Model Evaluation: An Adequacy-for-Purpose View

Parker, Wendy S.

doi:10.1086/708691

Cited by 146 publications

(115 citation statements)

References 38 publications

Supporting

Mentioning

112

Contrasting

Unclassified

Order By: Relevance

“…Delimiting the domain of validity, we believe, is a genuine token of understanding. Models are not supposed to be reliable in any circumstances but, since they are only partial and idealised representations of the target phenomena, they should be adequate for the specific purpose (or set of purposes) at stake (Parker 2020). If scientists don't know the domain of validity of the model they are using, they encounter the risk of misusing it, e.g., of running the simulation in a physical domain in which the underlying model fails to apply.…”

Section: Delimiting the Domain Of Validitymentioning

confidence: 99%

“…Climate models are best considered with respect to their adequacy for a certain purpose (Parker 2020), and many climate models (in particular regional climate models relying on statistical techniques and climate models incorporating machine learning methods) may not have understanding as their primary purpose. From this point of view, it can be asked to what extent and in what sense these climate models can nevertheless provide some understanding; addressing these questions will allow for a more detailed characterization of the understanding gap in climate modelling (highlighted in Held 2005).…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Understanding climate change with statistical downscaling and machine learning

2020

View full text Add to dashboard Cite

Machine learning methods have recently created high expectations in the climate modelling context in view of addressing climate change, but they are often considered as non-physics-based 'black boxes' that may not provide any understanding. However, in many ways, understanding seems indispensable to appropriately evaluate climate models and to build confidence in climate projections. Relying on two case studies, we compare how machine learning and standard statistical techniques affect our ability to understand the climate system. For that purpose, we put five evaluative criteria of understanding to work: intelligibility, representational accuracy, empirical accuracy, coherence with background knowledge, and assessment of the domain of validity. We argue that the two families of methods are part of the same continuum where these various criteria of understanding come in degrees, and that therefore machine learning methods do not necessarily constitute a radical departure from standard statistical tools, as far as understanding is concerned. * We thank the participants of the philosophy of science research colloquium in the Spring semester 2020 at the University of Bern for valuable feedback on an earlier draft of the paper. We also wish to thank the participants of the seminar 'Philosophy of science perspectives on the climate challenge' and the workshop 'Big data, machine learning, climate modelling & understanding' in the Fall semester 2019 at the University of Bern and supported by the Oeschger Centre for Climate Change Research. JJ and VL are grateful to the Swiss National Science Foundation for financial support (grant PP00P1_170460). TR was funded by the cogito foundation.

show abstract

Section: Delimiting the Domain Of Validitymentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Understanding climate change with statistical downscaling and machine learning

2020

View full text Add to dashboard Cite

show abstract

“…In an attempt to offer a narrowed-down and flexible characterization of the epistemic value of creativity in scientific practice, I associated such value with its potential to clarify a tradition, frequently suggesting fruitful ways of improving it, and sometimes also helping to fulfil exploratory aims. However, there is a plurality of other epistemic values at play in scientific practices (Brown 2013;Douglas 2013;Potochnik 2015;Parker 2020). So it might be the case that a creative model is not the most desirable type of model in certain epistemic circumstances, and that a creative methodology is not the most appropriate methodology with regard to certain goals.…”

Section: Plurality Of Epistemic Values In Sciencementioning

confidence: 99%

“…In fact, scientific models are only valuable to very specific respects (epistemically, socially, ethically), and even to each of these respects, only in very concrete ways. For this reason, philosophers like Parker (2010Parker ( , 2020 argue that to talk about the mere adequacy of a scientific model is rather ambiguous, insofar as the good performance of the model in the past (or regarding certain epistemic aims) does not guarantee its good performance in the future (or regarding other epistemic aims). Instead we should always think in terms of "adequacy-for-purpose" (Parker 2020: 457;see also Poznic 2018;Potochnik 2015;Douglas 2013).…”

Section: Plurality Of Epistemic Values In Sciencementioning

confidence: 99%

Novel & worthy: creativity as a thick epistemic concept

Sánchez-Dorado

2020

Euro Jnl Phil Sci

View full text Add to dashboard Cite

The standard view in current philosophy of creativity says that being creative has two requirements: being novel and being valuable (to which a third intentionality requirement is often added; Sternberg and Lubart 1999; Boden 2004; Gaut 2010). The standard view on creativity has recently become an object of critical scrutiny. Hills and Bird (2018) have specifically proposed to remove the value requirement from the definition, as it is not clear that creative objects are necessarily valuable or creative people necessarily praiseworthy. In this paper, I argue against Hills and Bird (2018), since eliminating the element of value from the explanation of creativity hinders the understanding of the role that creative products play in actual epistemic practices, which are fundamentally normative. More specifically, I argue that the terms ‘creativity’ and ‘creative’ function as thick epistemic concepts when employed by competent epistemic agents in practice, that is, these concepts have both a descriptive and an evaluative content that cannot be disentangled from one another. Accordingly, I suggest that philosophers should prefer thick accounts over thin accounts of creativity. A thick account of creativity is one that endorses the standard view at its basis, but further develops it in two ways: by stressing the entanglement of the value and novelty requirements; by permitting to encompass a range of domain-specific characterizations of such entanglement for different epistemic situations. In order to take the first steps in the development of such a thick account of creativity, I look at the domain of scientific practices as a case in point, and try to spell out what the thickness (or entanglement of novelty and worth) of creative instances typically entails here. Namely, I identify the worthy novelty of creative models and methods with their potential to clarify a tradition, with fruitfulness, and with the fulfilment of exploratory aims.

show abstract

“…That is, models are not just representations; they are also tools, which are constructed or selected, and manipulated, with an eye toward achieving specific epistemic or practical purposes (e.g., Morrison and Morgan 1999 ; Giere 2004 , 2010 ; Boon and Knuuttila 2009 ; Parker 2010 ; Knuuttila 2011 ; Currie 2018 ; Boon 2020 ). A closely-related view of model evaluation proposes that models be judged in terms of their adequacy or fitness for particular purposes, rather than by comparison to some abstract standard of perfect representation (Parker 2010 , 2020a ; Currie 2018 ; see also Teller 2001 ; NRC 2007 ; Taper et al 2008 ).…”

Section: Introductionmentioning

confidence: 99%

Data models, representation and adequacy-for-purpose

Bokulich

Parker

2021

Euro Jnl Phil Sci

View full text Add to dashboard Cite

We critically engage two traditional views of scientific data and outline a novel philosophical view that we call the pragmatic-representational (PR) view of data. On the PR view, data are representations that are the product of a process of inquiry, and they should be evaluated in terms of their adequacy or fitness for particular purposes. Some important implications of the PR view for data assessment, related to misrepresentation, context-sensitivity, and complementary use, are highlighted. The PR view provides insight into the common but little-discussed practices of iteratively reusing and repurposing data, which result in many datasets’ having a phylogeny—an origin and complex evolutionary history—that is relevant to their evaluation and future use. We relate these insights to the open-data and data-rescue movements, and highlight several future avenues of research that build on the PR view of data.

show abstract

Model Evaluation: An Adequacy-for-Purpose View

Cited by 146 publications

References 38 publications

Understanding climate change with statistical downscaling and machine learning

Understanding climate change with statistical downscaling and machine learning

Novel & worthy: creativity as a thick epistemic concept

Data models, representation and adequacy-for-purpose

Contact Info

Product

Resources

About