Higher-Order Responsibility

TL;DR

The paper tackles responsibility gaps in group and sequential decision making by introducing higher-order responsibility up to degree $d$, formalized within a sequential mechanism framework with deontic constraint $\gamma$. It defines $d$-order responsibility ${\sf R}^d_i$ and the corresponding $d$-gap-free class ${\sf GF}^d$, and proves two main results: (i) in any consecutive $n$-agent mechanism, the $n$-order gap is always empty, ensuring accountability at some $d\le n$; (ii) deciding emptiness of the $d$-order gap is $\Pi_{2d+1}$-complete, indicating high computational hardness. These findings position higher-order responsibility as a stringent, computation-aware alternative to group responsibility for ensuring accountability in multi-agent systems. The work leverages formal logic to connect deontic violations, counterfactual responsibility, and mechanism design, with implications for AI governance and ethical planning in hybrid human–machine settings.

Abstract

In ethics, individual responsibility is often defined through Frankfurt's principle of alternative possibilities. This definition is not adequate in a group decision-making setting because it often results in the lack of a responsible party or "responsibility gap''. One of the existing approaches to address this problem is to consider group responsibility. Another, recently proposed, approach is "higher-order'' responsibility. The paper considers the problem of deciding if higher-order responsibility up to degree $d$ is enough to close the responsibility gap. The main technical result is that this problem is $Π_{2d+1}$-complete.

Figures (1)

• Figure 1: Extensive form game between Devil and Moralist.

Theorems & Definitions (16)

• Definition 1
• Definition 2
• Definition 3
• Theorem 1
• Lemma 1
• Theorem 2
• Lemma 2
• Definition 4
• Lemma 3
• Lemma 4