Decoding Coherence with Evolutionary Computing

Abstract

This research explores the role of coherence in decision-making, using evolutionary computing techniques to understand why coherence considerations emerge. Coherence refers to how well information fits together and serves as a heuristic indicating truth. This study aims to unravel the cognitive processes behind coherence-seeking tendencies and their adaptive advantages in human cognition.

To address this, three hypotheses are formulated: Hypothesis 1 posits that coherence is a useful heuristic in most contexts. Hypothesis 2 suggests its utility in specific contexts requiring quick decisions or dealing with high uncertainty. Hypothesis 3 challenges the value of coherence, proposing alternative heuristics.

The methodology involves implementing a Genetic Algorithm (GA) with 100 agents divided into groups representing different decision-making strategies: Bayesians, Confirmationists, and two variations of Coherentists—one viewing coherence as relevance and the other as overlap. Each agent's decision-making process is encoded as a chromosome, and their performance is evaluated using a fitness function designed to measure how well they achieve the optimization objectives. The GA simulates decision-making scenarios to identify evolutionary mechanisms underpinning the importance of coherence in specific contexts.

Drawing from key studies, [1]Harris and Hahn's work on Bayesian rationality emphasizes coherence in aggregating information from multiple sources. [2],[3] Douven's research on group learning optimization supports using evolutionary computing to find optimal decision-making procedures, while his exploration of the ecological rationality of explanatory reasoning highlights the potential advantages of explanation-based update rules over traditional Bayesian principles.

A practical example of coherence in the wild can be seen in medical diagnostics, where a diagnosis is considered coherent if the symptoms fit well together, guiding doctors toward the correct decision even under uncertainty.

In synthesizing these perspectives, this research aims to deepen our understanding of coherence in decision making, elucidating when and why it is a valuable heuristic. By leveraging evolutionary computing techniques, this study seeks to uncover the underlying evolutionary mechanisms shaping decision-making strategies, offering insights applicable to psychology, economics, and artificial intelligence. Ultimately, this modeling project endeavors to decode the interplay between coherence and decision-making, informing the development of more robust frameworks.

References

[1] A. J. L. Harris and U. Hahn, "Bayesian rationality in evaluating multiple testimonies: Incorporating the role of coherence," Journal of Experimental Psychology: Learning, Memory, and Cognition, vol. 35, no. 5, pp. 1366–1373, 2009. DOI: [10.1037/a0016567](https://doi.org/10.1037/a0016567).

[2] I. Douven, "Optimizing group learning: An evolutionary computing approach," Artificial Intelligence, vol. 275, pp. 235–251, 2019. DOI: [10.1016/j.artint.2019.06.007](https://doi.org/10.1016/j.artint.2019.06.007).

[3] I. Douven, "The ecological rationality of explanatory reasoning," Studies in History and Philosophy of Science, vol. 79, pp. 1–14, 2020. DOI: [10.1016/j.shpsa.2020.04.001](https://doi.org/10.1016/j.shpsa.2020.04.001).