Multi-Dimensional Assessment for AI Cognition (MAAC): A Theoretical Framework for Process-Oriented Cognitive Evaluation of Text-Based AI Systems

This paper introduces the Multi-Dimensional Assessment for AI Cognition (MAAC), a theoretically grounded framework for evaluating AI systems through the lens of cognitive processes rather than task outcomes. MAAC defines nine cognitively motivated dimensions — Cognitive Load, Tool Execution, Content Quality, Memory Integration, Complexity Handling, Hallucination Control, Knowledge Transfer, Processing Efficiency, and Process-Outcome Alignment — each anchored in established cognitive science theory at Marr's (1982) algorithmic level, drawing on Baddeley's working memory model, Sweller's cognitive load theory, Barnett and Ceci's transfer taxonomy, and Newell and Simon's problem-solving architecture. Five theoretical analyses provide initial support for the framework's coherence and empirical testability: dimension-to-theory mapping against 108 retained sources (inter-rater κ = .78); a coverage matrix demonstrating exhaustiveness and non-redundancy across nine cognitive construct categories; a mechanistic gap-closure analysis addressing five critical limitations in current AI evaluation practice; a worked diagnostic illustration showing that identical benchmark accuracy (84% MMLU) can coexist with fundamentally different cognitive architectures; and seven a priori interdimensional correlation predictions for empirical testing in Study 3. Building on the complexity-validated scenario generation pipeline established in Study 1 (Doleh et al., 2026; https://doi.org/10.5281/zenodo.20114561), MAAC provides the theoretical and operational infrastructure for Study 3 (single-model empirical baseline) and Study 4 (multi-model comparative analysis). This work is currently under review for publication.