Bibliography on Philosophy of Cosmology

Objects to the inference from fine-tuning to the claim that the universe is “improbable,” focusing in particular on what “probability” could mean in this context.

Suppose a gambler is asked, after witnessing one roll of a pair of fair dice, whether this is the first roll or one of many.  The fallacy: if the outcome is double sixes, there must have been many rolls. Hacking argues that the inference from fine-tuning to many universes in sequence (as proposed by Wheeler) commits this fallacy.

This paper attempts to determine the extent to which “free” parameters in our current best theories (cosmology and particle physics) are free to vary. Argues that, in a final theory, at least some of these parameters will remain free.

Discussion of various examples of fine-tuning in physics and cosmology, in favor of explaining why the universe is just so based on anthropic selection from a multiverse.

This is a classic work of philosophy in the fine-tuning literature. The main purpose of this book is to use modern physics—particularly cosmology—to argue for a creator God. The influence of Leslie’s treatment of fine-tuning and anthropic reasoning have extended past theology; in particular, Leslie proposes a number of vivid test cases (such as the firing squad) that continue to be used in evaluating proposals regarding fine-tuning and selection effects.

Anthology including 21 papers on design arguments, from biology as well as cosmology. Still the best single source including critics alongside advocates of the design argument, reflecting the impact of cosmology.

Fine-tuning arguments in cosmology, as well as similar arguments (such as the “Doomsday argument”) reflect a basic confusion. According to Norton, within a Bayesian account a low probability assignment is ambiguous between neutral evidence and counter-evidence.  Avoiding these fallacious inferences requires adopting a different, non-probabilistic representation of neutral evidence.

Accessible overview of answers to two fundamental questions:  Why is there a universe at all?  And why does this universe (with the specific features it has) exist?

Sober formulates the design argument in terms of the “likelihood principle”.  He contrasts his preferred formulation with other versions of the argument and reviews several critiques. He rejects the cosmological design argument as unsound because it fails to account properly for selection effects

 The inference from fine-tuning to multiple universes is fallacious, reflecting a mistake regarding the content of our evidence.  Our evidence is that: (i) this universeis life-permitting, not (ii) there is a universe that is life permitting.  The existence of multiple universes is irrelevant to the probability of (i), but not (ii).  White further argues that hypotheses that render particular facts less “surprising” are not necessarily confirmed by those facts.

Influential systematic review, following discussions of fine-tuning and anthropics in cosmology starting in the early 70s.  Formulates a wide variety of distinct “anthropic principles,” running the gamut from selection effects to substantive teleological and metaphysical claims.

Considers a wide variety of cases where selection effects need to be taken into account, including review of previous discussions and several new cases. Proposes and evaluates several candidate principles to be used in reasoning about selection effects, within a Bayesian approach.

In this article, the authors work within the Bayesian framework of epistemology to propose constraints on credence functions for self-locating propositions. They settle on a compromise rule for cases where no unique simplest measure can be defined, for which one takes a weighted average of simple measures. They propose that the compromise rule should hold in the case of eternal inflation, where the self-locating proposition is that we inhabit a 14-billion year old universe. Their argument is a response to the disagreement over the appropriate choice of measure over pocket universes in an eternal inflation scenario.

Earman distinguishes different forms of the anthropic principle (weak, strong, participatory, and final). Valid uses of anthropic reasoning can be interpreted as natural extensions of ordinary reasoning in cosmology, and are therefore neither absurd (as critics might think) nor revolutionary (as proponents sometimes claim). In particular, anthropic reasoning fails to place the existence of conscious observers in a starring explanatory role with regard to the history of the universe

Discussion of the “Sleeping Beauty” paradox, which illustrates the difficulties in reconciling Bayesian conditionalization with indexical information.

Accessible review of how cosmologists aim to extract predictions from eternal inflation, and the main obstacle to doing so (the “measure problem”).

Argues against the “selection fallacy” – that we should take ourselves as “typical” members of some reference class.  Instead we should explicitly acknowledge further assumptions regarding how typical we are as another component in Bayesian analysis of the evidence.

Masterful historical survey, making the case that two contrasting principles that have played a significant role in cosmology: first, appeals to our existence as a selection effect (anthropic principle), and second, a preference for theories that are insensitive to detailed features of the initial state (indifference principle).

Thorough criticism of approaches to anthropic reasoning that introduce new principles to handle “indexical” information, such as Bostrom’s Self-Sampling Assumption.  Bayesian conditionalization on the full non-indexical information available is sufficient to solve various puzzling cases discussed in the literature.

Careful critical assessment of fine-tuning and anthropic principles.  Defends the weak anthropic principle as the appropriate way that the nature of the human subject should be taken into account in cosmology, and compares this to Kant’s “Copernican Revolution”.

