Similarity, causality and argumentation William Jimenez-Leal (w.jimenezleal@uniandes.edu.co) Christian Gaviria (cgaviria@uniandes.edu.co) Departamento de Psicologia, Universidad de los Andes. Cra. 1 No 18A- 12, Edificio Franco, Bogota, 111711. Colombia Abstract Similarity is a notion that is widely used both in cognitive science and in argumentation theory. These research programs have, however, developed in large part separately and in consequence rely on disparate notions of similarity. Only recently there has been a proposal for specifying how similarity actually plays a role in judging slippery slope arguments. We present here further theoretical discussion and empirical evidence in order to show how similarity can play a role in slippery slope arguments and in argumentation in general. In the experiment presented here, we manipulated the availability of causal information, and showed that people are sensitive to it when judging arguments’ strength. We conclude that similarity between causal properties of the elements presented in arguments is crucial for arguments’ strength assessments. Keywords: Argumentation, similarity, causality, analogical reasoning. Introduction The degree of conviction that an argument generates depends on many elements. The effectiveness of some arguments seems to depend on the perceived similarity between the elements presented in the premises and the conclusions that might follow. For example, such is the case of the argument based on precedent, where the similarity between past events and the one under discussion is such as to warrant following the same course of action as with the precedent (Walton, 2010). Similarly, some arguments fail because the relation of similarity between premises and conclusion is weak. The fallacy of false analogy (Tindale, 2007) is one example, where there is a comparison between situations based on superficial similarities that do not support the conclusion. Walton, Reed & Macagno (2008) recognize that judgements of similarity between a class and an exemplar are key for the quality of arguments from verbal classification (from definition, vagueness, arbitrariness) (See also Macagno, 2009; Walton, 2009). The notion of similarity is thus central to explaining why people deem some arguments good or bad, and it is taken as a primitive element for explaining how people evaluate arguments. Similarly, the typologies of arguments put forward by perspectives like the dialectical (Walton, 2010) and the pragma-dialectical (van Eemeren, Houtlosser, & Snoeck, 2007), rely on identification of similarity. In the pragma- dialectical perspective, for example, one of the three main types of arguments is the ‘argumentation based on comparison” (van Eemeren, et al., 2007), where the argument and the standpoint argued for refer to different things but share a predicate. In the example “It is not at all necessary to give James a 10 dollar allowance, because his brother always got 10 dollars a week”, the similarity between James and his brother regarding the money needed, is the justification that allows one to proceed from premise to conclusion (Hitchcock & Wagenmans, 2011). In fact, the questions proposed to identify this type of argumentation scheme presuppose the notion of similarity (e.g. “Are there enough relevant similarities in the things that are compared?”) Similarity thus plays a dual role in argumentation: not only is it proposed that similarity judgments are performed by people engaged in argumentation, but it is also suggested that argumentation schemes are to be identified by questions that imply similarity judgements. That is, similarity plays a role both in explaining what people do, and also as a tool that the argumentation scholar needs to identify arguments and evaluate its correctness. Even though there has been vigorous research on the role of similarity in several psychological processes (Goldstone & Son, 2005), and despite argumentation research consistently using this construct as a tool to characterise several argumentation schemes (Walton et al, 2008), little work has been done to integrate the findings of cognitive science into our understanding of how people reason with arguments. In what follows, we will briefly examine the most common notions of similarity currently in use in cognitive science and consider the only work we are aware of that explicitly makes use of this idea to explain argument strength (Corner, Hahn & Oaksford, 2011). This will lead us to consider causality as one of the key ideas that is missing when using similarity as an explanatory principle. We will then present some empirical evidence to support our claims. Similarity and cognitive science It is difficult to overstate the importance of similarity as an explanatory tool in cognitive science. From categorisation to analogy, similarity judgements are advanced to explain very diverse phenomena. Links between rules and similarity as well as the very need of appealing to similarity in explaining cognition have been widely discussed (Sloman & Rips, 1998; Goldstone, Day & Son, 2010). It is more or less accepted that alternative ways of conceiving similarity capture different intuitions about our use of this notion, and that all have different weaknesses and strengths. The multiplicity of contexts in which it is possible to use the notion of similarity is consistent with the diversity of ways in which people judge that objects are alike. One can distinguish three main models to conceive similarity: geometric models, featural models and alignment based