Two years ago, I was doing a Philosophy course in CTYI and we’d just come across Hume’s problem of induction, which basically invalidates science by saying we can’t conclude anything from experiments because there’s no guarantee that the laws of the universe didn’t change between the experiment and now. I was wearing a labcoat at the time as CTYI’s resident Scientist, and so all eyes turned to me. Oops. So here I am, finally reading a book about the philosophy of science.
I was pretty unsure about picking this up. Over Christmas break, I went back to my favourite bookshop, Waterstones in Drogheda, and bought a book I knew I’d have fun reading because it was by Ben Goldacre (review here), and this. I knew this might be less enjoyable, but thought it was important to get some kind of grip on the philosophy of science since I’m so big into science.
The first chapter of the book was my favourite, because it got right to the heart of what I was reading the book for -- the induction problem. How do we know science means anything? You can test a drug on 100,000 people and find it safe, but what if it kills the 100,001st person?
It goes into one way of getting around that, Karl Popper’s falsificationism, which states that science should always be trying to disprove hypotheses since you can never actually prove them, but disproving them does bring you closer to the truth by process of elimination. There are two big problems with that: (a) it means that the theory of relativity is on the exact same footing as my hypothesis that Theresa May is a lizard, because evidence for a hypothesis means nothing and neither of them have been conclusively disproven yet*, and (b) the fatal flaw is that the experiments you do to disprove something are generally based on other theories -- if you’re trying to prove that neutrinos move faster than the speed of light, you need to use a whole bunch of complicated methods to measure the speed and compare it and even find the neutrino, all of which rest on previous theories which in turn rely on inductive reasoning.
It went downhill from there, mainly because the first chapter was exactly what I wanted out of the book (well, apart from a solution to the problem), and the other chapters were different.
The book has two parts. The first part opens with that chapter and then goes into how we demarcate the boundaries between fields like science, history, economics and homeopathy (this was annoying -- it devil’s advocated for homeopathy by using the argument that it works by catering to each individual patient’s needs rather than average patients, and I don’t see how that makes it science), paradigm shifts and scientific realism.
The scientific realism chapter asks: is science the truth? Is it even approaching the truth? It examines one argument for that (the No Miracles argument, which says that surely if science wasn’t true we’d have run into problems since we rely on it all the time), and rebuts two arguments against it (underdeterminism and pessimistic induction).
Underdeterminism argues that we can never have enough data to decide between one theory that explains the results and has X implications, and another theory that explains the results but has completely opposite implications. He addresses that by saying, well, that just goes back to Descartes’ “there’s a demon making you imagine everything and nothing except yourself is real”, so it doesn’t specifically affect science. It’s just saying the world may not be as we think it is.
Pessimistic induction says that science is continually wrong -- Archimedes was disproven, then Newton was disproven (to an extent), eventually all will be disproven. Scientists drop old theories and replace them with new ones. It’s basically saying science is never fully right...although you could also look at it another way and say it’s always on the up.
The author has qualms about the No Miracles argument because it pretty much says that only a tiny percentage of false theories should be successful. But there are vastly more possibly incorrect theories than correct ones, so probability says that even if most true theories, most successful theories are probably false.
I didn’t like the second half as much -- it’s talking about the bearing of science on the philosophical questions in our lives, not the bearing of philosophy on science, which is what I bought it for.
It argues that science has values and has an interesting but seemingly irrelevant chapter called Human Kindness, where I was introduced to the concepts of psychological altruism (doing something that benefits others/being selfless) and biological altruism (doing something that benefits others and harms yourself). It’s possible to be psychologically altruistic without being biologically altruistic.
It then talks about nature vs nurture and whether humans have free will. The free will part was quite interesting and, again, disconcerting. Firstly, there’s the causal nexus argument, which says that we’re just atoms tumbling along in the universe and everything happens because of the knock on effects of whatever happened at the start of the universe. Our consciousness doesn’t have some magical influence -- it just does what the laws of physics make our atoms do.
Then there’s the neuroscientific argument from tardiness, based on experimental results. One of them, from a 2008 paper by John-Dylan Hanes et al, showed that researchers using brain scans could predict (with middling accuracy) the action of a participant 10 seconds before the participant had made the choice to act.
These experiments measure an increase in neuronal activity called the readiness potential which occurs just when someone is getting the urge to perform a “spontaneous” action. Experimenters asked participants to note when they felt the conscious urge to act and also measured the time of the action, and found that the readiness potential spike occurred 350 milliseconds before the urge, which is 200 milliseconds before the action.
That said, people might just have predecided that they’ll move when they feel some kind of signal in their mind, and it’s not certain that RP is actually the intention to move, it can just be about deciding things. Another experimenter could predict whether subjects could choose the left or right button 60% of the time up to 10 seconds in advance by monitoring their brain using fMRI scanning.
The book closes with a chapter on the reach of science, essentially just saying that facts aren’t everything and some things need to be learned through experience.
In short: It was an interesting book and the author is clearly very knowledgeable about the subject, which, as with Superintelligence, I appreciated. I liked how there were more twists and turns to the arguments presented than I would’ve thought of, and how the arguments were solid and not straw men. The second half of the book was less relevant but definitely interesting in parts. Would recommend if you’re looking to get a more nuanced version of the purpose and meaning of science.