Hard Science Vs. Harder Science

This is one of the essays I delivered to my patrons last month. If you want to support more work like this, and see it earlier, you can sign up here.

I started university life as a physics major and ended it with a degree in psychology. Along the way, I was a tutor and a teaching assistant in physics and a research assistant in psychology. Graduating with honors in psychology also meant I had to run an independent research project. I chose to replicate an important study in a novel population and was lucky enough to be able to recruit one of the original authors as my adviser.

Photo of a stone, probably quartz, with a feathery vein of gold running over its surface.
“Dendritic crystalline gold” by James St. John, CC BY 2.0

While I ultimately decided I didn’t want to work in either field, the whole experience gave me a–perhaps unhealthy–interest in the fuss over “hard science versus soft science”. I’ve spent an absurd amount of time arguing over whether there’s a real difference between types of science that falls along those lines, including a delightful bit of argument with former science journalist Susan Jacoby, which was unfortunately brief, as it happened in the middle of a workshop I was running on a different topic.

Just this past summer, I sat on and moderated a panel discussion on the topic at CONvergence, with physics, geology, and psychology represented. I was hoping the video would be available by now, but the short version of the panel goes like this: None of us recognize any meaningful distinction in the practice of science between fields that are generally classed as “hard” sciences and those classed as “soft” sciences. None of these fields are more science-y or less than the others, and we’re all kind of tired of saying so.

Yet the idea that only some of these fields are “real” science, and particularly the idea that social sciences are somehow not scientific, persists.

That’s right. Psychology isn’t science.

Why can we definitively say that? 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.

There’s some irony that the author of those words trained as a biologist, but we’ll get to that later. For now, I’ll just note that biology is generally lumped with physics and chemistry as a “hard” science, with psychology and sociology on the “soft” side and anthropology stuck in limbo because most people aren’t quite sure what it is.

I’ll also note that this definition of science is facile and, well, wrong. It misses quite a bit about how both “hard” and “soft” sciences have been and are practiced, as well as the factors that determine which disciplines are allowed to fall under the heading of “science”.

Old Science Vs. New Science

I’ll talk mostly about psychology here, because I have a stronger background in its history and issues, but the history of psychology as a scientific discipline is only slightly more brief than that of sociology and anthropology. All three were born out of Enlightenment philosophy, as scholars decided their ideas about humanity should be tested against reality. Psychology is the youngest of these sciences, with a history of rigor dating back only about a century and a half.

A century and a half into the Scientific Revolution, we were arguing about the notation to use for this newfangled calculus stuff, making educated guesses about the nature of light, and coming to grips with gravity and magnetism on a very terrestrial scale. There was very little consensus and several competing frameworks to explain experimental results that had been largely driven by, “What happens if we do this?”

And Isaac Newton was studying alchemy.

This is what science looks like when it’s young. It looks like building tools and making sure they measure what you think they do and figuring out what to call the things they accidentally measured and cranking away at the math on the side until it (probably) does what you need it to do. It’s saying, “Well, that’s weird. Will it do it again?”, then figuring out what it means if it does or if it doesn’t or if it only does sometimes. It’s poking at the world because the damned thing is big, and it might do something interesting. It’s wondering whether your disagreement with another scientists is over differing observations of one thing or of two closely related things you’re calling by the same name. It’s theorizing in the absence of an overarching framework at the same time you’re trying to figure out how all the pieces fit together to make one.

It’s messy. It has to be. So much is unknown, including the scope of the problem. If our tools and methods were fully adequate to the tasks we’d set ourselves, we’d already know what we’re trying to find out. If we had names for everything, we’d have already discovered it. If the patterns that create predictability were obvious from one piece of knowledge, we wouldn’t have a scientific enterprise at all.

It’s only when we know enough to be able to reasonably intuit the size of the gaps in our knowledge, when we go from unknown unknowns to known unknowns, that everything starts to settle down into the patterns we teach to children as the scientific method. Even then, though, it’s a simplified and inadequate representation of what scientists do.

That psychology today looks chaotic is a mark of its newness as a discipline. It doesn’t tell us that psychology has no discipline.

The Complicated Equations

Even in more settled, venerable, “hard” scientific fields, we still run into areas that test what we mean when we’re talking about science. Take, for example, theoretical physics:

But in the opening talk at the workshop, David Gross, a theoretical physicist at the University of California, Santa Barbara, drew a distinction between the two theories. He classified string theory as testable “in principle” and thus perfectly scientific, because the strings are potentially detectable.

