The Directions of Time: how physics develops its temporal asymmetries

The fundamental laws of physics do not make reference to the difference between past and future. If some sequence of states of the world solves the equations of fundamental physics, so does the very same sequence, run backwards. In occasional situations in large-scale physics (think of the motion of the planets, or of a table-tennis ball) the same is (approximately) true. But in most situations studied in science - and most situations encountered in the everyday world - there is a clear distinction between past and future. Ice cubes in a drink melt; they do not spontaneously form. Fires consume things; they do not spontaneously create them. Hot coffee cools down.

Since we normally think - for good reason - that the facts of the macroscopic world can ultimately be explained by the laws of the microscopic world, there is a paradox here. How can the past/future distinction emerge at the large scale unless it is already present at the small? This profoundly puzzled the physicists who, at the start of the 20th century, developed statistical mechanics, the branch of physics that allows us to work out the laws governing large-scale phenomena - like heat and pressure - given only the laws governing microscopic particles. Despite the practical success of statistical mechanics, it continues to puzzle physicists and philosophers today: there is no generally-accepted explanation of where the difference between past and future originates. The philosophy of statistical mechanics, to which this project aims to contribute, is concerned above all else with this continuing puzzle. Its implications in physics are profound and widespread, and it matters well beyond physics, influencing our approach to philosophical problems such as the distinction between cause and effect, or between memory and anticipation.

Too much research in philosophy of statistical mechanics, though, has fixated on the puzzle to the extent that it ignores the positive achievements of statistical mechanics, which are enormous. Whatever the conceptual difficulties of the subject, its methods succeed, with great reliability and great accuracy, at predicting just how large-scale phenomena turn out. This has been largely missed by a subject whose question has been 'how in principle could microscopic physics lead to a direction of time?' and not 'how can we understand, justify, make sense of the actual methods used by physicists to construct time-directed laws from a starting point which lacks that direction?'

This project aims to rectify this mistake. Its goal is to study, in some detail, just what the quantitative methods are that physicists use to perform calculations and make predictions in statistical mechanics, and to see exactly what must be added or assumed to make those methods philosophically justified. The hope is that doing so will give the insight that is needed to really succeed in understanding statistical mechanics, and through it, the direction of time.

Research in this field generally has impact beyond academia largely through its interface with popular science (in published and broadcast form) and I anticipate that the project may, in the long term, have an indirect impact outside academia in this manner. However, it would be disingenuous for me to assert that it would directly have any such impact, as it's significantly too technical for a lay audience.