Defends anthropic reasoning as an important step in the assessment of physical theories, with extensive discussion of anthropics applied to the cosmological constant (reflecting on his own 1987 prediction, as well as subsequent work).

Outlines some of the main philosophical issues in cosmology, including: What issues will a theory of cosmology deal with? What kinds of causation will be taken into account as we consider the relation between chance and necessity in cosmology? What methodology will we use to test theories beyond the bounds of observations? Should cosmology weaken the need for empirical testing?

Ellis et al. construct a model of an “ideal” set of cosmological observations, which represent the best empirical access we can have to astromonical objects. They show that even this set is insufficient to determine the spacetime structure on the ideal observer’s past lightcone. However, if one assumes that the Einstein field equations accurately describe the dynamical structure of spacetime, then the full set of ideal observations is both necessary and sufficient to determine the causal history of the ideal observer. This is a practical limit on how precisely one can determine cosmic spacetime structure, even if one assumes the absolute validity of general relativity.

Merritt argues that the addition of dark matter and dark energy to the standard model of cosmology are ‘conventionalist’, in the sense of Karl Popper. They were added as auxiliary hypotheses in response to observations that would have otherwise falsified the standard model in its previous form. He argues that this indicates that cosmology is currently a degenerating research program, and that the concordance arguments in favor of ΛCDM are weak.

An argument that evidence in fact does have the power to uniquely determine theoretical hypotheses, using the quantum discontinuity as a case study. The arguments in this paper are useful to consider in the case of cosmology.

Masterful analysis of the nature of the knowledge achieved in celestial mechanics. On Smith’s view, post-Newtonian orbital astronomy focused on identifying robust physical sources for discrepancies between calculated orbits and observations, and the knowledge regarding these physical details and their causal impact survives the transition to general relativity. This paper provides a convincing counterargument to the pessimistic meta-induction argument against scientific realism. 

Even if there are no fully equivalent rivals to all theories, cases of transient underdetermination (theories equivalent with respect to available data) challenge scientific realism. Stanford makes a historical case that transient underdetermination is prevalent: we often fail to conceive of alternative theories that are equally successful with respect to some body of data, yet are only uncovered by later inquiry.

Critical assessment of the extent to which proposals such as inflationary cosmology constitute a “unification” of particle physics and general relativity.

Very large N-body simulations such as Millenium-I and Millenium-II have been used to make detailed predictions about the distribution and properties of galaxies and clusters of galaxies, based on cosmological parameters constrained by WMAP. This paper presents the methodology and tools – notably halos and subhalos tracking tools – used in Millenium simulations and details some of its results.

Excellent review of the astrophysical observations that constrains the nature of dark matter particles. Brooks details the consequences of these constraints for rival models to CDM; i.e., Warm Dark Matter and Self-Interacting Dark Matter; and the extent to which these rivals could solve the small-scale challenges that appear in galaxy formation simulations given these constraints.

This review paper briefly introduces the ΛCDM model at different scales. It insists on the problems the model faces, according to N-body simulations, at scales smaller than 1 Mpc and the extent to which better understood baryonic physics could solve them.

An extensive review of the history of simulations in cosmology and the challenges upcoming for this discipline. This review focuses on dark matter simulations, and is complemented by Baldi’s review referenced below, dedicated to dark energy N-body simulations and the investigation of cosmic inflation scenarios through them.

Although Parker’s work focuses on climate science simulations, the reader will find in the following two references – and in Parker’s work in general – an excellent discussion of the epistemology and metaphysics of simulations. This paper criticizes the claim that the materiality – or lack thereof – of computer simulations forbids them to qualify as experiments.

Parker discusses the analogues in computer simulations to Franklin’s strategies for building confidence in an experimental result. She emphasizes the epistemic privilege in the domain of computer simulations of the so-called ‘Sherlock Holmes’ strategy, consisting in ruling out alternatives to a given hypothesis.

Illustris is the first generation of very large ‘hydrodynamic’ simulations; i.e., simulations tracking the evolution of dark matter haloes but also taking into baryonic physics feedback. This paper constitutes an excellent introduction to the methodological challenges motivating Illustris and its limits.

This paper summarizes three of the major challenges that the Cold Dark Matter model currently faces at small-scales in predicting the abundance and demographics of dark matter halos. It also offers some perspectives on the extent to which one can hope to overcome these challenges with more realistic simulations.

Weisberg distinguishes three kind of models; mathematical, physical and computational – the one of interest to us here. The reader can find in this taxonomy and in the wealth of examples provided a good starting point to understand how to evaluate the reliability of this distinct kind of model.

This book constitutes a nice introduction to the new challenges arising with computer simulations: in chapter 2 Winsberg discusses how to verify and validate a simulation and the impact of simplifications made in the models; the role of simulations between equation-solving and replacing experiments in chapters 3 and 4; chapters 5 and 7 discuss the role of fictions in science and criticize the widespread idea that scientific success implies truth.