Much more troubling, he says, are concepts such as the multiverse because the other universes that it postulates probably cannot be observed from our own, even in principle. “Just to argue that [string theory] is not science because it’s not testable at the moment is absurd,” says Gross, who shared a Nobel prize in 2004 for his work on the strong nuclear force, which is well tested in experiments, and has also made important contributions to string theory.

What do we do when our theories outstrip our abilities to make predictions from them, much less to test them? If the answer is physics, we hold a conference. If the answer were psychology, we’d probably be talking about defunding departments.

The same problem can happen when the predictions produced by our models become probabilistic rather than deterministic, or a least when the predictions from our “soft” sciences do. In reality, many of our “hard” science models also make probabilistic predictions. Take climatology, for instance.

Climatology combines physics, particularly dynamics, and physical chemistry to study and predict the behavior of global and local climates. The prediction portion of this work is hugely difficult, as we don’t have the mathematics to practically solve equations with that many variables, even if we could measure them all directly. (We can’t, just for the record.)

Instead, we use models, simplifications of the complex systems whose future behavior we want to know. Then we provide the models with our best, highly educated guesses about the variables that drive the behavior we’re interested in. Or we create bracketing scenarios that give us the range of climate behavior we’re most likely to experience or that it’s possible for us experience, and we draw on records of past variables to assign the likelihood of each scenario.

Does it sound like there’s a lot of uncertainty here? That’s because there is. In this discipline rooted in hundreds of years of scientific study, specific predictions about complex systems are still very, very hard. We still call this “hard” science.

Hopefully, this helps put complaints that psychology doesn’t make predictions in perspective. We’ve identified and continue to identify the variables that have an effect on the complex system that is human behavior. We’re figuring out how they interact with each other. We’re in the very earliest stages of building our models that will produce the kind of probabilistic predictions of individual or group behavior.

That we’re not there yet isn’t an argument that psychology isn’t a science. It’s a measure of the complexity of the task.

The Familiarity Factor

Given that finding differences between “hard” and “soft” sciences largely relies on comparing long-established, simpler fields to newer, more complex ones, how has the idea that these sciences are fundamentally different? Well, first of all, most of the people arguing for the distinction simply aren’t taking the long view. They’re looking only at the current situation because it’s easy.

There’s also the fact that few people are trained across disciplines in any depth. I’m certainly not, and I have more crossover experience than most people. It’s easier to cherry pick your scientific results when other people don’t know all of what you’re choosing from.

Still, there’s another factor that matters, and that’s “common sense” familiarity with the material in question.

I think it was a physics professor who introduced me to the problem. He told me that it’s simpler to teach physics than it is to teach many topics, because students came to his classes with very few preconceptions they had to unlearn. Once he convinced them that circular motion didn’t work the way their brains expected it to and that measuring the effects of gravity produced unexpected results, students showed up to class ready to learn.

By contrast, everyone thinks they know how to speak and write English. They’ve been doing it all their lives. Everyone thinks they understand psychology, because people are such an important part of their lives. Everyone knows how society works, because they’ve been living in it.

When it comes to topics we deal with every day without studying them formally, we tend to think we know more than we do. We overestimate our own expertise. More than that, we downgrade the expertise of people who have studied the topic when they disagree with our lay knowledge and theories. We’d rather believe our grandparents who spoke in folksy aphorisms.

We don’t like the idea that there exist people who are more expert in us than we are. It’s easier to accept the idea that their whole field of study is invalid.

How Gender Matters

I mentioned at the start of this post that I’d come back to biology. We talked about this a bit on the panel at CONvergence this summer, because biology as a field is currently in a fascinating state of flux.

As long as I’ve known the “hard”/”soft” science debate to exist, biology has been classed as a “hard” science. It’s held that post despite being roughly the same age/maturity level as the social sciences. It was saved, as far as I can tell, by being a physical science rather than a social science. No, that doesn’t match the distinctions people try to make between “hard” and “soft” sciences. No, it didn’t matter.

Recently, however, biology has come under more scrutiny. Medical science in particular, has been in the news over the last half dozen years for replication problems, for promising treatments that don’t deliver, for the publication of nutrition and epidemiological studies that contradict each other. For a science that has been held up as “hard”, biology and medical science have taken a lot of knocks lately.

By the way, all this criticism has followed hard on the heels of rapid gains in the percentage of life sciences doctorate degrees granted to women (pdf). Also as a point of interest, the social sciences have long seen a relatively high percentage of their doctorates go to women. Race plays a factor there as well.

Does that mean people are calling these fields “soft” just because they have a lot of women in them? Not necessarily. However, in a world where men can’t even handle using the same lip balm as women, it’s not unreasonable to hear “not really scientific” as “ew, girly science” when one of the major differences between the two is percentage of women. That’s particularly true as we watch a scientific field get the downgrade to pink collar.

