Cues and Insinuations: Indicating Affordances of Non-Player Characters Using Visual Indicators

This study examines how the visual appearance of non-player characters (NPCs) in video games shapes players' interpretations of their functional roles, and what criteria players use to reach those judgments. Grounded in narrative theory, including Jenkins' embedded and emergent narratives, Calleja's scripted narrative and alterbiography, and Gibson's concept of affordances, the work argues that visual cues function as indicators that set player expectations about NPC behavior. Believable NPCs depend on those expectations being met, so the question of how players read appearance has direct design relevance.

The research relied on an online survey of 213 valid responses drawn from a self-selected sample recruited through social media, with respondents ranging in age from 18 to 52 and drawn largely from Sweden and the United States. Participants viewed in-game screenshots of 27 NPCs from four 2014 titles, namely Assassin's Creed: Unity, inFamous: Second Son, WATCH_DOGS, and World of Warcraft: Warlords of Draenor, and classified each according to Bartle's (2004) functional roles while explaining their reasoning. Frequency analysis captured classification patterns, and thematic analysis identified the factors behind them. Respondents relied on three broad factors: the NPC's surrounding area and location, the actions the NPC performed, and its attributes and visual presentation. Shopkeepers were recognized through counters and wares, guards through weapons, armor, and wariness, and quest givers through distinctive, face-forward presentation. Generic, masked, or clustered figures were read as filler or as enemies meant to be killed for loot. Respondents often drew on user interface elements left in the screenshots, which the authors acknowledge as an uncontrolled limitation. The authors note that Bartle's categories were difficult to separate, and that the framework may need to be revised for genres beyond MMORPGs. Demographic comparisons between Swedish and American respondents, and between male and female respondents, showed only weak differences, and the sample's heavy skew toward male and Western respondents limits generalization.