When Alignment Is Not Enough: Cognitive Budget Framework for Game Design

Abstract

The Learning Mechanics–Game Mechanics (LM–GM) model helps serious game designers align game mechanics with intended learning functions. But pedagogical alignment alone is not enough, game design should also consider how the addition of mechanics costs learners before it can support learning. This paper develops the Cognitive Budget Framework from an unexpected result in a randomized controlled trial: an LM–GM-aligned sales-training game failed to outperform a content-matched conventional module. The framework extends LM–GM with three concepts: mechanics acquisition cost as the working-memory demand required to operate rules, tokens, cards, and interfaces before they teach anything; expertise offset as the relief provided by prior domain schema; and effective load, the learner-specific burden that remains after mechanics cost and expertise interact. It reframes pedagogical alignment as necessary but insufficient, especially when learners differ in prior readiness. The empirical basis is a randomized controlled trial with 63 completers, comparing a scaffolded, card-based digital sales-training game with a conventional digital module covering identical content. The study used a-priori-specified outcomes, validated self-report measures, ADAPTS-SV as a proximal adaptive-selling outcome, and in-game telemetry. Contrary to design expectations, the game did not outperform the control. It produced higher perceived cognitive load, d = 0.71, and higher attrition, 41.8% versus 16.2%. The game’s value was conditional on learner readiness: outcomes were least favorable for low-readiness learners and approached parity as baseline expertise increased, with a significant adaptive-selling performance interaction, F(1, 59) = 11.65, p = .001. Within the game condition, engagement related to performance through self-efficacy, indirect effect = 1.10, 95% CI [0.43, 2.24], while cognitive load did not enter this pathway. These findings suggest that an aligned mechanics layer can impose real acquisition cost, and that learners with stronger prior schema are better able to absorb it. The paper contributes a theory-building extension to LM–GM and offers practical guidance for game-based training design: use readiness checks, separate mechanics onboarding from content learning, and route learners adaptively based on prior expertise.