ABSTRACT
Chater, 2007), David Over (2009) announced the new paradigm in reason-
ing. This nomenclature had such clear intuitive appeal as a description of
recent developments in the psychology of reasoning that it has already found
its way into the textbooks (Manktelow, 2012). There are two central features of the new paradigm. First, it abandons binary truth functional logic and
adopts probability logic in its place as the computational-level model of how
people should reason (e.g., Adams, 1998). Second, it adopts a dual (or tri-)
process view of the cognitive architecture in which reasoning is carried out
(Evans, 2007; Stanovich, 2011). We have discussed the relationships between
these two aspects of the new paradigm elsewhere (Oaksford & Chater, 2011,
2012). In this paper we explore how the new paradigm deals with non-
monotonic or dynamic inference. As Oaksford and Chater (2012) argued, the principal motivation for both the probabilistic approach (e.g., Oaksford
& Chater, 1994, 2001, 2007) and the dual process approach (e.g., Evans,
2002) has been problems with non-monotonic inference.
