Skip to main content
Log in

Generic Generalizations in Science

A Bridge to Everyday Language

  • Published:
Erkenntnis Aims and scope Submit manuscript

Abstract

This article maintains that an important class of scientific generalizations should be reinterpreted: they have typically been understood as ceteris paribus laws, but are, in fact, generics. Four arguments are presented to support this thesis. One argument is that the interpretation in terms of ceteris paribus laws is a historical accident. The other three arguments draw on similarities between these generalizations and archetypal generics: they come with similar inferential commitments, they share a syntactic form, and the existing theories to make sense of them are alike. Once these generalizations are properly understood as generics, the recent cognitive approach to generics can be extended to the study of the relevant sciences. The last section indicates ways in which this extension is fruitful for the two strands of research that we combine: the philosophy of science literature on generalizations and the semantics literature on generics.

This is a preview of subscription content, log in via an institution to check access.

Access this article

Price excludes VAT (USA)
Tax calculation will be finalised during checkout.

Instant access to the full article PDF.

Similar content being viewed by others

Notes

  1. “There is evidence that” is a hedge, see below.

  2. When George Lakoff coined the term ‘hedges’, it referred to those “words whose job is to make things fuzzier or less fuzzy” (Lakoff 1973, p. 471), where ‘fuzzy’ is to be taken in the logical sense. We are not directly concerned with this reading.

  3. Of course, this conflation is exactly what economists are sometimes accused of making (Morgan 2012, pp. 405–409).

  4. There are also reasons to reject the attribution of ‘loose talk’ to WUGS. For arguments in this direction about generics, see Nickel (2016, pp. 26–30).

  5. We will not discuss in this article generics that are said to be habituals, e.g., Mary smokes after lunch, and kind predicating generics, e.g., Dinosaurs are extinct.

  6. For a comprehensive presentation of the multiple attempts to analyse the semantics of generics, see Leslie and Lerner (2016).

  7. Nickel (2009, 2016) acknowledges the need to take the properties and kinds at stake into account, but does so with set-theoretical tools. As a result, his truth-conditions for generics are not oriented towards the cognitive traits of the utterer.

  8. Leslie’s project, like other empirically informed research in semantics, involves a transition from the study of patterns in the use of language to a theory about the truth conditions guiding the patterns. In her early work, Leslie used hesitantly the terminology of ‘truth conditions’ (see her distinction between “semantic truth conditions” and “worldly truth makers” in Leslie 2008, p. 43), but has been more outspoken recently (e.g., Lerner and Leslie 2016, p. 406).

  9. In formal terms, we can write: “Ks are F” is true if and only if (NAC ∧ (CDF ∨ SF ∨ EPF)).

  10. We thank an anonymous referee for helping us articulate this important distinction, which is not made in Leslie’s work.

  11. Nickel (2010, sect. 3.1) discusses such similarities, except he dubs as “open-ended” what we construe as being non-monotonic. Unterhuber (2014) acknowledges the non-monoticity of inferences from generics (insofar as he accepts Delgrande’s logic for generic sentences) as well as their tolerance towards exceptions.

  12. It is neither in On the Definition of Political Economy (Mill 1844) nor in Book VI of A System of Logic (Mill 1886), but he used the phrase 16 times in a non-philosophical context in his Principles of Political Economy (Mill 1848) and his well-known methodological characterization of economics as a deductive, inexact science is a primary influence on Cairnes (1888, see especially endnote 21), who seems to be the first to use ceteris paribus in methodological work. For examples of explicit use of the ceteris paribus clause prior to this period, see Persky (1990, pp. 87–89) and Reutlinger et al. (2015, sect. 2).

  13. The general pattern that we found is that such phrases are used in the context of discussing theoretical models or results of statistical estimation. In this context, the phrase ‘ceteris paribus’ refers in a straightforward fashion to what would happen if the value of one variable in an equation was changed while keeping intact the other variables (for examples, see Cingano 2014, p. 28; Halteret al. 2014, p. 84). We also note that some scholars have an unusual fondness for the phrase (e.g., Piketty 2014 who uses it nine times in his admitedly lengthy best seller).

  14. The context condition is important: if interpretive defaultness was not a function of the context, we could not explain why unquantified generalizations about mathematical models (see Sect. 2.3) are not generics although they share their syntactic form. We submit that, in this case, the generalization can be unmarked because, in the context of specialist-to-specialist conversations over mathematical systems, there is no great risk of misinterpretation when the quantifier is omitted.

  15. A recent version of a statistical interpretation of cp laws is provided by Roberts (2014). For a detailed survey of the cp literature with references to numerous accounts of each type, see Reutlinger et al. (2015).

  16. These are what he calls absolute generics, but there are also relative generics. For detailed presentation, discussion, and refutation of Cohen’s view, see Nickel (2009, sect. 5, 2016, chap. 4.2), Leslie (2008, pp. 7–13), and Leslie and Lerner (2016, sect. 2.5).

  17. Cognitive traits always play a role insofar as they determine what counts as negative.

  18. If we were to focus on Leslie’s notion of ‘strikingness’, political and moral commitments would seem a more promising place to start the empirical enquiry since what is deemed ‘dangerous’ depends on what we value. Due to space constraints, we do not further explore this hypothesis here.

  19. There is indeed increasing psycholinguistic evidence that generics are easier to process than quantifiers, even for adults (Meyer et al. 2011; Leslie et al. 2011).

  20. This phenomenon of “defaulting to generic” has already been documented in children and in adults (Leslie 2012, pp. 36–41).

Bibliography

Download references

Funding

This research has been financially supported by the Canada Research Chair in Applied Epistemology (Grant Number 950-230644).

Author information

Authors and Affiliations

Authors

Corresponding author

Correspondence to François Claveau.

Rights and permissions

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Claveau, F., Girard, J. Generic Generalizations in Science. Erkenn 84, 839–859 (2019). https://doi.org/10.1007/s10670-018-9983-x

Download citation

  • Received:

  • Accepted:

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1007/s10670-018-9983-x

Navigation