How do you see philosophy entering into your work on cosmology?
I suspect most scientists see philosophy as entering their field of inquiry only after they have done their daily work. Generally our approach is to respond to the literature: looking at new measurements, new observations, new ideas and trying to see whether your own expertise gives you a way of contributing to those developments, or even just trying to fit it all together into a clear picture that makes sense to you. It’s like solving a jigsaw puzzle − and only after you have done, do you stop and think “well was that the right jigsaw? What is a jigsaw anyway?” And so on. So for me, philosophical inquiry is a cleaning-up exercise after you have done your practical business − it is not actually a tool for doing the science.
Can you tell us about your current research in cosmology?
Almost everyone in cosmology is trying to solve the same problem: how did we get here? Some areas of physics are rich in unexpected phenomena, such as superconductivity. When you deal with sufficiently complex systems, they can often behave in a way that is unpredictable in the light of the fundamental laws and have to be treated as new ‘emerging’ phenomena (for example: the human brain operates according to the laws of quantum mechanics, but human consciousness is hardly predictable given just Schrödinger’s equation). That happens much less in cosmology: the universe in its initial conditions was in a certain simple state of small fluctuations around uniform density, and much of what we see around us is still very strongly formed by those initial conditions – on large scales, it is recognizably the same physical entity. And that is good in a way, because if we are interested in the question of how it all began, then by studying our universe today, we can get a fairly accurate picture of these initial conditions. So we are able to make a time machine of this sort just by observing the large-scale distribution of matter − i.e. making 3D maps of where all the different galaxies are located in space. We do this first by cataloguing galaxies in their 2D positions on the sky. Then we estimate a radius via spectroscopy: measuring the redshift of spectral lines that arises from the expansion of the universe, which is an effect that increases with distance. So then we have a 3D picture that lets us measure the size and density of superclusters, chains of galaxies hundreds of millions of light years long. With supercomputers, we can predict the patterns that would be expected if the universe started with particular initial conditions − especially those predicted by the ‘inflationary’ theory of initial conditions − and the match turns out to be impressively good.
Do initial conditions count as laws of nature in your view? For example, what do you think about initial conditions such as the Past Hypothesis?
Initial conditions are simplifying tasks: we can write down a recipe for how things were in terms of an amplitude of fluctuations and how that depends on scale, and the statistics of the process (whether it is random); and if that all works, we have reduced a complex thing to a simple thing − and then the next question is where does the simple thing comes from, and some people rush ahead in this process. We have a fairly good empirical knowledge of what the initial conditions must have been, and there is already a lot of debate about whether this knowledge generates a big problem, in the sense that the initial state must have had an unusually low entropy. Is such a statement to be considered a law in the same sense as Maxwell’s equations? When we say that the state had an “unusually” low entropy, we already seem to be implicitly suggesting a stochastic process where there is a set of alternatives, and somehow our universe is selecting one out of a number of possibilities at random. And then you can in principle ask whether the choice for our actual universe was unusual or unexpected out of the available range. But you have already smuggled in a lot of assumptions there. And the danger is some infinite regress: if you say that the initial conditions for our universe were chosen as one of many possibilities, then where do those possibilities come from, what set them up and so on. But in the end I think the only satisfying answer to this has to come from existing physics: rather than inventing some principle that supplements the laws that we have with some additional initial conditions that are treated as an axiom, you want a concrete picture of why those particular conditions arose.
Presumably there is the problem of testing initial conditions. Can novel predictions be used as a way of testing the validity of initial conditions?
For the last 30 years in cosmology we used to think of initial conditions in very special terms because we were dealing with models with a single origin, i.e. the Big Bang model of the universe. And everyone was very worried of having to specify quite a lot about how the universe was at a point with no way of ever being able to talk about what happened before that point. There was no way of having a physics-based discussion of how initial conditions could be set up – they were just a given. You could try to give them the status of a law of nature, where laws of nature are themselves just recipes from which you can calculate the behaviour of physical entities. Some initial conditions play the same role and as such, they can arguably be counted as laws. But the more information we have to write down as part of a law of nature, the less satisfied we are that this can be truly fundamental. We have that situation in particle physics where the standard model so far explains everything that the LHC has seen, but to do that you need to write down the value of 30 numbers, and these values seem pretty well random − so you would like (if possible) to see if these numbers are related to each other, so you can then treat a smaller set of numbers as fundamental. So initial conditions for the Big Bang come under this heading of having to supply what feels like too much information. But then what happened in 1980 was that the classical problems with the Big Bang, especially the issue of causality (the universe being created uniform, even though no light signals can reach from one end to the other), somehow ushered in inflationary theory, which attempts to explain where these apparent initial conditions at the Big Bang came from. Now, you could ask the same question about inflation: where does it come from? Does it start with a singularity? But at least inflation does a few things that make it feel like a genuine advance in explanation compared to the old view. Firstly, it lets you understand how the causality problem is solved, because the universe is older than we thought; secondly, it converts the singularity issue into something of a non-question, because events prior to inflation cannot be observed; but finally, it makes a number of testable predictions. Some of these are very specific, such as the generation of relic gravitational waves; for a moment in 2014, the BICEP experiment thought it had detected these, and this would have been a development of historic importance. But more broadly, some models of inflation deal with the initial conditions issue by being eternal: inflation continues forever on average, but it seeds bubbles where inflation ends − and our observable universe would be a small domain inside one such bubble. Because we have a multiverse of different bubbles, there is now the possibility of solving the problems of the small vacuum energy if these bubbles each have this set at a different level, allowing for the fact that structures capable of hosting observers will only grow when it is small. There is a good deal of debate about whether a model that predicts many causally disconnected bubbles is testable: how can we test it without experiencing these bubbles? Well, the vacuum density that we experience is measured, so a multiverse model that fails to predict this value is ruled out. So there is testability, but it is limited in scope. But astronomers are familiar with this restriction: we live with the data provided by the universe, and can’t treat it as a piece of experimental apparatus that can be modified to yield other results. Nonetheless astronomy is still science − I hope.