Explainable artificial intelligence in cognitive learning psychology: A psychometric meta-analysis

Abstract

The purpose of this study is to examine explainable artificial intelligence (XAI) in cognitive psychology within higher education institutions. Data were retrieved from Scopus, Web of Science, and PsycINFO (1 January 2015–30 October 2025) in accordance with the PRISMA 2020 guidelines. Our initial search identified 3426 records. Of these, 1936 papers were excluded, and 616 studies underwent full-text review. In the final stage, 188 studies met the selection criteria and included 62,000 participants from higher education institutions worldwide. Consistency was demonstrated by a mean R 2 of 0.79 ( p < 0.001), as revealed by a science mapping analysis of three clusters—cognitive load and neural efficiency (C = 0.82, D = 0.76), attention and working memory (C = 0.64, D = 0.58), and metacognition and affective interaction (C = 0.42, D = 0.39)—which indicates positive pooled correlations ( r = 0.27–0.36). Model fit diagnostics were satisfactory ( CFI = 0.96 ; RMSEA = 0.042 ), with moderate heterogeneity observed ( I 2 ≈ 58 % ), and a global hypermean ( μ 0 ) centered on a positive association ( r ≈ 0.32 ). The psychometric meta-analytic model showed that the posterior distributions across XAI in the cognitive psychology domain were as follows: μ = 0.33 ≈ r = 0.32 , 95 % CrI 0.27,0.38 ; σ μ = 0.05 0.01,0.12 ; τ ∼ = 0.11 . The findings indicated that XAI demonstrated validity within cognitive psychology, structured across three principal domains: cognitive load and neural efficiency, attention and working memory, and metacognition and affective interaction, within adaptive learning systems in higher education.

Ähnliche Arbeiten