So let’s drop this “hard”/”soft” distinction between scientific fields, particularly as a gatekeeping mechanism to define what is and isn’t science. It doesn’t work on either a historical or practical basis, and we’ve got good reasons to mistrust even the impulses that lead us there.

Want to see more work like this? Support me on Patreon.

Hard Science Vs. Harder Science

18 thoughts on “Hard Science Vs. Harder Science

  1. 1

    I agree that string theory and the multiverse hypothesis are difficult or impossible to test at present, but they are logical extensions of sciences which have an excellent record of precise prediction over decades or centuries. By contrast psychology has never been capable of such precise prediction. It’s arguable that it never will be, because its subject matter is complex in so many relevant ways. Hard sciences such as celestial mechanics have it much easier. The Earth, with its biosphere, is an incredibly complex system, but most of that complexity is irrelevant for the purpose of calculating planetary orbits.

  2. 3

    Trained as a geologist, I have realized that the value of my science is in the beholder, depending upon what they’re worried about. My science is obviously deficient to the average young-earth creationist, being “non-observational”… and yet, why can’t geologists predict earthquakes better? Psychology is the same way, a useless waste of time and money… until someone needs help.

    A field that I often read diatribes against is evolutionary psychology. And I don’t have a problem with the notion that a huge majority of the current hypotheses are total junk. They’re in the messy beginnings of a new science.

  3. 4

    There is a fairly clear distinction that appears to have been overlooked. Physics, chemistry, geology and such are termed ‘hard’ because the subject matter under study is purely objective; the rules of the universe, the rules of atoms and such. Psychology and other ‘soft’ sciences instead study things that are partly subjective, in that there are elements that rely on personal experiences and perceptions. There is no measuring device that a psychologist can turn to to determine the illness of a patient, because the thing being measured / evaluated is the personal experience of the patient in question. Taking a large number of patients and looking at probabalistic trends in the relationship between X past event and Y exeperience, taking people as an aggregate and attacking the problem with mathematics, does not do away with the fact that the basic thing being studied is subjectivity itself.

    I would suggest that anthropology, history and such are in classification limbo because while they may use objective measurments such as radioisotope dating to determine the time at which A civilisation did B, the thing being measured is still not a universal property of matter but rather is entirely contingent on the relatively whimsical behaviours of humans. Geologycould be considered similar in that it is the study of the history of the planet, but it is distinct as it is determined by the behaviour of agency-less matter, and thus avoids the mess that is human behaviour.

    The one point where I agree with you is in terminology. The terms ‘hard’ and ‘soft’ science play into the hands of the sneerers, as those particular words carry connotations that imply ‘difficult’ versus ‘easy.’ I think the distinction is a reasonable one to make, but much better terms would be something like ‘objective’ and ‘subjective’ or similar. They at least make clear the nature of the distinction being made.

  4. 5

    Unless you’re endorsing dualism here, which I assume you’re not, you’ve made a fundamental error. You’re confusing the brain as a measuring instrument with the brain as an object of study.

    Even there, though, you’re underestimating the value of the human brain as an instrument. It remains one of our best tools for making sense of complex, ambiguous data. This is why scientists employ data visualization and why projects like Galaxy Zoo exist and have been successful. Brains don’t handle the data perfectly, but nothing does. That’s why ambiguity is ambiguous. And let’s not get into the work required to calibrate other measuring instruments or the subjectivity that goes into it.

    Also, where do you think our understanding of the limitations of the human brain came from, if not from the science of psychology? Accurately describing a complex system is (part of and not exclusive to) science. Wanting that system to be less complex because that fits our notions of how the world should work is not.

  5. 6

    The distinction I put forth was that the things being studied are differently objective / subjective. The rebuttal you put forth is that an investigation into anything at all must perforce have some element of subjectivity due to the human investigators. But this does not address the point I made: that the subject matters under investigation are different in terms of objectivity / subjectivity.

  6. 7

    No, my objection is that you’re telling me we cannot study the brain as an organ in which behavior objectively does or does not happen because it is also an organ that adds subjectivity when it evaluates behavior and that this is an absurd conflation of two different phenomena.

  7. 11

    In fact I am not. Far from saying we cannot study behaviour, I even mentioned studying behaviour twice. The closest I get to saying such is the sentence beginning“There is no measuring device that a psychologist can turn to to determine the illness of a patient,…” and there I think the main issue is that I may have used psychologist where I should have used psychiatrist.

