Many years ago, as a psychology student, I learned all about operant conditioning, which was very fashionable at the time. I learned how a hungry small rodent in a Skinner box would wander around the box, active because it was hungry, and would eventually press a lever, which would give it a food reward. Gradually, I learned, the animal would come to associate the two events – the pressing of the lever and the receipt of food – at which point it would approach the lever when it was hungry, and begin pressing. It all seemed very straightforward and logical.
In those days it was compulsory for BSc psychology students to undertake some coursework involving animals – not vivisection, but usually some kind of learning experiment. I was rather reluctant to do this, but wasn't able to articulate why: the idea of ethical objections to 'harmless' animal coursework had not yet been voiced, and I only understood later where my own reluctance came from. Eventually, therefore, I found myself in the animal lab with two hamsters, charged with the task of seeing that they obtained their food by means of the Skinner box. The hamsters had previous experience, so didn't have to learn the task from scratch. They had been reduced to four-fifths body weight to make sure they would be hungry, as per the textbook, and my task was to place them in the box, and record how they behaved.
The first hamster, it appeared, had been reading the same books as I had. It explored a bit, and then began pressing the lever, breaking off every now and again to nose into the food box. This rapidly produced the predicted response, and that hamster would hammer away on its lever throughout the experimental session, pausing now and then to eat, then going straight back to pressing its lever. So far, so good. I came away with a nice clear frequency chart, and an even clearer conviction that doing an animal experiment as coursework was just pointless. After all, we knew what was going to happen.