Readers may recall that I've been writing of late about Deborah Mayo, a professor at Virginia Tech.
To restate, Mayo has written Statistical Inference as Severe Testing; a book discussing statistics, induction, and the philosophy of science. Much of the book is obscure to me, though I skim through what I have to in order to get to stuff where I can gain something.
To move forward nonetheless: near the middle of the book she has a discussion of the Higgs boson. Allow me to talk about that today.
Back in 1964 Peter Higgs predicted the eventual discovery of a certain particle that would be responsible for other particles having mass. This became known as the "Higgs boson" or, more grandly as the "God particle"!
Over the following decades the Higgs boson came to be regarded as an important piece of the Standard Model of particle physics, and the fact that no one had experimentally confirmed its existence became something of a scandal.
Then came 2012, and CERN, the European Organization for Nuclear Research, announced that it HAD discovered the Higgs boson after all. Yippee! Except that scientists don't really like completing and confirming models. They like shaking things up, and the existence of a particle that was exactly what Higgs had ordered up soon came to seem too boring to get them all out of bed in the morning.
As Mayo reminds us, the tests at the CERN lab in Switzerland seemed to confirm the existence of the previously hypothetical particle. The discovery inherently involves the construal of statistics, because there are trillions of particle collisions underway, and the existence of a Higgs boson turned on what is signal in all of this, what is background noise.
The publicity declared that the Higgs boson had been discovered by a 5-sigma event. This means that they were looking at data from the Large Hadron Collider that was more than 5 standard deviations away from the mean. That is impressively unlikely to be an accident or, in this case, depressingly unlikely to have occurred in the absence of a Higgs.
In the midst of laying this out, Mayo cites a fellow statistician/philosopher named Dennis Lindley. Right after CERN's announcement, Lindley asked: "Why such an extreme evidence requirement? We know from a Bayesean perspective that this only makes sense if (a) the existence of the Higgs boson has extremely small prior probability and/or (b) the consequences of erroneously announcing its discovery are dire in the extreme."
Five sigma means, to be precise, that only one experiment in 3 million would have seen an apparent signal that strong in the absence of a Higgs boson. If we were confident that only one Packers-Vikings game in 3 million would see a Vikings victory we would never bother watch the game. Why do particle physicists set the bat quite this high?
Mayo's view is that we don't need to invoke Bayesianism for this. Part of the answer is that the positive result (with lesser sigma) might have been triggered by something not exactly a Higgs boson, something Higgs-like. Indeed, even with all five sigma in place with the July 2012 data there were "hints" of an anomaly -- the possibility that it was a Higgs-plus, something that would do more than fill out the standard model, could take particle physics beyond the standard model.
The exciting revolutionary move BSM didn't happen. Mayo's discussion of this line of experimentation ends with a sense of disappointment. "The promising bump or 'resonance'disappeared as more data become available, drowning out the statistical implications." The real news here is not just that Higgs was verified but that the reality of the Higgs-plus boson was falsified. It was the drive to and beyond five sigma that achieved both of those things, though, and Lindsey's query is ... unhelpful for practitioners of the science.
I think I've got that right. At any rate: it is fascinating, is it not?
And I think she is making a point about how the distinction between confirmation on the one hand and falsification on the other is not as clear a cut as one might have been led to think by the old Vienna Circle debates.
Well presented, CF. I read something on this, years ago.The author moved from the west coast to my neck of the woods---good move, IMHO. He is a *big picture* guy. By this point, you know who I am talking about. The God particle, I think, is fascinating. However, if Physics, Mathematics, Science and Philosophy cannot come to grips with this, we have either *made it up*, or, theorized the hell out of it, or presented an irresolvable paradox. I like the big picture guy a lot. If he is still with Johns Hopkins, I wish him the best.
ReplyDelete