Ulrike Hahn
@UlrikeHahn@fediscience.org  ·  activity timestamp 3 weeks ago

4/ specifically, we take three interrelated notions of ‘argument’: as a single reason, as a unit comprising premises and a conclusion, and, finally, as a dialectical exchange. We argue that these form interrelated, but separable dimensions of a 3 dimensional argument space, which allows us to locate extant modelling frameworks relative to each other while also providing a tool for thinking about argument in different real world contexts.

https://link.springer.com/article/10.1007/s11245-025-10215-2
Ulrike Hahn
@UlrikeHahn@fediscience.org  ·  activity timestamp 3 weeks ago

5/ Argument “space” as defined by the representational complexity of individual reasons, the inferential complexity of premise-conclusion relations, and the dialectical complexity of the strategies by which agents select arguments

…we hope this work will lead to better modelling choices and greater conceptual clarity in argumentation research more generally. In particular, we hope it will facilitate exchange across “rival” frameworks for thinking about argument by allowing greater clarity on what a given framework does and does not capture

@cogsci @philosophy #epistemology#ABM#ABMs
Nick Byrd, Ph.D.
@ByrdNick@nerdculture.de  ·  activity timestamp 4 weeks ago

How are concepts like 'reasoning' and 'inference' defined?

#Philosophy has definitions that go way back (e.g., in #epistemology and #PhilMind). Now #computerScience are realizing a need for definitions.

This #openAccess#CS review takes a crack at them: https://doi.org/10.48550/arXiv.2504.15900

"...we formalize reasoning as a structured, multi-step process that dynamically decomposes complex problems, generates intermediate hypotheses, and iteratively refines solutions through logical and evidence-based transformations." "While inference I constitutes a single-step conditional probabilÎ𝑡=1 𝑇 ity computation 𝑃( 𝑦 |𝑥) = 𝑃( 𝑦 𝑡 |𝑥, 𝑦 <𝑡 ), reasoning R implements a meta-process coordinating multiple inference calls through explicit state management R(𝑥) = Φ 1 ◦ Φ 2 ◦ · · ·◦Φ 𝑛(𝑥). T"
"...we formalize reasoning as a structured, multi-step process that dynamically decomposes complex problems, generates intermediate hypotheses, and iteratively refines solutions through logical and evidence-based transformations." "While inference I constitutes a single-step conditional probabilÎ𝑡=1 𝑇 ity computation 𝑃( 𝑦 |𝑥) = 𝑃( 𝑦 𝑡 |𝑥, 𝑦 <𝑡 ), reasoning R implements a meta-process coordinating multiple inference calls through explicit state management R(𝑥) = Φ 1 ◦ Φ 2 ◦ · · ·◦Φ 𝑛(𝑥). T"
"...we formalize reasoning as a structured, multi-step process that dynamically decomposes complex problems, generates intermediate hypotheses, and iteratively refines solutions through logical and evidence-based transformations." "While inference I constitutes a single-step conditional probabilÎ𝑡=1 𝑇 ity computation 𝑃( 𝑦 |𝑥) = 𝑃( 𝑦 𝑡 |𝑥, 𝑦 <𝑡 ), reasoning R implements a meta-process coordinating multiple inference calls through explicit state management R(𝑥) = Φ 1 ◦ Φ 2 ◦ · · ·◦Φ 𝑛(𝑥). T"