Argumentation by Analogy and Weighing of Reasons

ISSA Conference Proceedings , 2023

The thesis of this paper is that what characterises argumentation by analogy is that it is based on a comparison of argumentative relationships. I distinguish two types: intraargumentative relationships and interargumentative relationships. By the former I mean the relationship between what is presented as a reason and the claim that reason allegedly support. This is usually marked by expressions such as 'so,' 'therefore,' 'because,' 'consequently,' etc. By interargumentative relationships I mean the relationship between two or more reasons. This is usually marked both by coordinative locutions such as 'in addition,' 'on the other hand,' 'moreover,' etc., and by adversatives expressions such as 'but,' 'although,' 'having said that,' and so on. On the basis of this, two varieties of argumentation by analogy are distinguished: argumentation by parity of reasons and argumentation by parity of weighings.

Human reasoning applies argumentation patterns to draw conclusions about a particular subject. These patterns represent the structure of the arguments in the form of argumentation schemes which are useful in AI to emulate human reasoning. A type of argument schema is that what allow to analyze the similarities and differences between two arguments, to find a solution to a new problem from an already known one. Researchers in the heavily studied field of analogies in discourse have recognized that there is not a full and complete definition to indicate when two arguments are considered analogous. Our proposal presents an initial attempt to formalize argumentation schemes based on analogies, considering a relationship of analogy between arguments. This will contribute to the area increasing such schemes usefulness in Artificial Intelligence (AI), since it can be implemented later in Defeasible Logic Programming (DeLP).

AILACT Essay Prize., 2022

Winning paper for the 2022 AILACT Essay Prize.

In this chapter, I inquire into the structure of analogical arguments. I begin by considering several historical meanings of “analogy”, understood as a semantic relation, an act of cognition, and a kind of argument. I proceed then to provide a general characterization of arguments from analogy and identify four essential aspects thereof: the problem situation, the prima facie similarity, the relevant similarity, and the solution. In the remaining part of the chapter I analyze these aspects in more detail.

The reasoning process of analogy is characterized by a strict interdependence between a process of abstraction of a common feature and the transfer of an attribute of the Analogue to the Primary Subject. The first reasoning step is regarded as an abstraction of a generic characteristic that is relevant for the attribution of the predicate. The abstracted feature can be considered from a logic-semantic perspective as a functional genus, in the sense that it is contextually essential for the attribution of the predicate, i.e. that is pragmatically fundamental (i.e. relevant) for the predica-tion, or rather the achievement of the communicative intention. While the transfer of the predicate from the Analogue to the analogical genus and from the genus to the Primary Subject is guaranteed by the maxims (or rules of inference) governing the genus-species relation, the connection between the genus and the predicate can be complex, characterized by various types of reasoning patterns. The relevance relation can hide implicit arguments, such as an implicit argument from classification , an evaluation based on values, consequences or rules, a causal relation, or an argument from practical reasoning.

Argumentation , 2023

In this article I defend what I call a 'particularist approach to arguments by analogy.' Particularism is opposed to generalism, which is the thesis that arguments by analogy require a universal principle that covers cases compared and guarantees the conclusion. Particularism rejects this claim and holds that arguments by analogy operate on particular cases. I elaborate on two ideas that support this position. On the one hand, I contend that an analogy can be seen as a parallelism of argumentative relationships, drawing on the distinction between similarity and analogy (Gentner 1983) and on the meta-argumentative account of arguments by analogy (Woods and Hudak 1989). On the other hand, I argue that universal principles are not necessary neither for the analysis nor the evaluation of arguments by analogy (Govier 1989) and that, rather than being a requirement, they can be seen as by-products of good analogies.

Argumentation Library, 2015

The paper contributes to the debate about arguments by analogy, especially the distinction between 'deductive' and 'inductive' analogies and the question how such arguments can be 'deductive', yet nonetheless defeasible. It claims that 'deductive' and 'inductive' are structural, not normative categories, and should not be used to designate argument validity. Based on Aristotle's analysis of enthymemes, examples, and metaphors, it argues that arguments from analogy are complex arguments that involve inductive, abductive, and deductive components.

Systematic Approaches to Argument from Analogy, ed. H. J. Ribeiro, Heidelberg, Springer, 2014, 23-40.

This paper shows why a double scheme approach to argument from analogy is justified. It shows how the one scheme applies to the famous violinist case of argument from analogy in philosophy. It shows how factors are weighed in systems of case-based reasoning. It analyzes a famous case of argument from analogy in legal rhetoric using a notion of similarity using script-based technology from artificial intelligence.

Reasoning and Public Health, 2015

The Philosophical Review, 2012

To the extent that the worth of scientific or philosophical efforts can be assessed by the number of productive research avenues they open up, this is definitely an important book. It deserves careful consideration by scientists, mathematicians, psychologists, and philosophers. Since it does not fit neatly into any usual category but rather stands athwart many research areas, its reception may depend on precisely who attends to its bold claims. This book aims to answer two questions: "What criteria should we use to evaluate analogical arguments used in science?" and "How can we provide a philosophical justification for those criteria?" (ix). Paul Bartha recognizes that analogies are widely used in all areas of human action-but claims: "We have no substantive normative theory of analogical arguments" (3). He persuasively argues that none of the theoretical approaches to analogical argumentation that previously have been developed is generally applicable. But he holds that the uses of analogies in science and mathematics are "key or 'leading' special cases that provide an excellent basis for a general normative theory" of analogical reasoning (3). This book proposes a systematic theoretical treatment, and a set of evaluation criteria, that (Bartha claims) apply to all varieties of analogical reasoning-both in science and elsewhere. This assertion is not modest, but careful arguments support it well. The claim seems quite plausible. Analogical arguments involve "source" (S) and "target" (T) domains that are similar to each other in certain respects. Positive analogies occur when property P and relation R pertain to domain S , and corresponding property P * and relation R * pertain to T. If the target domain T has feature A * but the source domain S lacks that feature (so that , A applies to S), this constitutes a negative analogy. The question at issue is: Under what conditions (and with what degree of confidence) would it be correct to infer that if S has a feature Q , then T has a corresponding feature Q * ? In favorable cases deductive reasoning may lead to conclusions that are considered correct with a high degree of certainty. In contrast, analogical reasoning at its best leads to results that are 'plausible'-that is "they have some degree of support" (15). Plausibility can be interpreted probabilistically, so that plausible statements are understood to have a rather high probability of being true, and additional relevant evidence may increase that probability. In