Ability and knowledge: from epistemic transition systems to labelled stit models.

Area of Science:

• Formal epistemology and logic
• Artificial intelligence and multi-agent systems
• Philosophy of action and ability

Background:

• Distinguishing between knowing *that* one can achieve a goal and knowing *how* to achieve it is crucial in understanding ability.
• Causal ability refers to having the power to bring about an outcome, while epistemic ability relates to knowing the means to achieve it.
• Existing formalisms for modeling ability often struggle to capture this nuanced distinction.

Purpose of the Study:

• To formally model and differentiate between causal and epistemic ability.
• To establish a correspondence between epistemic transition systems and labelled stit models for representing ability.
• To demonstrate the enhanced expressiveness of labelled stit logic over epistemic transition systems.

Main Methods:

• Utilizing epistemic transition systems to represent states of knowledge and actions.
• Employing labelled stit (STIT) models, a formalism for strategic ability, to capture agency and control.
• Developing mappings between the language and structures of epistemic transition systems and labelled stit models.

Main Results:

• A formal framework is established that successfully models both causal and epistemic conceptions of ability.
• A strong correspondence is demonstrated between epistemic transition systems and labelled stit models.
• The extended labelled stit logic proves to be more expressive than the logic of epistemic transition systems.

Conclusions:

• The study provides a unified approach to understanding different types of ability within formal logic.
• The findings contribute to the development of more sophisticated models of agency and knowledge in artificial intelligence.
• The enhanced expressiveness of labelled stit models offers new possibilities for analyzing strategic reasoning and control.