Emotion-Based Interactive Storytelling with Artificial Intelligence

Abstract

Artificial Intelligence (AI) techniques have been widely used in video games to control non-playable characters. More recently, AI has been applied to automated story generation and game-mastering: managing the player’s experience in an interactive narrative on-the-fly. Such methods allow the narrative to be generated dynamically, in response to the player’s in-game actions. As a result, it is more difficult for the human game designers to ensure that each possible narrative trajectory will elicit desired emotional response from the player. We tackle this problem by computationally predicting the player’s emotional response to a narrative segment. We use the predictions within an AI experience manager to shape the narrative dynamically during the game to keep the player on an author-supplied target emotional curve.

Keeping the Player on an Emotional Trajectory in Interactive Storytelling

“…In this paper we implement and evaluate PACE, an AI experience manager (Poo Hernandez, Bulitko, and St.Hilaire 2014). To make the paper self-contained, we briefly present PACE algorithm before detailing our implementation of it in a video game and the resulting evaluation.…”

“…Which plan would elicit the desirable emotional response from the player? Poo Hernandez, Bulitko, and St.Hilaire (2014) answer this question by computationally predicting the player's emotional reactions to automatically planned narratives and then choosing the narrative predicted to elicit the reaction closest to that specified by the author. The advantage of this approach is that the author of such an interactive narrative does not even need to specify concrete narrative events/goals (as was required by previous managers (Riedl et al 2008;Ramirez Sanabria and Bulitko 2014)).…”

Designing Story-Centric Games for Player Emotion: A Theoretical Perspective

“…Interactive narratives present opportunities for players to become active participants in rich, engaging story experiences. Furthermore, intelligent narrative technologies enable interactive story designers to create experiences that dynamically adapt and respond to individuals (McCoy et al 2014;Rowe, Mott, and Lester 2014;Yu and Riedl 2015;Poo Hernandez, Bulitko and Hilaire 2014). Recent work on intelligent narrative technologies has spanned a broad range of topics, including data-driven techniques for drama management (Rowe, Mott, & Lester 2014;Yu and Riedl 2015), planning techniques for plot-centric story generation (Ware and Young 2014;Porteous et al 2015), and playable social simulation models (McCoy et al 2014).…”

“…The OCC model of emotion is a commonly used theory for modeling agent emotions in intelligent narrative technologies, including recent work on moral emotions for virtual characters (Battagliano, Damiano, and Dias 2014) and character emotion in story generation (Sarlej & Ryan 2014). Recent work has also begun to investigate computational models of emotion to inform drama management decisions (Hernandez, Bulitko, and Hilaire 2014). In this paper, we contribute to this literature by presenting preliminary work examining player affect in interactive narratives from the perspective of psychological theories of emotion.…”

Standardizing the Evaluation of Digital Managers for Better Interactive Experiences

“…The outcome is that research in the field is widely fragmented and mostly independent, making it tough to draw conclusions from research that generalize well across the field (Koenitz et al 2011;Szilas, Boggini, and Petta 2011;Szilas et al 2012). Some prior work has offered structural comparisons of experience managers (e.g., in terms of their design and expected capacities) (Roberts and Isbell 2008;Thue 2015) and different work has compared one manager's performance to that of a homemade baseline (Thue et al 2007;Ramirez and Bulitko 2014;Hernandez, Bulitko, and Hilaire 2014). We are unaware of any previous efforts to directly compare the performance of two or more independently-created managers.…”