Philosophers of science have offered various accounts of climate model evaluation which have largely centered on model-fit assessment. However, despite the wide-spread prevalence of process-based evaluation in climate science practice, this sort of model evaluation has been undertheorized by philosophers of science. In this paper, I aim to expand this narrow philosophical view of climate model evaluation by providing a philosophical account of process evaluation that is rooted in a close examination of scientific practice. I propose dynamical adequacy as a (...) metric by which scientists test and evaluate models that represent and produce key regional climate processes and features. I argue that process-based evaluation confirms the adequacy of a regional climate model for simulating and predicting future changes of a specific regional climate feature. I offer a case study of how, in practice, scientists establish the reliability of model projections by assessing the dynamical adequacy of such models. I also show how process-based evaluation mitigates some well-known shortcomings of model-fit assessment and supports the adequacy and reliability of climate model projections against philosophical objections, like confirmational holism. Such adequacy is especially important for decision makers who need reliable regional model projections to guide their climate change mitigation and adaptation policies. (shrink)

Many philosophers have argued that a hypothesis is better confirmed by some data if the hypothesis was not specifically designed to fit the data. ‘Prediction’, they argue, is superior to ‘accommodation’. Others deny that there is any epistemic advantage to prediction, and conclude that prediction and accommodation are epistemically on a par. This paper argues that there is a respect in which accommodation is superior to prediction. Specifically, the information that the data was accommodated rather than predicted suggests that the (...) data is less likely to have been manipulated or fabricated, which in turn increases the likelihood that the hypothesis is correct in light of the data. In some cases, this epistemic advantage of accommodation may even outweigh whatever epistemic advantage there might be to prediction, making accommodation epistemically superior to prediction all things considered. (shrink)

Computer simulations of the formation and evolution of large-scale structure in the universe are integral to the enterprise of modern cosmology. Establishing the reliability of these simulations has been extremely challenging, primarily because of epistemic opacity. In this setting, robustness analysis defined by requiring converging outputs from a diverse ensemble of simulations is insufficient to determine simulation validity. We propose an alternative path of structured code validation that applies eliminative reasoning to isolate and reduce possible sources of error, a potential (...) path that is already being explored by some cosmologists. (shrink)

We provide a novel articulation of the epistemic peril of p-hacking using three resources from philosophy: predictivism, Bayesian confirmation theory, and model selection theory. We defend a nuanced position on p-hacking: p-hacking is sometimes, but not always, epistemically pernicious. Our argument requires a novel understanding of Bayesianism, since a standard criticism of Bayesian confirmation theory is that it cannot represent the influence of biased methods. We then turn to pre-analysis plans, a methodological device used to mitigate p-hacking. Some say that (...) pre-analysis plans are epistemically meritorious while others deny this, and in practice pre-analysis plans are often violated. We resolve this debate with a modest defence of pre-analysis plans. Further, we argue that pre-analysis plans can be epistemically relevant even if the plan is not strictly followed—and suggest that allowing for flexible pre-analysis plans may be the best available policy option. (shrink)

In climate science, climate models are one of the main tools for understanding phenomena. Here, we develop a framework to assess the fitness of a climate model for providing understanding. The framework is based on three dimensions: representational accuracy, representational depth, and graspability. We show that this framework does justice to the intuition that classical process-based climate models give understanding of phenomena. While simple climate models are characterized by a larger graspability, state-of-the-art models have a higher representational accuracy and representational (...) depth. We then compare the fitness-for-providing understanding of process-based to data-driven models that are built with machine learning. We show that at first glance, data-driven models seem either unnecessary or inadequate for understanding. However, a case study from atmospheric research demonstrates that this is a false dilemma. Data-driven models can be useful tools for understanding , specifically for phenomena for which scientists can argue from the coherence of the models with background knowledge to their representational accuracy and for which the model complexity can be reduced such that they are graspable to a satisfactory extent. When citing this paper, please use the full journal title Studies in History and Philosophy of Science. (shrink)

This is the second of three parts of an introduction to the philosophy of climate science. In this second part about modelling climate change, the topics of climate modelling, confirmation of climate models, the limits of climate projections, uncertainty and finally model ensembles will be discussed.

Over the last twenty-five years, climate scientists working on the attribution of climate change to humans have developed increasingly sophisticated statistical models in a process that can be understood as a kind of calibration: the gradual changes to the statistical models employed in attribution studies served as iterative revisions to a measurement(-like) procedure motivated primarily by the aim of neutralizing particularly troublesome sources of error or uncertainty. This practice is in keeping with recent work on the evaluation of models more (...) generally that views models as tools for particular tasks: what drives the process is the desire for models that provide more reliable grounds for inference rather than accuracy to the underlying mechanisms of data-generation. (shrink)

