I’ve been a Popperian for years. That’s not to say I’m a fan of Robert Popper (although I am); rather that I susbscribe broadly to Karl Popper’s philosophy of science. The basic tenet of it is that science advances only through making audacious predictions; that such predictions are ones that ought, given our existing experience, be easy to discredit; that hypotheses best serve science by being easy to disprove yet are not disproved; that scientific hypotheses worthy of the name can in a fundamental way never be really proven; that even the most successful hypothesis might one day face its black swan.
I’m only banging on about this again after all these years because I’ve just read Salmon’s Foundations of Scientific Inference, which talks about trying to solve Hume’s problem of induction: what logic permits us, what exist a priori in the universe that lets us make the leap from the sun rising today, or the pips coming through on BBC Radio 4 at 6pm, or indeed the fact that atomic fine structure worked last microsecond, from the confident prediction that all or any of this will happen again in the immediate future? I’m still not completely sure, by the end of Salmon’s book, that there is any, outside of the Cartesian idea that the will of God supports the entire edifice from time signal to time-and-space. Being an atheist, I can’t decide whether or not I should start believing in a divine power or just cancel our home contents insurance as a waste of money.
When I first discovered the idea of falsifiability, I loved the paradoxical way that it seemed to embrace nihilism and somehow conjure forth panism, or at any rate a fully formed empiricism. But at the time, researching in the physics laboratories, I could only dimly see its relationship to what I was doing from day to day. Most worryingly, I couldn’t quite square that principle of relentless falsification with the way that, when we obtained utterly unpredicted results from our apparatus (or simply null results, a dead signal), we wouldn’t tear up the rulebook of quantum physics with a Popperian flourish, but would instead investigate the electrics and the lenses, tweak the lasers, go back to the data, try again and see if it worked that time. Overturning the western scientific tradition every time the cameras were quiet would of course have felt like an overreaction to the circumstances, if not a throwing of toys out of the pram: but in a sense, if falsification is a key idea behind practical experiment, without which empiricism becomes positivism and eventually metaphysics, then adherence to that rule could surely be justified. Besides, if your philosophy is going to loosen its belt every time circumstances feel uncomfortable, then it just gets fat.
Nowadays I’m definitely fatter, but I think Salmon’s discussion of Bayesian priors and all the givens that tag along with the inductive process, and his talk of suppressed premises and the like, have given me a more pragmatic perspective on what was going on. The smooth running of our apparatus—the almost transparent transmission of atomically tiny physical events to our hulking experimenter selves—was in fact already bound up with our original hypotheses. When we predicted that we would see a single atom first brightly illuminated by a laser, then dark, then bright again, there was hidden in that single word “see” a whole range of engineering assumptions: we did not, after all, ever believe we might see it unaided (even though lucky experimenters on barium atoms could indeed sometimes manage that.)
That genuine apotheosis of scientific falsifiability, rather than technical fallibility, could only come about if we ruled out all those other potential falsifications tied up in the language of our scientific hypothesis. In a sense we’d trapped ourselves in our own word game, but that was OK: it’s one that researchers, as long as they continue to balance honesties about on the one hand their engineering skill and on the other the possible untruth of their pet theories, can embed themselves in without too much harm. It’s a forgetfulness that helps you carry out research without endless, obsessive recourse to the underlying fundamental philosophy of science. Once you’ve set up the scenario to test your unlikely hypothesis, then diligence will ultimately not risk the purity of the pursuit, and rarely ends up a waste of time.
I’ve no idea what the point of all of this is, except that perhaps not being a full-time physicist any more arguably means I have more time to think about these things that at the time would have made me feel much happier and more fulfilled about the experience. Empirical studies will only ever discover empirical truths, which are too humble themselves to claim any priority over other sorts. But falsification, its nose-thumbing hubris at the vacuum at the core of induction itself, lends something noble to what might ultimately, in the pursuit of the Ding an sich, be simply bootless labour. While “not only are the drops of rain mere appearances, but even their round shape—nay even the space in which they fall—are nothing in themselves, but merely modifications or fundamental forms of our sensible intuition, and that the transcendental object remains unknown to us…” while that might be true, it’s ultimately no criticism of those who direct attention to the arresting nature of the rainbow.