Why do we push the green button? Because whenever we do, the train door opens. That’s it, and it’s all the explanation we need to understand a seemingly pointless bit of human behaviour.

Edward Thorndike first proposed the law of effect more than 100 years ago. Simply put, the law says that successful behaviour will tend to be repeated. In the language of behavioural psychology, a behaviour – anything that we do – that’s followed by a reinforcer – something good happening, like a reward – will be more likely to occur again in the future. When we learn, that’s what we’re learning – that a particular action of ours was followed by success, so perhaps it will be next time too, so let’s do it again. When we push the green button, that action is followed by the reinforcer of the train door opening, and we’re likely to push the button again next time.

Does it matter that the door would have opened anyway, and that our action actually had no effect? Not at all, because the law of effect doesn’t say anything about causality. All that’s needed is that the behaviour is followed by the reinforcer. The behaviour doesn’t have to cause the reinforcer for the law to work.

Really, how could it be otherwise? Unless we unscrew the green button and have a look at the wiring, we don’t know what it does. Our best evidence for causality is usually correlation of events in time – when one event follows another, we tend to assume that the first event caused the second. That may not be very logical or scientific of us, but perhaps it’s a good thing. We’re more likely to learn new successful behaviours if we’re biased in favour of attributing causation to our behaviour than if we take the position that the world is random and that nothing we do can make a difference.

Animals show the same bias. BF Skinner showed, in the 1940s, that providing food at regular intervals to a pigeon led to it repeating the behaviour it had been performing just before the food was delivered. That is, it behaved as if what it was doing had caused the food reinforcement, even though it didn’t. Skinner called this phenomenon superstition, and it’s a good explanation of superstitions in humans too.

Most of our superstitions are based on preventing something bad happening, and “nothing bad happening” is another kind of reinforcer. Whenever we avoid stepping on a footpath crack, our mother’s back doesn’t break, and the bears don’t eat us. (It depends on our cultural background which of those we’re trying to avoid.) So our behaviour is reinforced, and so is the superstition. This is just what the law of effect, and our evolutionary heritage, predict.

And usually it doesn’t matter. We do ourselves almost no harm by avoiding footpath cracks and pushing unnecessary buttons on trains. But sometimes it might matter.

When we can’t inspect the world’s wiring then there’s only one way to find out whether apparent causality is real or just correlation, and that’s to conduct an experiment – to do science. What happens if we don’t push the button? Let’s find out! At least when the consequences of a mistake are minor, it has to be good to understand more about the world’s wiring, about real causes rather than illusion, and perhaps to discover some bits of our behaviour that are pointless relics of our past.