The last few years have seen increasing research interest in moods and atmospheres. While this trend has been accompanied by growing interest in the history of the wordStimmungin other disciplines, this has not yet been the case within philosophy. Against this background, this paper offers a conceptual history of the wordStimmung, focusing on the period from Kant to Heidegger, as this period is, presumably, less known to researchers working with notions like mood, attunement or atmosphere today. Thus, considering this period (...) might provide conceptual resources not yet considered in current debate.Stimmunghas the remarkable feature of encompassing the entire semantic field of mood and atmosphere, insofar as both subjects and objects can literally be inStimmung.Stimmungmight refer to the state or condition of being attuned, which is understood as a dispositional state, as well as the process or act of attuning, which includes self-activating and foreign-determined forms of attuning. The word was first used for the tuning of musical instruments, but was quickly transferred to the fields of aesthetics, psychology, and physiology. This paper will focus on the contrast between the psychological canonization ofStimmungas a type of mental state, and the use ofStimmungas an untranslatable, irreducible metaphor with unique semantic force allowing for original theorizing. (shrink)

Many examples of calibration in climate science raise no alarms regarding model reliability. We examine one example and show that, in employing Classical Hypothesis-testing, it involves calibrating a base model against data that is also used to confirm the model. This is counter to the "intuitive position". We argue, however, that aspects of the intuitive position are upheld by some methods, in particular, the general Cross-validation method. How Cross-validation relates to other prominent Classical methods such as the Akaike Information Criterion (...) and Bayesian Information Criterion is also discussed. (shrink)

Climatologists have recently introduced a distinction between projections as scenario-based model results on the one hand and predictions on the other hand. The interpretation and usage of both terms is, however, not univocal. It is stated that the ambiguities of the interpretations may cause problems in the communication of climate science within the scientific community and to the public realm. This paper suggests an account of scenarios as props in games of make-belive. With this account, we explain the difference between (...) projections that should be make-believed and other model results that should be believed. (shrink)

The paper discusses modelling uncertainties in climate models and how they can be addressed based on physical principles as well as based on how the models perform in light of empirical data. We ar...

The central question of this thesis is how one can learn about particular targets by using models of those targets. A widespread assumption is that models have to be representative models in order to foster knowledge about targets. Thus the thesis begins by examining the concept of representation from an epistemic point of view and supports an account of representation that does not distinguish between representation simpliciter and adequate representation. Representation, understood in the sense of a representative model, is regarded (...) as a success term. That is, a representative model is one relatum in a relation of adequate representation (Chapter 2). When a representative model represents a target, it allows users of this model to learn something about the target. It is argued that a representative model has this epistemic function because it shares relevant features with the target. This presupposes a similarity view of representation. Similarity views of representation face serious objections, which will be rebutted (Chapter 3). One way of spelling out a similarity view of representation is to defend an indirect view of representation. In this thesis, which does not argue for an indirect view, it is assumed that the indirect view is a good option, if not the best, for articulating the similarity view. It is demonstrated how such an indirect view can be expanded to account for cases of technological modeling. A case study in bioengineering is used to show that the indirect view of representation must acknowledge a distinction between two directions of fit in relations between vehicles and targets. In this context the notion of design is interpreted as a relation between a vehicle and a target, thereby connecting ideas from philosophy of science with ideas from philosophy of technology (Chapter 4). Fictionalist accounts of models are intended to tackle the issue of the ontology of models. In this thesis, however, two prominent fictionalist accounts are discussed from an epistemological point of view in light of the central question regarding how one can learn about targets by using models. This question is addressed from the standpoint of Waltonian fictionalism. The result of the discussion is that the two discussed Waltonian fictionalist accounts cannot sufficiently answer the question. These accounts are criticized for their inability to deliver a satisfactory epistemology of representation (Chapter 5). Although Waltonian fictionalism is criticized, the present thesis also shows that the foundational theory of Waltonian fictionalism, the theory of make-believe can nevertheless be used to account for the distinction between projections and predictions that is made by the Intergovernmental Panel on Climate Change (Chapter 6). (shrink)

I examine error diagnosis (model-model disagreement) in climate model intercomparisons including its difficulties, fruitful examples, and prospects for streamlining error diagnosis. I suggest that features of climate model intercomparisons pose a more significant challenge for error diagnosis than do features of individual model construction and complexity. Such features of intercomparisons include, e.g., the number of models involved, how models from different institutions interrelate, and what scientists know about each model. By considering numerous examples in the climate modeling literature, I distill (...) general strategies (e.g., employing physical reasoning and using dimension reduction techniques) used to diagnose model error. Based on these examples, I argue that an error repertoire could be beneficial for improving error diagnosis in climate modeling, although constructing one faces several difficulties. Finally, I suggest that the practice of error diagnosis demonstrates that scientists have a tacit-yet-working understanding of their models which has been under-appreciated by some philosophers. (shrink)

Mathematical models are often expected to provide not only predictions about the phenomenon that they represent, but also explanations. These explanations are answers to why-questions and particularly answers to why the predicted phenomenon should occur. For instance, models can be used to calculate when the next total solar eclipse will happen, and then to explain why it will take place on July 2, 2019. In this regard we can obtain explanations from a model if we can solve the model equations (...) which govern the phenomenon under study. But some equations have no explicit solution or are too complicated to solve. In these cases it is difficult for a... (shrink)