580 California St., Suite 400
San Francisco, CA, 94104
Key research themes
1. How can geometric and mathematical properties of event representations explain qualitative reasoning about causality, action control, and learning generalizations in events?
This theme investigates the cognitive and mathematical constraints on representing events as mappings between action spaces and result spaces. It focuses on properties like monotonicity, continuity, and convexity, which underpin causal thinking, control of action, and generalization, respectively. Understanding these constraints provides insights into qualitative causal reasoning and event semantics that differ from traditional temporal segment-based accounts.
Key finding: This paper proposes a geometric two-vector model of events involving an action space and a result space, and identifies three mathematical constraints—monotonicity supporting qualitative causal reasoning; continuity... Read more
Key finding: Building on the two-vector geometric event model, this paper argues that human reasoning about actions and their outcomes is fundamentally based on expectations grounded in geometric and topological properties—rather than... Read more
Key finding: The paper bridges reasoning about actual causation with formal action representation (using action languages and Answer Set Programming), distinguishing laws governing states and events causing transitions, and showing that... Read more
2. What formal logical frameworks effectively support hierarchical and abstract action reasoning, including refinement of abstract actions into concrete plans?
This research theme focuses on extending dynamic logic and related formalisms for representing and reasoning about structurally complex actions. Particularly, it addresses how abstract actions can be refined into more specific composite actions in planning, offering proof systems and formal semantics that handle underspecification, hierarchical task networks, and soundness/completeness properties for abstractions enabling modular multi-level planning.
Key finding: This work extends propositional dynamic logic with a refinement relation allowing abstract actions to be refined into more specific composite actions, supports hierarchical planning via a formal proof system... Read more
Key finding: This paper develops a general framework for sound and complete abstractions between high-level and low-level agent action theories using situation calculus and ConGolog. It introduces refinement mappings and bisimulation... Read more
Key finding: This paper integrates defeasible argumentation with partial order planning to handle contradictions and dynamic information during plan construction. Actions' effects and preconditions interplay with defeasible rules, and new... Read more
3. How can reasoning about actions and change incorporate epistemic, intentional, and normative considerations to explain intentional action and enable ethical, accountable autonomous behavior?
This theme explores how reasoning about actions extends beyond causal and logical dynamics to include knowledge-how, ability-constituting knowledge, intentions, and ethical frameworks. It includes philosophical accounts reconciling causalist and epistemic views of intentional action, frameworks for ethical decision-making in autonomous agents, and the role of knowledge in practical reasoning geared towards epistemic normativity and practical agency.
Key finding: This paper proposes a reconciliation of causalist and Anscombian traditions in philosophy of action by appealing to ability-constituting knowledge (knowledge-how) as epistemic features distinguishing intentional action from... Read more
Key finding: Argues a functionalist epistemic normativity of reasoning where practical reasoning serves not only prudential goals but an epistemic function of generating knowledge of what one ought to do. This unifies the epistemic norms... Read more
Key finding: Develops a modular logic-based computational framework using event calculus and answer set programming to model and reason about ethical theories governing autonomous agents’ actions. The framework supports systematic... Read more
Key finding: Integrates the knowledge-first epistemology with intellectualism about knowledge-how, emphasizing the role of belief-based desire and intention in practical reasoning. The paper clarifies the cognitive architecture underlying... Read more