Dependencies in evidential reports: The case for informational advantages

Area of Science:

• Cognitive Science
• Decision Making
• Information Theory

Background:

• Dependencies in expert forecasting, group communication, and evidence evaluation are often treated as redundancy.
• Existing frameworks conflate the structure of dependency networks with the observations within them.

Purpose of the Study:

• To demonstrate that dependencies can provide an informational advantage over independence in reasoning.
• To disentangle dependency network structure from network observations.

Main Methods:

• Mathematical proofs were employed to analyze the informational value of dependencies.
• Specific examples were used to illustrate theoretical findings.
• Comparison of hypothesis support under dependency versus independence with partial or contradicting observations.

Main Results:

• Dependencies can yield an informational advantage over independence.
• Partial or contradicting observations provide stronger hypothesis support in the presence of structural dependency.
• Lay reasoners understand assumptions for these advantageous structures but not their probabilistic implications.

Conclusions:

• Rethinking the role of dependencies in reasoning is crucial.
• Dependencies are not inherently redundant and can enhance information processing.
• Further research is needed on how to leverage these dependencies in belief revision and judgment.