Managing Uncertainty in Rule Based Cognitive Models

TL;DR

This paper investigates methods for modeling and propagating uncertainty in rule-based cognitive systems, emphasizing psychologically plausible techniques and mathematical methods for combining certainty factors.

Contribution

It introduces a unified approach for propagating uncertainty within rules and across rules, extending previous findings with a focus on psychological realism.

Findings

Maximum and minimum functions effectively summarize antecedent certainty.

Multiplication of certainty factors models conclusion certainty.

Heckerman's method improves certainty combination across rules.

Abstract

An experiment replicated and extended recent findings on psychologically realistic ways of modeling propagation of uncertainty in rule based reasoning. Within a single production rule, the antecedent evidence can be summarized by taking the maximum of disjunctively connected antecedents and the minimum of conjunctively connected antecedents. The maximum certainty factor attached to each of the rule's conclusions can be sealed down by multiplication with this summarized antecedent certainty. Heckerman's modified certainty factor technique can be used to combine certainties for common conclusions across production rules.