“A new approach to contrary-to-duty obligations” characterized

Damir Dzhafarov, Stefan Kaufmann, Bjørn Kjos-Hanssen, Dave Ripley, et al., at the 2016 ASL Annual Meeting at UConn.

Slides

José Carmo and Andrew J.I. Jones have studied contrary-to-duties obligations in a series of papers.

They develop a logical framework for scenarios such as the following:

1. There ought to be no dog.
2. If there is a dog, there ought to be a fence.

One conjecture from Carmo and Jones 1997 was refuted in a rather technical way in my 1996 term paper at University of Oslo.
The conjecture stated that one could simply add the condition
$\DeclareMathOperator{\pii}{ob}$
$$
(Z \in \pii(X)) \land
(Y \subseteq X) \land
(Y \cap Z \ne \emptyset ) \rightarrow (Z \in \pii(Y )) \tag{5e}
$$
for the conditional obligation operator ob.
In a follow-up paper (2001) they argued that (5e) could be added by weakening some other conditions.
In a new paper in Studia Logica, and presented at the Association for Symbolic Logic Annual Meeting 2016 at UConn, I argue that (5d) and (5e) are in conflict with each other. The argument is a generalization and strengthening of the 1996 argument.

2018: Benzmüller et al. have implemented Carmo and Jones’ logic in the proof assistant Isabelle and Jake Fennick’s MA project is the implementation of my follow-up paper in Isabelle.

During my sabbatical in 2025 I found a surprising characterization of Carmo and Jones’ system from their 1997 article A new approach to contrary-to-duty obligations in terms of a single forbidden world. The formal proof and accompanying paper is currently under revision with a journal.

A computable measure of irregularity

My book “Automatic complexity: a computable measure of irregularity” was published by de Gruyter on February 19, 2024. It concerns a version of Sipser’s complexity of distinguishing that was introduced by Shallit and Wang in 2001 and has been developed by me and others since 2013. The book’s exercises have formalized solutions in the proof assistant Lean.

Errata

  1. Exercise 1.12: the word “no” should be removed.
  2. Theorem 1.41 (3) refers the reader to Chapter 2 for the function $A_N$. In fact, $A_N$ is introduced in Definition 1.21.
  3. Claim 4.43: $b’\in\Sigma\setminus\{w_1,\dots,w_s\}$ should (obviously) be replaced by $b’\in\Sigma\setminus\{b_1,\dots,b_s\}$.
  4. Theorem 4.49: The text claims Our witnesses for $A^{\mathrm{perm}}(x)\le |x|+1$ are doubly transitive since the generated group is $S_n$. However, the generated group is not $S_n$ in general. A minimal counterexample is the witness for the word $00$, which generates the group $\mathbb Z/3\mathbb Z \ne S_3$. Therefore, whether $A^{\mathrm{tra2}} \le A^{\mathrm{perm}}$ should be considered open. Python calculations suggest it is true for binary words $w$ with $|w|\le 4$, and for 00000, at least.
  5. Theorem 6.6: Two constructions $M_1\times_1 M_2$ and $M_1\times_2 M_2$ are described. The text suggests that $\times_1$ is used for the theorem, but actually $\times_2$ should be used. The operation $\times_1$ can be used to prove the inequality $A_N(x)\le A_N(x\mid y)\cdot A_N(y)$ instead.

Math 100 student’s protein folding research

Madison Koskey had the highest score in the class on this project task:

Fold an idealized protein known as S64 to obtain a highest possible score in the hydrophobic-polar protein folding model in three dimensions.

This problem was studied by Lesh, Mitzenmacher, and Whitesides (2003).

Above we see Madison’s intricate folding of S64 which earned a score of 27.

To try your hand at rotating it in 3D, go directly to this link.