Blindness

Question

How does "argumentative completeness," as described, differ from simpler forms of bias like confirmation bias or selective exposure? Why is this distinction crucial for understanding its impact on learning within both standard and hierarchical Bayesian frameworks?

Synthesized answer

"Argumentative completeness" describes a body of testimony where the authors can respond to and argue against any doubts or data that might threaten the credibility of their narrative [Passage 1, Passage 3]. This differs from simpler biases as it implies an active defense against contradictory information, rather than just a passive selection or interpretation of evidence. Agents holding onto an argumentatively complete theory will eventually find a reply to any attack or doubt they might encounter [Passage 2].

This distinction is crucial because argumentative completeness can preclude learning even in hierarchical Bayesian frameworks. Such testimony can undermine higher-order constraints and epistemic practices designed to promote good learning [Passage 3, Passage 4]. Argumentative completeness collapses the case of higher-order Bayesian frameworks to a first-order setting, meaning a learner may hold onto an argumentatively complete theory regardless of its inadequacy by other standards, as it will always find a reply to objections [Passage 2]. The passages do not explicitly detail how this differs from confirmation bias or selective exposure beyond the active argumentation…

Synthesized from the book passages below. Chat with the book on Feynman for follow-up.

From the book

ion to few sources of testimony and a natural co-dependence between beliefs and interpretation (Asher and Paul, 2018 ) . Relying on testimony from a restricted set of sources to update one’s beliefs leads to the mutual reinforcement of our confidence in the source and our belief in ; this creates a bias that can preclude learning when an agent tries to exploit new data that are incompatible with or simply distinct from . Agents who are interpretively blind will discount any evidence that challenges their beliefs. We use Wolpert’s 2018 extended Bayesian framework to prove our results. While IB…
Passage [4]
ypotheses are rational and are updated on . If such confirms a hypothesis that does not, then is incapable of learning . Claim 2 of Proposition 6 shows that . Then apply Proposition 4 . Argumentatively complete testimony thus collapses the case of higher order Bayesian frameworks to our first order setting. What is troubling about IB is that our learner may hold onto an argumentatively complete regardless of how inadequate it is in the eyes of others or standard epistemic criteria; an argumentatively complete theory will always eventually find a reply to any attack or any doubt might acquire.…
Passage [26]
can respond to and argue with any doubts raised by other data or arguments in a body that might threaten ’s credibility. A skillful climate denier, for example, will always find a way to undercut the most scientifically careful argument. Argumentatively complete testimony thus can undermine higher order constraints and good epistemic practices that should guide first order learning. Our paper starts in Section LABEL:sec:testimony by discussing testimony. We then introduce the codependence of belief and interpretation and apply it to the situation of testimony and the sources that support it.…
Passage [5]
Interpretive Blindness Nicholas Asher 1 1 1 CNRS, IRIT and Julie Hunter 2 2 2 Linagora GSO Abstract We model here an epistemic bias we call interpretive blindness (IB). IB is a special problem for learning from testimony, in which one acquires information only from text or conversation. We show that IB follows from a co-dependence between background beliefs and interpretation in a Bayesian setting and the nature of contemporary testimony. We argue that a particular characteristic contemporary testimony, argumentative completeness , can preclude learning in hierarchical Bayesian settings, even…
Passage [2]
bilities updated on . In addition suppose there is a . We first show (1). Since is argumentatively complete, such that . We need to show that for some such , relative to . Suppose that , for all such that . By rationality, for each such , . Thus, all the non-0 probability mass of falls on undercutting sequences . But for each such undercutting of length , since is argumentatively complete, there is an evaluation hypothesis supported by such that . Since has only finitely many levels, at some level all T undercutting sequences get probability. This, together with the fact that , contradicts…
Passage [25]

More questions about this book