Truth is the aim of inquiry, but some theories seem to better approximate that goal than others. One may think, for instance, that Copernicus’ model of the solar system is closer to the truth than Ptolemy’s. Since Copernicus assumed that planetary orbits were circles, his theory is, strictly speaking, false. Thus, we have here a case of a false theory which better approximates the truth than another false theory. Moreover, Copernicus’ theory also seems better than the plain claim that planets move in orbits: so, false propositions may sometimes be more verisimilar even than true but uninformative ones. This suggests that verisimilitude (or truthlikeness, as it is also known) is a mixture of truth and informative content: a theory is verisimilar when it tells many things about the world, and many of these things are true. Explicating this intuition in a rigorous way amounts to solving the so-called logical problem of truthlikeness. In normal circumstances, we simply don’t know what the truth about a given matter is. Thus, we cannot directly assess the relative verisimilitude of competing theories. However, we can estimate their verisimilitude on the basis of available evidence: explaining exactly how this can be done means answering to the so-called epistemic problem of truthlikeness. Interestingly, estimated verisimilitude doesn’t follow the rules of probability. In particular, it may be that a theory with low probability is estimated as highly truthlike; and, a logically stronger theory may be estimated as more verisimilar than a logically weaker one. This makes (estimated) verisimilitude a mathematically less well-behaved notion than probability; still, it provides the philosopher of science with a flexible and useful tool to analyze such issues as theory change, scientific progress, convergence to the truth, and the realism/antirealism debate.
|Key works||The notion of verisimilitude was first introduced, and clearly distinguished from probability, by Popper 1962 (see also Popper 1972 and Popper 1983), who also provided the first formal definition of truthlikeness. Ten years later, Tichý 1974 and Miller 1974 independently proved that Popper’s definition was untenable. This triggered the still ongoing post-Popperian research program on explicating verisimilitude, which currently include a number of different, and partially conflicting, proposals. These can be classified within three different perspectives (Zwart 2001, Oddie 2014): the content approach (Miller 1977, Miller 2006; Kuipers 1987), the likeness or similarity approach (Oddie 1986, Niiniluoto 1987), and the consequence approach (Schurz & Weingartner 1987, Schurz & Weingartner 2010). All such approaches address at least the logical problem of truthlikeness; for discussions of the epistemic problem see especially Niiniluoto 1987 and Kuipers 2000.|
|Introductions||Niiniluoto 1998 provides an excellent survey and discussion of the main accounts of verisimilitude; a more recent survey is given by Oddie 2014. Book-length treatments include Oddie 1986, Niiniluoto 1987, the collection edited by Kuipers 1987, Kieseppä 1996, Kuipers 2000 (parts III-IV), and Zwart 2001. Cevolani & Tambolo 2013 is a gentle introduction to the methodological applications of the notion of truthlikeness.|
Using PhilPapers from home?
Create an account to enable off-campus access through your institution's proxy server.
Monitor this page
Be alerted of all new items appearing on this page. Choose how you want to monitor it:
David Bourget (Western Ontario)
David Chalmers (ANU, NYU)
Rafael De Clercq
Ezio Di Nucci
Jack Alan Reynolds
Darrell P. Rowbottom
Learn more about PhilPapers