In this paper, we introduce the reader to the field of cognitive network science, that is, the application of network science methods to study human cognition and knowledge structures. Cognitive networks are representations of associative knowledge between concepts in a cognitive system apt at acquiring, storing, processing and producing language, that is, the mental lexicon. In a cognitive network, nodes represent concepts with links expressing relations, such as semantic, syntactic, phonological and visual connections, for example, “canine” and “dog” (nodes) linked by “being synonyms” (link). Hence, cognitive networks represent associative knowledge in mathematical, measurable and quantifiable ways. Can such structure be used to gain insights over cognitive phenomena? We explore this research question by reviewing recent, pioneering key applications and limitations of cognitive networks across visual, auditory, and semantic language processing tasks, either in healthy or clinical populations. We also review applications of cognitive networks modeling language acquisition, reconstructing text content and assessing creativity or personality traits in individuals. Our paper also gently introduces the reader to mathematical notations, definitions and measures about single‐layer and multiplex networks as well as hypergraphs. Last but not least, across phonological, semantic and syntactic networks, we guide the reader through relevant psychological frameworks, datasets and software packages that might all aid current and future cognitive network scientists. This article is categorized under: Psychology > Memory Psychology > Theory and Methods Linguistics > Cognitive

Creativity is a fundamental skill of human cognition. We use textual forma mentis networks (TFMN) to extract network (semantic/syntactic associations) and emotional features from approximately one thousand human-, GPT3.5-, and Sonnet 3.7-generated stories. Using Explainable Artificial Intelligence (XAI) we test whether features relative to Mednick’s associative theory of creativity can explain creativity ratings assigned by humans or AI raters. Using XGBoost, we examine 5 scenarios: (i) human rating human stories, (ii) GPT-3.5 rating human stories, (iii) GPT-3.5 rating GPT-3.5 stories, (iv) Sonnet 3.7 rating human stories, and (v) Sonnet 3.7 rating Sonnet 3.7 stories. Our findings reveal that GPT-3.5 and Sonnet 3.7 ratings differ significantly from human ratings not only in terms of correlations but also because of feature patterns identified with XAI methods. GPT-3.5 and Sonnet 3.7 favour “their own” stories and rate human stories differently from humans. Feature importance analysis with SHAP scores shows that: (i) network features are more predictive for human creativity ratings but also for ratings by GPT-3.5 and Sonnet 3.7 for human stories; (ii) emotional features played a greater role than semantic/syntactic network structure in GPT-3.5 and Sonnet 3.7 rating their own stories. These quantitative results underscore key limitations in the ability of GPT-3.5 and Sonnet 3.7 to align with human assessments of creativity. We emphasise the need for caution when using AI models to assess and generate creative content, as they may not yet capture the nuanced complexity that characterises human creativity.

Fostering creativity is vital for tackling 21st-century challenges, and education plays a key role in nurturing this skill. According to the associative theory, creativity involves connecting distant concepts in semantic memory. Here, we explore how semantic memory changes following an educational intervention intended to promote creativity. Specifically, we examine how a scientific education curriculum—Scientific Creativity in Practice (SCIP) program—impacts the semantic memory networks of 10–18-year-old students in a chemistry class (n = 176). Students in an Intervention group who received the SCIP intervention, and a Control group who did not, completed creative thinking tests, as well as verbal fluency tasks to estimate semantic networks in science-specific (chemistry) and domain-general (animal) categories. Results showed that the SCIP intervention significantly improved fluency scores on a general creative thinking test. Using network science methods, we observed increased interconnectedness in both science-specific and domain-general categories, with lower path distances between concepts and reduced modularity. These traits define a ‘small-world’ network, balancing connections between closely related and remote concepts. Notably, the chemistry semantic network showed substantially more reorganization, consistent with the chemistry contents of the SCIP intervention. The findings suggest that semantic memory reorganization may be a cognitive mechanism underlying successful creativity interventions in science education.

In this tutorial paper, we discuss cognitive networks as powerful models for understanding human cognition and knowledge. Cognitive networks are representations of associative knowledge between concepts in a cognitive system apt at acquiring, storing, processing and producing language, i.e. the mental lexicon. In a cognitive network, nodes represent concepts with links expressing relations, such as semantic, syntactic, phonological and visual connections, e.g. “canine” and “dog” (nodes) linked by “being synonyms” (link). Hence, cognitive networks represent associative knowledge in mathematical, measurable and quantifiable ways. Can such structure be used to gain insights over cognitive phenomena? We explore this research question by reviewing recent, pioneering key applications and limitations of cognitive networks across visual, auditory, and semantic language processing tasks, either in healthy or clinical populations. We also review applications of cognitive networks modelling language acquisition, reconstructing text content and assessing creativity or personality traits in individuals. Our tutorial also gently introduces the reader to mathematical notations, definitions and measures about single-layer and multiplex networks as well as hypergraphs. Last but not least, across phonological, semantic and syntactic networks, we guide the reader through relevant psychological frameworks, datasets and software packages that might all aid current and future cognitive network scientists.