August 13, 2013 | 21
Fellow Scientific American blogger Melanie Tannenbaum is flustered by allegations that psychology is not a science and I can see where she is coming from. In this case the stimulus was a piece by Alex Berezow, a microbiologist, who in a short and provocative piece in the LA times argued the case that psychology is not a real science. I think he’s right. I also think that he misses the point.
Berezow’s definition of science is not off the mark, but it’s also incomplete and too narrow. Criticism of psychology’s lack of rigor is not new; people have been arguing about wishy-washy speculations in fields like evolutionary psychology and the limitations of fMRI scans for years. The problem is only compounded by any number of gee-whiz popular science books purporting to use evolutionary and other kinds of “psychology” to explain human behavior. Neither is the field’s image bolstered by high-profile controversies and sloppy studies which can’t be replicated. But it’s hardly fair to kill the message for lack of a suitable messenger. The same criticism has also been leveled at other social sciences including economics and sociology and yet the debate in economics does not seem to be as rancorous as that in psychology. At the heart of Berezow’s argument is psychology’s lack of quantifiability and dearth of accurate terminology. He points out research in fields like happiness where definitions are neither rigid nor objective and data is not quantifiable.
Happiness research is a great example of why psychology isn’t science. How exactly should “happiness” be defined? The meaning of that word differs from person to person and especially between cultures. What makes Americans happy doesn’t necessarily make Chinese people happy. How does one measure happiness? Psychologists can’t use a ruler or a microscope, so they invent an arbitrary scale. Today, personally, I’m feeling about a 3.7 out of 5. How about you?
This is absolutely true. But you know what other fields suffer from a lack of accurate definitions? My own fields, chemistry and drug discovery. For instance there has been a longstanding debate in our field about how you define a “druglike” molecule, that is, a chemical compound most likely to function as a drug. The number of definitions of “druglike” that have sprung up over the years are sufficient to fill a phonebook. The debate will probably continue for a long time. And yet nobody will deny that work on druglike compounds is a science; the fact is that chemists use guidelines for making druglike molecules all the time and they work. In fact why talk about druglike compounds when all of chemistry is sometimes regarded as insufficiently scientific and rigorous by physicists? There are several concepts in chemistry – aromaticity, hydrophobic effects, polarizability, chemical diversity – which succumb to multiple definitions and are not strictly quantifiable. Yet nobody (except perhaps certain physicists) denies that chemistry is a science. The accusation that “softer” fields are less rigorous and scientific than your own is common enough to be captured in this xkcd cartoon, but it’s more of an accusation than, well, a quantifiable truth.
Now chemical definitions are still admittedly more accurate and quantifiable than definitions of happiness or satisfaction. But the point is that not everything measurable needs to be quantifiable to the sixth decimal point to call itself scientific. What matters is whether we can come up with consistent and at least semi-quantifiable definitions that are useful enough to make testable predictions. Psychological research is useful not when it’s quantifiable but when it says something about human nature that is universal and revealing. A few days ago I watched a new movie about the life of psychologist and political thinker Hannah Arendt and mulled over the “banality of evil” that Arendt made famous. Now the banality of evil is not exactly rigorously quantifiable like the angular momentum of a figure skater, yet few people would deny that Arendt made an enormously valuable contribution to social science. The contribution worked because it was testable and repeatable (in Milgram-style experiments for instance) and true, not because you could accurately measure it with an fMRI machine. Or consider Daniel Kahneman’s seminal work in behavioral economics which has led to real insights into decision making and biases; very few people would call what he did unscientific.
In fact one can argue that social scientists tread on dangerous ground when they start trying to make their discipline too accurate; the proliferation of mathematical models of finance that led to disaster on Wall Street are good testaments to what happens when financiers start longing for the rigor of physics. As the particle physicist turned financial modeler Emanuel Derman puts it, “Physicists are trying to discover 3 laws that will explain 99% of the universe; financial modelers should be content with discovering 99 laws that explain 3% of the universe”. So is finance a science? The point is that we still know too little about biology and social systems to achieve the kind of quantitative prediction that sciences like physics do (on the other hand, physics – depending on what kind of physicist you are talking to – does not have to deal with emergent phenomena on a routine basis). But that does not mean that everything we say about human nature is completely unquantifiable and useless.
