Predicting and Analyzing High-Level Cognitive Traits Using Computational Multiplex Networks and Vector Representations

Abstract: High-level cognition, such as intelligence and creativity, are considered the hallmark of human cognition; however, their complexity hinders the identification of underlying common mechanisms. We focus on one such likely mechanism - mental navigation. We utilize converging computational methods to demonstrate how mental navigation - operationalized via verbal fluency tasks—predicts individual differences in creativity, intelligence, and openness to experience (the personality trait most closely related to them). Participants’ (N = 479) responses to two tasks- a 2-min animal fluency task and a 2-min generating synonyms of the word “hot” fluency task—were modeled over a multidimensional model (a cognitive multiplex network) of the mental lexicon. Quantitative measures of their mental navigation were used to build regression models that significantly predicted their assessed high-level cognition (replicating across both fluency tasks). In addition, we enriched the multiplex structure beyond the original four layers (phonological similarity, free association, hypernyms, synonyms) by testing new layers- Word2Vec, LWOW, image-based CLIP, and Lancaster sensorimotor norms - each capturing complementary lexical dimensions. All extensions improved predictive accuracy, with the Lancaster norms layer yielding the strongest gains, particularly for creativity-related outcomes. These findings show that short fluency tasks, modeled through enriched multiplex lexical networks, offer a robust and scalable framework for predicting high-level cognition. They also underscore the importance of integrating semantic, distributional, and sensorimotor knowledge in computational models of thought. Finally, we introduce an online tool the High-level Cognitive Prediction tool that applies this framework.