Dialectical Argumentation for Reasoning about Chemical Carcinogenicity

We aim to build intelligent systems which can reason autonomously about the carcinogenicity of chemicals. Scientific debates in this area draw on evidence from multiple, and often conflicting sources, both theoretical and experimental, and participants use various modes of inferential reasoning. In seeking to automate such reasoning, we have first articulated precisely the multiple modes of inference used when an assertion of human carcinogenicity is made from experimental animal evidence. Because such inferences are often contested, scientific debate in this domain can be vigorous. To model such debates, we propose the use of a form of dialectical argumentation, drawing on Habermas’ philosophy of Discourse Ethics [9] and Pera’s philosophy of science [18]. The resulting formalism permits the representation of uncertainty and disagreement regarding the modes of inference used, as well as the claims being asserted.