Safety Logics II: Normative Safety.
John Bell, Zhisheng Huang · 1996
Abstract. In this paper we continue our analysis and formalisationof reasoning about safety. In a companion paper we distinguish be-tween absolute safety and normative safety, and use Dynamic Logicto develop a formal possible-worlds semantics and a logic for abso-lute safety. In this paper, we begin by outlining this theory. We thenintroduce Defeasible Dynamic Logic in order to give possible-worldssemantics and a logic for normative safety. We then define a prefer-ential entailment relation defined in order to represent commonsensereasoning about the normal termination of actions. We conclude witha discussion of the relationship between safety, obligation, rationalityand risk, and outline some extensions to the present work. 1 Introduction Attempting to develop a formal theory of reasoning about safety isa complex task; as reasoning of this kind may involve commonsensereasoning about action and change, time, beliefs, goals, obligations,rationality, etc. In an earlier paper we began this task by abstractingaway from many of these features in order to develop a formal the-ory which captures what we consider to be the essential features ofwhat we call absolute safety [9]. In order to make this paper self-contained, we outline the essential features of this theory in the nextsection. We then extend this theory to include normative safety. Webegin by introducing Defeasible Dynamic Logic and use this to givea possible-worlds semantics for normative safety. We then present alogic of normative safety. We then show how commonsense reason-ing about the normal termination of actions can be represented bymeans of a preferential entailment relation. In conclusion, we discussthe relationships between safety, obligation, rationality and risk, andoutline some extensions to the present work.