(fetzer)dr

Defeasible Reasoning: A Philosophical Analysis in Prolog
by Donald Nute

p 251
We reason defeasibly when we reach conclusions that we might be forced to
retract when faced with additional information. [...] It is the kind of
``other things being equal'' reasoning that proceeds from the assumption
that we are dealing with the usual or normal case.

The method of my analysis will be formal, but it will not be any of the
usual methods of the philosopher logician. Instead, I will take a different
approach and formulate my analysis of defeasible reasoning as a computer
program.

Philosophical Analysis and Logic
Philosophical Analysis and Logic Programming
Prolog

Defeasible Reasoning, p 261
This is the familiar ``good news-bad news'' story. ``I submitted a paper for
a meeting of the Hawaiian Philosophical Association. Unfortunately, they
didn't accept it. But they invited me to comment on another paper. My
department didn't have money for my travel expenses. Luckily the dean came
up with the money. By that time, all the flights to the meeting were booked.
But I don't really mind. They were meeting in Cleveland.''

Defeasible Warrants and Rules
Absolute Rules, Defeasible Rules and Presumptions
Absolute Derivability
Defeasible Derivability

Might Conditionals and Defeaters, p 276
[A might conditional is a denial of an ordinary conditional - a defeasible
rule. A defeater is special case where the rule might be defeated] a sick
bird might not fly.

Chaining
Summary of the Analysis
Expressive Power and Meaning of the Defeasible Language

Advantages of the Method and the Analysis, p 285
One advantage [is] to extend the community of inquirers in two [ways].
First, it provides a common ground for exchange [...]. [Many cognitive
sientists and AI researchers] often prefer models of a more concrete kind,
specifically models that actually work [...]. Second, [it] allows anyone
[to] propose test cases [...].

Systems built around probabilistic approaches [are] difficult to maintain
because we may have to adjust probabilities for existing rules in a system
whenever we add new rules.