    But the point I was making did not hinge on that. It remains that the object of study for a chemist is an angency-less molecule that is governed by forces rather than stimuli; the behaviours being studied are merely those of a passive object being pushed about by immutable laws. Something that a human is not.

  8. 13

    I also agree that there’s a problem with the terminology (“hard” and “soft”). But I’m not sure that the terms “objective” and “subjective” are better. Those words also have connotations, and I think people tend to associate “objective” with science and “subjective” with not-science.

    That said, I think Holmes@4 is right to bring up “agency”, and I think that’s the more important distinction here. Our brains/minds aren’t perfect, and the data we can collect depends on the technology we have to work with at any given moment, and we have to interpret the data, and our interpretations might be [always are?] faulty or incomplete or biased in some way.*

    But there is still a difference between studying things that are “agency-less” and things (persons?) that have some sort of agency.

    The stars/planets/etc don’t have minds of their own; they are not agents. Humans (persons) *do* have agency (to some extent). There’s no need to try to trick a planet into thinking it’s being looked at for one reason rather than another. You don’t need to put a planet in a room with a “confederate” that the planet thinks is just another planet of investigation, but is really working for the investigator.

    “Human sciences.” I’ve read things (originally published in French) that refer to “the human sciences”. Maybe that’s a better term. Unfortunately, I can’t really understand French (despite years of instruction in elementary and high school). But I think that the first two sentences of this french wikipedia page** seems to oppose “hard sciences” to “human sciences”. Maybe that’s a more helpful pair of terms. We can pretend to be hard and not soft, or objective and not subjective, but it’s hard for anyone to put themselves on the side of the non-human and feel superior. At least that will be the case until the cylons take over.***

    *So for various reasons, one might conclude that the earth is at the centre of the universe. And that was a reasonable conclusion for a long time. The Ptolemaic model made decent predictions and gave the most reasonable explanations based on the knowledge people had at the time.

    The first two sentences are the only ones I’m referencing:
    Les sciences humaines et sociales sont un ensemble de disciplines étudiant divers aspects de la réalité humaine. On les oppose souvent aux sciences dures en raison de leurs sujets d’investigations.

    “I don’t want to be human! … I’m a machine. And I can know much more.”

  9. 14

    I’m embarrassed to ask this (and will be embarrassed by the errors in my previous comment when I find them) but…

    What do you mean by “dualism”?

    When you say “dualism”, what are the two things you have in mind?

  10. 15

    Hmm, radioisotope dating used a universal property of matter to determine a civilization did not wrap Buddy Jebus in a rag that ended up in Turin. That the shroud is the whimsy of some 14thC adherent of the longest of the long cons doesn’t imbue it with agency, any more than finding coins with.a Caesar on them makes them subjective. Carbon dating is used to verify the antiquity of manusvripts further ensuring what we read are actually the words & thoughts of a Caesar or Heron of Alexandria. Hard evidence iow.

  11. 16

    Anon1152 – iirc dualism is the idea that your brain and mind are distinct entities. The brain is the radio & the mind is the signal. From somewhere.

    Yes, even scientists can be biased. Science addresses that and these principles can be applied to other fields. People having agency doesn’t preclude objective study of the talking monkey or the introduction of bias by the observer.

  12. 17

    That we’re not there yet isn’t an argument that psychology isn’t a science. It’s a measure of the complexity of the task.

    Exactly. The progression of complexity from particle physics upwards is obvious.

    The Standard Model has just four kinds of particles; quarks, leptons, gauge bosons, and the Higgs, with well-defined interactions. Of course things can get complicated even there, with couplings varying at different energies, some energy ranges being hard to analyze computationally, etc, but the basic entities and interactions are really simple.

    Chemistry is hellishly more complicated, biology even more so. A human cell has, on the average, 1e14 atoms. There are about 4e13 cells in a human body, again, doing horribly complicated stuff. That ‘human’ as an entity in any given theory would have behaviour difficult to describe should come as no surprise. But working towards that description is certainly no less science than lower-level fields.

  13. 18

    Thanks al kimeea. So mind/body dualism, perhaps with a Rupert Sheldrake-ish spin? I hope I don’t have to commit to that to distinguish between sciences like physics and sciences like psychology… I think I still like the “hard sciences” vs “human sciences” distinction. It lets us acknowledge that “hard sciences” aren’t exactly inhuman, and that “human sciences” aren’t exactly easy (or unscientific).


    For some reason I’m reminded of an episode of “The Big Bang Theory” where Dr. Sheldon Cooper cries out “Geology isn’t a real science!” in front of a bunch of geologists with paintball guns…. https://www.youtube.com/watch?v=_CBQv8Y3V4o

Comments are closed.