- How hypotheses and experimental design rely on theory
Originally, I was just going to write one post on the logic underlying studies in the sciences. By that I mean the ways in which scientists determine how to set-up experiments such that the results can actually tell them something about the phenomenon under investigation. After reading what I wrote, I decided that it was too dry and I could shorten it if I introduced some notions in a much more interesting example (and demonstrate how different sciences rely on fundamentally different principles when carrying out their research). This, then, is the story of the Rise and Fall of Physics: Quantum Mechanics (or How Modern Physics Requires Researchers to be Insane).
Young’s experiments with light
Way back before science was science (when it was natural philosophy, and Newton was a giant, and not really standing upon giants’ shoulders), light was already a problem. Newton believed it was made up of parts (the “corpuscular theory” of light), but couldn’t manage to “prove” it. At the turn of the 19th century, Thomas Young sought to show this theory of light was wrong and that light was a wave. He wrote a couple convincing papers, but 1) most physicists were still treating Newton’s ideas as gospel and 2) he didn’t have experimental evidence. So, in 1803, he devised a way to “prove” light was a wave.
Waves in classical physics aren’t “things”. A wave of water or a sound wave are really effects on physical substances (like vibrations). More technically, waves propagate through a medium, which is a fancy way of saying that because to make e.g., a water wave you need some kind of force that acts on the water, causing the water molecules to act in a particular way. The ideal example here is water ripples. These are technically waves, and you can understand Young’s experiment better if you think about ripples. Imagine that you and a friend dropped to big stones into a pond. Ripples would cascade outwards from the points the rocks hit, and you and your friend dropped them close enough for the ripples “hit” one another. You get an interference effect like this:
That’s actually a drawing of water wave interference that Young made. The problem he faced was showing the light did this kind of thing too. Here is his description of how he performed his experiment in 1803 as described in front of the Royal Society:
I made a small hole in a window shutter, and covered it with a piece of thick paper, which I perforated with a fine needle. For greater convenience of observation I placed a small looking-glass without the window shutter, in such a position as to reflect the sun’s light, in a direction nearly horizontal, upon the opposite wall, and to cause the cone of diverging light to pass over a table on which were several little screens of card paper. I brought into the sunbeam a slip of card, about one-thirtieth of an inch in breadth, and observed its shadow, either on the wall or on other cards held at different distances. Beside the fringes of color on each side of the shadow, the shadow itself was divided by similar parallel fringes, of smaller dimensions, differing in number, according to the distance at which the shadow was observed, but leaving the middle of the shadow always white. Now these fringes were the joint effects of the portions of light passing on each side of the slip of card, and inflected, or rather diffracted, into the shadow. For, a little screen being placed a few inches from the card, so as to receive either edge of the shadow on its margin, all the fringes which had before been observed in the shadow on the wall, immediately disappeared, although the light inflected on the other side was allowed to retain its course, and although this light must have undergone any modification that the proximity of the other edge of the slip of card might have been capable of occasioning.
That “diffraction” pattern (the “fringes”) were not something particles can do. It was clear, convincing evidence that light was a wave. So what did the scientific community do? Mostly ignore Young and continue to trust in Newton (“God said ‘let Newton be’ and all was light”- Pope. Apparently ol’ Alex Pope didn’t pick a very good metaphor). However, by the middle of the 19th century Maxwell not only sealed the deal on light, but did so by explaining in terms of classical electromagnetism. Light wasn’t just a wave, it was an electromagnetic wave.
Waves were fundamentally different from “particles” (matter) in numerous ways. For example, they don’t “move” in straight lines, they aren’t localized in space, and they don’t collide like physical “things”, they interfere with one another. Constructive interference is what happens when two waves “meet” and form a bigger wave, whereas destructive interference results in a smaller wave (I’m being extremely liberal with terms here). Those very expensive earphones that dampen our cancel out external noise work by using destructive interference: “combining” incoming sound waves with sound waves just right such that they cancel each other out. If you and your friend (the one who helped you with the rock experiment) managed to throw snowballs such that they hit in midair, bits of snow would go flying everywhere. What would never happen is for the snowballs to disappear. They are made up of particle, and so we will never see wave interference. So said the physics community.
Einstein and light “quanta”
But while physicists were all congratulating each other on having pretty much wrapped up physics (nothing left but, to paraphrase Lord Kelvin, two small problems). Part of one problem was called the photoelectric effect. Physicists performed several experiments shining light on metal and observing how this caused electrons to shoot out of the metal. I’ll spare you the details of why this was a problem; suffice it to say, a little known physicist whose name you’ve probably never hear (Einstein-something-or-other) won a Nobel prize for explaining these experimental results. How? By showing that light was a particle (composed of “quanta” or “parts”).
Now we have a problem. On the one hand, we have a century of experimental results (not to mention an entire theory) showing light was a wave. Now we have other empirical data showing it’s a particle. This is impossible, so what was the problem? The problem was that nobody realized that the hypothesis that light was either a wave or a particle, as well as the experimental evidence for both hypotheses, all relied upon theory. Classical physics said everything was either made of articles or was a wave, so these two possibilities were the only ones anybody tried to demonstrate. Nobody thought to test whether light was something else. Nobody thought that, as it turns out, nothing is made out of “particles” (rather, what we now call “particles”, like electrons or photons, are really more like classical waves, but as we go from the subatomic level to the atomic to the molecular and beyond, the wave-like effects quickly become so small that we don’t even have the technology to detect them).
The Takeaway:Hypotheses don’t come out of thin air, and neither do the experiments performed to test them. This is the most dramatic example of how this is true. It didn’t just show us an example in which two empirical findings showed conflicting results because the theory behind the wave vs. particle hypothesis was wrong and all experiments designed to determine which one light was therefore doomed to fail. It tore down most of physics and completely rearranged our notions of reality.
Today, most experiments are far more complex and really a great deal more on theory (as do hypotheses). Experimental designs in psychiatry, for example, involve carefully designing clinical trials to determine the efficacy of e.g., X medication to treat Y mental illness. But all psychiatric disorders were defined into existence and remain entirely defined by symptoms (in fact, the biological evidence suggests that, at best, whatever pathologies underlie mental illness, they don’t match our diagnoses). Studies in neuroscience, sociology, psychology, etc., are mostly concerned with testing hypotheses about phenomena that we’ve defined to exist (such as religiosity, political position, intelligence, etc.). Most of the time, we’ve good reason for defining these constructs. But even in such cases, the fact that we are researching phenomena that we’ve defined into existence (or at least demarcated into existence) can create enormous problems. Take memory: Are there 256 different kinds of memory? The ways in which we’ve categorized different types of memory as well as the changes in our ability to test the relationship between brain activity and memory have resulted in conflicting definitions and models. Thus memory research can differ because researchers are using different classification schemes and designing experiments to test the “types” of memory that exist in their particular classification.
Research design starts with theory (or theories). Nobody is designing experiments to test whether light is a wave or a particle anymore, not because the question was finally decided but because it turned out nothing is either a particle or a wave, and our experiments were constructed based upon false theoretical assumptions.