Action Types in Stit Semantics

Abstract

Stit semantics grows out of a modal tradition in the logic of action that concentrates on an operator representing the agency of an individual in seeing to it that some state of affairs holds, rather than on the actions the individual performs in doing so. The purpose of this paper is to enrich stit semantics, and especially epistemic stit semantics, by supplementing the overall framework with an explicit treatment of action types. We show how the introduction of these new action types allows us to define a modal operator capturing an epistemic sense of agency, and how this operator can be used to express an epistemic sense of ability.

Epistemic Oughts in Stit Semantics

“…Although we will work with this treatment of indistinguishability throughout the paper, it is worth noting that indistinguishability is typically cast as a relation between indices of evaluation, which, in the case of stit semantics, are moment/history pairs, rather than moments. The decision to treat indistinguishability as a relation between moments is a simplification, equivalent to understanding indistinguishability as a relation between moment/history pairs, in the more standard way, but subject to the (C4) constraint from Horty and Pacuit (2017).…”

“…But this objection turns on an ambiguity. There is a sense, captured by the truth of 3[α stit: W], in which you do have the ability to win-this is referred to in Horty and Pacuit (2017) as the causal sense of ability. But since you do not know whether the coin is heads up or tails up, there is another sense of ability-the epistemic sense-in which you do not have this ability, and it is this latter sense that seems to be crucial for assessing the legitimacy of criticism.…”

“…A robot might execute the action type of raising its 15. Although we focus here only on the approach of Horty and Pacuit (2017), it should be noted that this approach continues a line of research initiated by several computer scientists on epistemic concepts of action and ability in stit semantics; see, for example, Broersen (2011), Herzig and Troquard (2006), and Lorini, Longin, and Mayor (2014). This line of research was itself motivated by the problem of arriving at a satisfactory analysis of ability for agents with imperfect information in game theory and multi-agent systems; see, for example, Ågotnes (2006), Jamroga and van der Hoek (2004), and Schobbens (2004).…”

“…18. The perfect information constraint at work here is the constraint (C3) from Horty and Pacuit (2017).…”

“…Eric Pacuit and I have recently argued, in Horty and Pacuit (2017), that it is helpful to enrich the standard framework of stit semantics with an explicit set of Epistemic Oughts in Stit Semantics • 73 action types, in addition to the action tokens already present, in order to represent an epistemic sense of ability; the result is a new framework that we refer to as labeled stit semantics, where each action token is assigned a label, indicating the type of which it is a token. What the current paper shows is that an appeal to action types is likewise helpful in the analysis of epistemic oughts.…”