Cognitive Thermometers: Machine Learning and Logical Complexity

Title: Cognitive Thermometers: Machine Learning and Logical Complexity Source: arxiv Machine learning models can serve as agnostic proxies for measuring semantic complexity when traditional logical definability proves sensitive to language choice. Evidence shows learning and logic converge on semantic typology, but machine learning better explains divergences where symbolic complexity fails to account for statistical patterns.