Ten years ago I tried to learn to predict the future. I trained myself the same way I was training rats at the time. Instead of pressing levers for food pellets, I pressed computer keys for points. How did I get points? By predicting a random number that the computer was about to generate.
In around 1998 I was working in a behavioral research lab at Reed College, writing computer programs and training rats to press levers for food. I went to a talk in given by Daryl Bem, who is making waves with his recent article about predicting the future. He said the first lesson of ESP research is, don't talk about it before you get tenure, so I should probably keep my mouth shut. But I found it fascinating.
So one morning I wrote a simple computer program. The program would wait for me to press one of four buttons, 1, 2, 3, or 4. Then it would generate a random number. If my response was the same as the computer's, I got a nice sound as a reward, and a point. If I was "wrong," I got a bad sound and no point. I did blocks of 100 trials. By chance, I would get 25% correct. To do better than that, I figured, I had to predict the future.
I tried really hard. What I was trying to do, I didn't know. I was just trying to feel out, in my mind, what the right answer was. It was downright weird.
But my goal wasn't to do well at first, it was to get better. Bem had talked about there being a certain subset of the population that seemed to have "psi" abilities, and (if I remember correctly) that the CIA might have hired some of these people to intuit the location of hidden bunkers, etc. I didn't think I was one of those people, but I wondered whether I could become one--that is, whether people have ESP abilities that they can access with training. Training of ESP seemed distinctly missing from the research Bem talked about.
As it turns out, I did surprisingly well. After a period of training, I tested myself. On some 100-trial blocks, I'd try to get the answer right. On those, I averaged 27%. On others, I tried to get the answers wrong. On those I averaged 23%. How could that be?
The most likely explanation is that it was just random chance. Accidents to happen, after all. The point of statistics is to determine whether something was an accident or a reliable phenomenon. I didn't keep at it long enough to find out.
A second explanation was that I had accomplished a remarkable feat: I had learned to predict my computer's random number generator. That sounds pretty hard, but my own mentor/boss at the time, Allen Neuringer, had, coincidentally, done some research that made it seem possible.
The third explanation is that I was predicting the future. I think the chance of this being true is basically zero. Then again, Daryl Bem knows as much as anyone about criticisms of psi phenomena, and the pitfalls of doing psi research (including random-number generation issues). And if it is true, talk about a major step forward for science!
When I told my boss, Allen--a wonderful person and advisor, and a behaviorist interested in hard facts--he looked at me, paused for a moment, said "ESP doesn't exist," and walked away. He's probably right. But it's a question worthy of scientific investigation.
Post-Script: Here's another take on ESP that I wrote with my brother Sam. http://www.miller-mccune.com/science-environment/i-foresee-an-uproar-over-an-esp-study-26715/