One valuable contribution that Berezow makes is to indicate the criteria that a field of study should satisfy to call itself a science. I think these criteria are incomplete and too rigid, but I think they provide a useful ruler for psychology to examine its own gaps and goals.
Why can we definitively say that (psychology is not a science)? Because psychology often does not meet the five basic requirements for a field to be considered scientifically rigorous: clearly defined terminology, quantifiability, highly controlled experimental conditions, reproducibility and, finally, predictability and testability.
I have already talked about the first two criteria and indicated that lack of clear terminology and quantifiability does not automatically consign a field to the bin of pseudoscience. The third criterion is actually interesting and important and it’s not completely clear how to get around it. Since human beings are not electrons, it’s indeed very hard to do an experiment with them and get the exact same results every single time. But that is why psychology relies heavily on statistics, to determine precisely whether the variability in results are due to chance or whether they reflect a real difference between samples. Admittedly this is a limitation that psychology will always have, but again, that does not mean it will preclude it from ever being useful. That’s because as Melanie accurately notes, even fields like particle physics rely heavily on statistics these days. Nobody observed the Higgs boson directly, it was only visible through the agency of complex tests of statistical significance. And yet particle physics has always been regarded as the “purest” science, even by other physicists. Or consider non-linear dynamics where dependence on initial conditions is so extreme that systems like weather and biological populations become completely chaotic after a while. And yet you can apply statistics to these systems, make more or less reliable predictions and call it science. Which brings us to Berezow’s last two points. Testability and prediction are indeed two cornerstones of science. I have already indicated that testability can often be accurate enough to be useful. As for prediction, firstly it can lie within a window of applicability. In my own field we routinely predict the activity or lack thereof of novel drug molecules. Sometimes our predictions are 90% successful, sometimes they are 40% successful. Even when they are 40% successful we can get useful data out of them, although it’s also clear that they have some way to go before they can be used on a completely quantitative basis. And all this is still science.
But more importantly, prediction is not actually as important to science as Berezow thinks. The physicist David Deutsch has noted that after watching a magician perform a magic trick ten times you would be able to predict what he would do next, but it doesn’t mean at all that you have actually understood what the magician is doing. Contrary to popular belief, in science understanding is at least as or more important than prediction. And psychological studies have definitely provided some understanding of how human beings behave under certain circumstances. It has helped us understand questions like: Why do smart people believe weird things? Why do otherwise decent people turn into monsters under certain circumstances (the banality of evil)? What is the basis of the bystander effect in which empathetic people don’t come forward to stop a crime? Psychology has provided intriguing clues and explanations in all these areas, even if those explanations are not one-hundred percent reproducible and quantifiable. Is this science? Well, it’s not a science like physics, but why should physics be the yardstick for measuring the “sciencyness” of various fields?
At the same time, I agree with Berezow that science cannot be redefined to such an extent that it no longer obeys time-honored criteria like testability and reproducibility; if you gradually start relaxing foundational requirements like hypothesis testing and observation you quickly slide down a slippery slope, at the bottom of which lie creatures like creationism, the Piltdown Man and astrology. But this was also the case with the beginnings of modern science when data collection was dominant, explanations were few and nobody had any idea what hypothesis testing meant. Yet we call what Linnaeus was doing science, and we call what Brahe was doing science. For crying out loud, even some of the work done by alchemists classifies as science; they did refine processes like distillation and sublimation after all. In my view psychology is in what we might call the Linnaean stage, collecting and classifying data and trying to find the right theory for describing its complexities. To me the acrimonious debates about evolutionary and positive psychology reflect the trial-by-fire that every field goes through in its early days to separate the chaff from the wheat. If you apply a narrow-minded definition of science then it might indeed be hard to call psychology a science. But what matters is whether it’s useful. And to me the field certainly seems to have its uses.