In 1963, Harvard psychologist Robert Rosenthal ran an interesting experiment. He sent some students a group of rats which he called “maze bright” rats. He told them that these were rats bred to run mazes more quickly than the average rat. He asked the students to confirm these results, and they did. Then Rosenthal sent other students “maze dull” rats, which, as you might imagine, were rats bred to run mazes more slowly than the average rat. The students once again confirmed the rats’ labels.
So Rosenthal ended up with data that showed that about half of his rats ran mazes incredibly quickly, and half of his rats ran mazes incredibly slowly, exactly as he told the students they would. Nevertheless, this was surprising data. It was surprising, because he never bred the rats at all. They were all from the same original population, and there was no difference between them.
Rosenthal ran this experiment to prove what is called the “observer-expectancy” effect. In short, this effect is about how every human being, including scientists, want certain things to be true. And, because we want certain things to be true, we will slightly “nudge” the evidence, either in real life or in our minds, to make it true. It’s basically the academic version of your love-struck friend. If Jessica says “hi” to him, he’s convinced that it means she’s flirting with him, even if she says hi to everyone at work. He willfully ignores the evidence to confirm what he wants to be true.
Rosenthal’s students wanted their professor to be right. Or, more likely, they didn’t want to look like incompetent experimenters in front of their august professor. Either way, they changed the data to fit what they were told was true. It’s possible they changed it in real life, by changing their measurements to fit in with the “maze bright” and “maze dull” hypothesis. Or they might have changed it in their analysis, by ignoring the real data as “outliers”. Regardless, these were well-trained scientists who all, in the end, accomplished the exact same wrong result. Ironically, this ended up being a pretty successful experiment for Rosenthal, if not for his students.
It’s interesting that it took until the 1960’s for psychology to formally recognize this problem, because it had been known for a lot longer. As you might have already guessed, one philosopher in particular described this issue to a T, all the way back in the 16th century. His name was Francis Bacon, or, if you prefer his full name, Francis Bacon, 1st Viscount St Alban.
Francis Bacon came to his theories a very different way than did Robert Rosenthal. Francis Bacon had been taught in the intellectual methods of Aristotle, who we’ve talked about previously. As we know, this meant that Francis Bacon had been trained in a logical, robust system that had little relation to the real world. Even worse, it didn’t even try to have relation to the real world.
Bacon decided there was a need for a “new instrument of science”, one which would gather observations from the real world and turn them into theories. Unlike Aristotle’s system, everything in it would be true and supported from observation. Bacon just needed to make sure that his instrument worked.
Bacon had a problem, though. Coming up with theories based on observations seemed like the right way to go, but other people had tried that before. After all, that’s exactly what people who didn’t know about Aristotle tried to do. They would see a natural phenomenon, like lightning, and then come up with a theory, like lightning is Zeus throwing thunderbolts. This, of course, wasn’t science, but they didn’t know that. So Bacon needed something better than just “observation leading to theories”.
Bacon’s specific method was a bit silly, so we won’t go into it. But his general idea was sound. He said we had to be very, very careful when we let our observations become theories. We had to build up our theories carefully, starting with specific instances, and only gradually working our way up to universal laws, making sure that we don’t let mental tricks lead us astray along the way.
What were these mental tricks? Well, Bacon called them “idols”, and they included the “observer-expectancy” effect of Rosenthal. He had four of them: idols caused by human nature, idols caused by personality, idols caused by language, and idols caused by philosophy.
Idols caused by human nature include the observer-expectancy effect, as Rosenthal mentioned. They also include what we call today “confirmation bias”, which is how we tend to form a conclusion, then draw evidence to support it. To paraphrase Andrew Lang, we use evidence like a drunk uses a lamppost, for support rather than illumination. Bacon also identifies an amazing number of other biases which were only named hundreds of years later, like the availability heuristic and the gambler’s fallacy.
Idols caused by personality are much simpler, and tend to just be how different people like different ways of looking at the world, and tend to frame hypotheses in their preferred manner, whether rightly or wrongly. For instance, an engineer might explain why a bridge collapsed in terms of how it was built, while a historian would explain it in terms of how political conflicts led to shoddy workmanship on the bridge.
Idols caused by language are simpler still, and simply pertain to how we sometimes twist our observations to fit our language, which can result in oversimplifying, complicating, or simply miscommunicating what was actually observed. Rosenthal, in coming up with the catchy names “maze bright” and “maze dull”, may have helped to cause his students to fall prey (pun somewhat intended) to this idol. Alternatively, early sailors who saw manatees and insisted they saw mermaids may also have fallen prey to this, as they had the word for manatee but not for mermaid.
Finally, idols caused by philosophy are Bacon’s ways of attacking those who he thought were actively holding science back: those who followed Aristotle, those who insisted on putting Christianity into science, and the “experimentalists”. Those who followed Aristotle, in Bacon’s view, twisted their observations to fit their logic, when they should have taken pains to make their logic fit their observations. Those who put Christianity into science “ruined both”. And the experimentalists, those who tried to use experiments to learn science rather than observations (like Galileo), never really understood what they thought they did.
Needless to say, I’m much more sympathetic to Bacon’s attack on the first two than the last one. Bacon’s disdain for experiments is part of the reason why his specific instrument of science never really works. However, his insistence on carefully building up a theory from facts, and his identification of so many of the cognitive biases we must avoid when doing so, cement his reputation as a truly great thinker.
What should you take away from this? Well, it wouldn’t hurt for you to actually read Bacon’s list of idols. Even though they were written 400 years ago, they’re surprisingly understandable, and pretty interesting. However, even if you don’t, you should think about how to form opinions and theories of your own. Don’t go charging in with a preconceived notion, and gather your facts in response to your opinion. Be open to the possibility that your first reaction is mistaken, and be aware of any mental shortcuts you take along the way. Finally, when you communicate your results or thoughts, watch your language closely, and make sure your audience understands your ideas as they were when you came up with them.
In fact, one place you might do this today is with your political opinions. You probably fall on side of a hot political issue, and you are likely aware of the other side’s view. What I want you to do is to carefully consider the other side. Pretend you were an alien, coming down from outer space, and you had no opinions on this whatsoever. If you were given the other side’s evidence and arguments, would you be convinced by them? Or, if you look at your own side, are you convinced by your side’s arguments?
The only way to actually carry out this exercise, by the way, is to actually go through with it. Belly-crawl your way through the muck and the grime, and peer carefully at the facts, the other side’s and your own. Then, carefully, like you’re building up a stack of cards, build your theories from the facts.
That’s how Francis Bacon would have had it. But, if you find it hard to pretend you know nothing, and to build your card castle starting from the ground floor, you are not alone. Luckily, there’s another philosopher waiting in the wings to help you, me, and all of humanity question our knowledge. Not just our theories, like Bacon would have it, but all of it, even our very existence. Who is this radical skeptic? Well, it’s the one and only Rene Descartes, and we’ll talk about him next!