Do ducks lay eggs? How people interpret generic assertions

Do Ducks Lay Eggs? How People Interpret Generic Assertions Sangeet Khemlani 1 , Sarah-Jane Leslie 2 , Sam Glucksberg 1 , and Paula Rubio Fernandez 1 {khemlani, sjleslie, samg, prubio}@princeton.edu 1 Department of Psychology 2 Department of Philosopy Princeton University Princeton, NJ 08540 USA Abstract West Nile virus; if anything, the norm is for a mosquito to be virus-free. Many theorists have proposed variations on such an account (e.g. Asher & Morreau, 1995; Pelletier & Asher, 1997; Greenberg, 2003). These accounts are discussed at much greater length in Leslie (2007; forthcoming a). One might suppose that generics are interpreted as existen- tial statements, such that the generic ‘Ks are Fs’ is said to be true whenever some Ks are Fs. One might hope to thus explain the otherwise puzzling truth of ‘mosquitoes carry the West Nile virus.’ Yet this interpretation fails to account for generics such as ‘birds are female’, which is intuitively false even though, of course, some birds are female. As these examples suggest, generics cannot be accounted for in terms of quantifiers such as ‘all’, ‘most’, or ‘some’. Further, language acquisition studies suggest that while generics lack explicit quantification, children acquire the con- ceptual framework of generics and employ them in their speech at least a year before they use explicit quantifiers (Hol- lander, Gelman, & Star, 2002; Gelman, Geotz, Sarnecka, & Flukes, forthcoming). These observations suggest that generics are not quantifi- cational. Quantificational statements are about how much or how many in a way that generics are not. Notice that, upon being asked ‘how many tigers are striped’, one might reply ‘most tigers are striped’, or ‘some tigers are striped’, but one cannot reply ‘tigers are striped’. The generic is not an appro- priate answer to this question (Carlson, 1997). Leslie (2007; forthcoming b) argues at length that the truth and falsity of generics does not depend on how many of the relevant in- dividuals possess the predicated property. There is no sense, she claims, in which generics are dependent on such quantita- tive considerations. The results presented here offer empirical support for this claim. Generics are statements that express generalizations, such as ‘ducks lay eggs’. Intuitively, such statements seem true. Even the universal form of such statements e.g., ‘all ducks lay eggs’ seems true, despite our knowing that the majority of ducks do not. We conducted an experiment to verify these intuitions, and found that people overwhelmingly judged generic asser- tions true. People also judged universally quantified assertions true, but to a lesser extent. These results are consistent with the hypothesis that generics express cognitively primitive general- izations, and so require fewer cognitive resources than do ex- plicitly quantified assertions (Leslie, 2007). Hence when peo- ple encounter universally quantified assertions they treat them as generics in order to minimize cognitive effort. Keywords: generics; semantics; linguistics; philosophy Introduction Generics are statements such as ‘tigers are striped’, ‘cars have radios’ and ‘ravens are black’. Such statements lack explicit quantificational operators (e.g., ‘some’, ‘all’, or ‘most’) but nevertheless express generalizations, rather than claims about specific individuals. This is true not just for English, but for all known languages for which generics have been explored (Krifka et al., 1995; Dayal, 1999). Generic statements are interesting for a number of reasons. Though they occur fre- quently in everyday speech, it is not immediately apparent how generics are interpreted. They do not express excep- tionless universal generalizations; while it is false that all tigers are striped (a few unfortunate albino tigers are quite stripeless), the generic claim ‘tigers are striped’ remains true in the face of these exceptions. The generic claim ‘tigers are striped’ is thus not equivalent to its universal counterpart ‘all tigers are striped’. Perhaps, then, generics might be as- similated to the quantifier ‘most’ — certainly it is true that most tigers are striped. Such a proposal runs into difficulties when we consider generics such as ‘lions have manes’ and ‘mosquitoes carry the West Nile virus’. The mature maned males make up less than half of the lion population, so it is false that most lions have manes, yet the generic strikes us as true. Further, over ninety-nine percent of mosquitoes are per- fectly virus free, yet ‘mosquitoes carry the West Nile virus’ is a true generic. If one were tempted to understand generics as meaning “all normal”, the last two examples should give one pause for thought. There is nothing abnormal about mane-less fe- male lions, so it is surely not true that all normal lions have manes. Nor is it true that all normal mosquitoes carry the Leslie’s Account of Generics As the foregoing discussion might suggest, the truth condi- tions of generics are an extremely complex matter. Generic truth conditions are so divorced from quantitative considera- tions that ‘mosquitoes carry the West Nile virus’ is true even though less than 1% of mosquitoes carry the virus, while ‘books are paperbacks’ is false, even though over 80% of books are paperbacks. The philosophical and linguistic litera- ture on generics offers accounts that are often so baroque as to take a half a page to simply state. These accounts make use