215 | Barry Loewer on Physics, Counterfactuals, and the Macroworld
What this covers
Barry Loewer and Sean Carroll discuss the foundations of statistical mechanics and its philosophical implications for understanding time, causation, and the nature of physical law. Loewer presents a unified framework in which the past hypothesis—the claim that the universe began in a low-entropy state—combined with a uniform probability distribution over compatible initial conditions yields what he calls the Mentaculus: a probability map from which the arrow of time, causation, counterfactuals, and even free will can all be derived. Central to his argument is a Humean account of laws: rather than governing forces imposed on the world, laws are the simplest, most informative descriptions of actual facts. This conception strips away the theological baggage inherited from seventeenth-century thinkers like Descartes and Newton, who understood laws as God's volitions.
The conversation ranges across why temporally symmetric fundamental dynamics nonetheless produce a directional arrow of time, how to evaluate counterfactual claims without invoking David Lewis's law-violating miracles, and whether the special sciences—biology, psychology, and others—introduce genuinely new laws or merely reorganize Mentaculus probabilities for human purposes. Loewer also addresses objections from anti-Humean philosophers who treat laws as extra fundamental entities, tackles the puzzle of free will by distinguishing what one can do conditional on macroscopic versus microscopic states, and considers complications that arise when applying his framework to quantum mechanics's many-worlds interpretation. Throughout, the discussion oscillates between technical philosophy of physics and the question of how to reconcile what science says the world is with how it appears to lived experience—a tension at the heart of what Wilfrid Sellars called the Manifest Image vs Scientific Image.
Loewer argues that statistical mechanics—specifically the past hypothesis combined with a uniform probability distribution over initial conditions (the Mentaculus)—is the foundational source from which the arrow of time, causation, counterfactuals, the special sciences, and even free will can all be explained, without positing laws as governing entities.
- Laws are best understood in Humean terms as the simplest, most informative descriptions of the actual facts, not as God-like governing forces.
- The past hypothesis (low entropy at the Big Bang) plus a uniform distribution over compatible microstates yields a probability map of the universe that grounds time's arrows and counterfactuals.
- Counterfactuals and causation can be analyzed via Mentaculus probability distributions over macrostates, avoiding Lewis's appeal to law-violating miracles.
Emergence depends on physics-based explanations for why macroscopic descriptions work despite microscopic incompleteness, locatable through different Hilbert space factorizations.
- The question of how higher-scale descriptions work so well despite not being microscopically complete is the crucial question of emergence; contrary to Loewer's appeal to biology and psychology, there are likely physics-based explanations for which macroscopic descriptions are usable, though a complete account does not yet exist.
“this is a crucially important question also, and it's the question of emergence. How do we come up with these higher scale descriptions that work so well, despite the fact that they're not microscopically complete? I am not actually gonna lean on biology or psychology, I think there are physics-based explanations for this”
- The fundamental stuff of the universe is sui generis—the world itself—but is mathematically represented as a vector in Hilbert space; this vector can be factorized in different ways, and there is an operational question about how best to factorize Hilbert space, which helps locate emergent descriptions of things.
“It's a vector in Hilbert space... No, it's the... The world is sui generis. It's the world. But it is mathematically represented as a vector in Hilbert space.”
The physics community ignored Feynman's low-entropy past hypothesis for decades despite its clear statement, not from skepticism but from failure to think carefully about it.
- The idea that the arrow of time requires a low-entropy past hypothesis was clearly stated by Feynman in the Feynman Lectures and Character of Physical Law, yet the physics community largely ignored it for decades—not because they disbelieved it but because they had not thought carefully about the question.
“it's right there in the Feynman lectures, it's there in the Character of Physical Law, and no one paid attention. Yeah, it's really, it's really bizarre.”
Physicists cannot avoid doing philosophy of physics and can only do it well or badly, as Feynman exemplified despite disparaging philosophers.
- You cannot avoid doing philosophy of physics; you can only do it badly—as illustrated by Feynman, who was actually quite good at it despite his explicit statements disparaging philosophers.
“I always say, you can't not do philosophy of physics, you can only do it badly, but he is actually pretty good at it, except for his explicit statements about it.”