PEACH: um modelo de avaliação de interfaces de jogos educacionais baseado em princípios cognitivos

Abstract

Educational games have become relevant tools for supporting the learning of subject-matter content by integrating cognitive, affective, behavioral, and sociocultural elements. However, their evaluation still faces challenges, particularly in identifying aspects that may generate cognitive overload during interaction. In this context, this dissertation proposes PEACH (Prototype Evaluation Analysis for Cognitive Heuristics), a model designed to evaluate the interfaces of educational games during their prototyping phases. The model comprises three dimensions: educational game design, multimedia learning, and interface design, and is grounded in Cognitive Load Theory (CLT), the Cognitive Theory of Multimedia Learning (CTML), and interaction design guidelines. The research adopted the Design Science Research (DSR) methodology, which was structured in four stages: (i) a systematic literature mapping highlighted the predominance of evaluations conducted at the end of development, mainly through self-report questionnaires such as NASA-TLX. It also identified four types of measures reported in the literature: subjective measures, focused on the perception of mental effort; behavioral measures, which record interactions such as response time and error frequency; performance measures, which verify accuracy, progression, and dropout rate; and physiological measures, which analyze signals such as pupil dilation, heart rate, and brain activity. These findings show that, although there is a diversity of methods, there is little integration among them and little direct relationship with game design and interface elements; (ii) based on these findings, an instrument grounded in CTML was developed and validated to examine the visual and textual organization of interfaces; (iii) the analysis of Educational Game Design Documents (EGDDs) from two games and validation with experts consolidated the dimensions of game design and interaction design, with item refinements, scale standardization, and the inclusion of a field for visual evidence. Part of these results has already been published in scientific papers that examine how design decisions can affect players’ understanding and experience. In addition to the construction and conceptual validation, (iv) the model was applied to three educational games. The evaluations revealed recurring issues, including the use of verbal–visual redundancies without consideration of accessibility goals or cognitive reinforcement, inadequate information organization, inconsistencies in progression, unclear tutorials, and interaction points that require greater cognitive effort. The questionnaire responses were analyzed quantitatively, with the calculation of scores by item, measures by dimension, and the generation of per-game dashboards, which facilitated the visualization of performance in each PEACH dimension and the localization of these issues in the interfaces. Semi-structured interviews with developers contributed to refining the model’s criteria and to understanding how it is used in the prototyping process of educational games. In summary, the results show that PEACH can support design revisions still in the prototyping phase and, in future work, be combined with approaches to learning analytics in digital educational games.