- How did science become distinct from philosophy?
- Who was most responsible for the shift?
Maybe more than anyone else, Hume was responsible for separating science from philosophy. He argued that nothing in philosophy is derived from reason alone; you need to start with axioms.
Even with sound reasoning, someone practicing philosophy can reach very different conclusions about the universe or human behavior or anything, depending on his axioms.
By definition, axioms are chosen, which means they’re sort of arbitrary. Hume chose for his philosophy a couple of axioms that have always been popular and practical:
- there is a causal relationship between events — any event must have cause(s)
- the laws that determine these cause/effect relationships are universal and immutable — they apply everywhere and for all time
He called his philosophy empiricism. It basically was and still is the philosophy of science. It says that you shouldn’t try to get knowledge by reason alone — you need evidence as well. He acknowledged that its axioms couldn’t be derived, but felt they were the most practical ones to use.
The first implies that laws connect events into cause/effect pairs. The best way to know what the laws are is to experiment, i.e. create conditions where cause/effect pairs happen in a controlled way and study them until a theory can be made to describe the law(s) that connect them.
The second implies that an experiment that shows that X causes Y under conditions Z performed today in Mexico City, will also show that X causes Y under conditions Z tomorrow in Oakland, and so on. This means that doing experiments is useful, because they produce knowledge that is generalizable.
The first doesn’t need justifying, because a world without cause/effect relationships isn’t something we can imagine. The second is shakier — there’s nothing that says laws can’t change from one day to the next. Hume argued that although there’s nothing that could stop them from changing, they probably won’t change, and he based his assumption on the idea of induction. In math, induction is a method for proving a statement S(n) for all natural numbers N, in two steps:
- show that the statement is true for n = 1
- show that if the statment if true for n, then it must be true for n + 1
Hume applied this to the immutability of physical laws. He said that when we look back in time we see that laws were the same 100 years ago as they are now, and the fact that the laws have gone so much time without changing imbues them with a sort of inductive inertia, and makes it unlikely that they will change soon. Sort of hand-wavy, but this idea is popular and practical. No one but a crazy person jumps off a cliff to test whether the laws of physics have changed — the result will be painful, because that’s how it always is.
That this idea sounds banal goes to show that all of us are scientists in a way =)
Skipping ahead 200 years we get to Karl Popper, who was known for trying to demarcate the problem domain of science. He wanted to answer this question: What assertions can be evaluated by science?
In Popper’s view science is a tool for assessing the truth value of assertions, but he realized that many assertions, while perhaps true and useful, can’t be evaluated by science, and thus have no scientific value. He came to the conclusion that an assertion must be falsifiable in order to be scientific. The more that an assertion tries to “evade” falsification, the less scientific it is, and assertions that evade falsification entirely have no scientific value. Science is agnostic on these assertions.
Here are some unfalsifiable assertions:
- A tree falling in a deserted forest makes no sound
- Somewhere between Earth and Mars is a china teapot that goes in an elliptical orbit around the Sun, but it is too small for even our most powerful telescopes to reveal
- What goes around comes around
- All the ones made by string theory
Hume’s small axiom set and Popper’s criteria help science avoid the knots that philosophy gets into — in Popper’s version, science doesn’t make pronouncements on assertions it can’t evaluate. What Hume knew, and what anyone that reads philosophy knows, is that axiom sets change from one philosopher to another. Many times this means that when philosophers set out to critique one another, they’re comparing apples with oranges, which is frustrating for yours truly.
Although it came long before, I see Occam’s Razor, aka the principle of parsimony/succinctness, as a sort of footnote to Popper’s criteria.
Occam’s razor is a truth-finding heuristic which says that among competing hypotheses, the one that makes the fewest assumptions is often the best. It’s based on a frequent correlation in good explanations between style and utility, form and function, elegance and truth. Here’s an example of what it says about a poor hypothesis — namely that homeopathy is an effective way to treat a gnarly stomach infection.
Your friend says that you should drink drops of some homeopathic solution, which you do, and your infection doesn’t improve. He says no, you were doing it wrong, you have to shake the bottle before drinking the drops, so you do, but still there’s no improvement. He asks you what you’ve been eating, and you say
lots of tacos chicken broth and apple sauce, and he says of course, that’s the problem, the chicken mixed with the apple is deactivating the solution… By this time you’ve given up on the drops, so you take a course of antibiotics and the infection goes away.
This isn’t proof that homeopathy is no good at treating stomach infections. After all, your friend might be right, and maybe the tacos you weren’t eating and the music you were listening to and the cloudy weather were harshing the essence of the drops. But it’s easy to see that if your friend keeps giving ad hoc reasons for why the drops don’t work, their supposed effectiveness can evade scrutiny forever. Popper says that as your friend does this his assertion’s scientific value (SV) approaches zero. Occam’s razor says something similar — all those caveats make your friend’s hypothesis awkward and probably untrue.
For Popper, the theories with highest SV are very “falsifiable”, i.e. they can be tested for falsification in many ways, and they have withstood most of the attempts at falsification. Newtonian gravity is a great example — it can be tested just about anywhere, and for hundreds of years it was resilient and reliable (it still is in everyday settings). Eventually astronomers observed things, like the precession of Mercury’s orbit, which weren’t explained by Newtonian gravity. In part, observations like these led to Einstein’s general relativity. This doesn’t mean Newtonion gravity is a bad theory — its SV is great, but it’s been superseded by something better. Similarly, physicists today don’t imagine that general relativity is the final word on gravity, but it’s the best they can do for the moment.