editing. His angle, which I agree with, is that it is unlikely to work as neatly as in sf (which resulted in this short list of memory-sf stories as an aside). We will have side effects, limited efficiency, affecting other memories and, most importantly, emergent effects that we cannot currently predict. This doesn't mean it will be bad or dangerous, but it is likely to be much messier than we like to think. For example, my and Matt's paper on memory editing ethics explicitly assumes that the technology works as advertised - because to do otherwise would make the philosophy hard to do. In stories, unless the flaws of technology are used to add to the setting or form a plot point, it is the flawless working of memory modification that usually is the key plot point.

Maybe science fiction is biasing us to think that technology either works exactly like it should or that it is totally flawed. Think of the Star Trek transporter, which seems to be perfectly safe except for when it kills, merges or dimensionally translocates people. Real technology is different: it acts up every day, often in minor ways. Cars have become safer in many ways, but airbags can hurt you anyway. I'm using a fantastically advanced word processor that cannot (easily) paste text without formatting it like on the originating webpage, and suggests mistaken grammatical errors. Keycards work - except when they stop working for obscure reasons. And so on.

In reality we muddle through. We can use imperfect technology, find ways around the problems (the main door doesn't let me in? I use the side door, which does), accept some limitations and often solve problems by adding new layers with their own problems (how much of our processing power is used by antiviral software?) Usually new technologies introduce completely logical but hard-to-predict problems (spam follows logically from the email model we choose; airplanes are by their nature potential projectiles) that we then spend energy solving. There is no reason to think we can design ourself out of this except in very special cases - design takes much time, intelligence and energy, and is limited by the combinatorial explosion of possible interactions in the real world.

In the end, I think Joel Garreau was right in Radical Evolution when he predicted that instead of a clean singularity or apocalypse we are just going to muddle through the transition to a technologically transformed species. But that should not be a comforting thought: muddling through still requires a lot of hard work.

In the case of memory editing there is a lot of research that needs to be done, not just in neuroscience but in applied psychology. The best ways of using it are probably something like hybrids of cognitive behavior therapy and judicious applications of memory-altering drugs. But how to do that, how to measure results, how to decide what alterations produce the best outcomes - that is going to be very tricky. And we are going to learn a lot simply by failing in various ways.

But who wants a sf story where the protagonist has a non-working memory treatment, followed by four others that kind of work before finally finding a solution using a completely different method? Almost Total Recall?

Posted by Anders3 at April 11, 2009 02:10 PM