This morning I'm sitting in front of a computer, watching a recording of brainwaves move across the monitor. While I was checking for eye blinks and muscle twitches -- events that can mess up the recording -- I started to think about all the things that can go wrong in an experiment. I started to wonder about how experiments ever work out.
In this experiment, I'm looking at how music affects the brain while people are doing a working memory task. Sounds simple enough, but what kind of music should I choose? What kind of task should I use? What kind of images and/or words should people be trying to remember? How many trials? How many electrodes on their head?
Then, when I get data from enough people, how to analyze it? Do I filter certain frequency ranges? Do I average stimuli that are similar (but not exactly the same)? Do I take into account whether people are musicians or not? If so, how should I define a "musician?"
It seems that many more experiments actually have useful results than would be expected to, based on the number of things that could go wrong or decisions that could have been made incorrectly. For instance, there's Gregor Mendel's original experiment. Mendel was a monk who observed pea plants change as he bred them, and from his observations he famously recorded the basic rules of genetic inheritance. But it turns out the outcome of his very influential breeding experiments was a fluke. If Mendel had chosen traits other than the ones he did, he wouldn't have obtained his very simple and intuitive result.
Maybe we only hear about the experiments that work, and we don't hear about the many experiments that go wrong, so we have an inaccurate view of the proportion of experiments that work. This likely explains some of the imbalance. But I know about all of my own experiments, including the ones that I don't publish because they go wrong. At least in my experience and the experience of other scientists I talk to, the ratio of useful-to-useless results is nowhere near what one would expect based on the hundreds (thousands?) of decisions that need to be made to perform a single experiment.
Instead, I think this lopsidedness in experimental outcome is a case of intuition at work. Scientists are experts at rationalizing everything, that's sort of the job description. In a way, that makes us least likely to notice when an idea arises from our intuition. In fact, it may be that in the well-trained scientists' brain, intuition has a field day because we just come up with a rational explanation for why we'd make one experimental choice over another, and because we're capable of rationalizing anything, voila! Our intuitions can masquerade as well-thought out choices.
On the other hand, this lopsidedness is probably not particular to science at all. In all fields, those who keep doing the work seem to develop expert intuitions about what will make things work and what will mess them up. These intuitions are often not algorithmic -- they live in the seemingly nonlinear space of the subconscious mind.
Regardless of how common this kind of lopsided success rate is, what I think is most important here is that intuition doesn't have to name something mysterious and unexplainable. It can just be a name for the work the subconscious mind does every day -- gathering data, making connections, and coming to conclusions -- without the bulky and bossy conscious mind to get in the way. In that sense, intuition is just a way of letting the sometimes smarter and often wiser subconscious have its say without setting off the alarms of the conscious mind. Scientists or not, it seems to me that listening to our intuition may be like having our own internal oracle.