Probabilistic assessment of user's emotions in educational games

Abstract

We present a probabilistic model to monitor a user's emotions and engagement during the interaction with educational games. We illustrate how our probabilistic model assesses affect by integrating evidence on both possible causes of the user's emotional arousal (i.e., the state of the interaction) and its effects (i.e., bodily expressions that are known to be influenced by emotional reactions). The probabilistic model relies on a Dynamic Decision Network to leverage any indirect evidence on the user's emotional state, in order to estimate this state and any other related variable in the model. This is crucial in a modeling task in which the available evidence usually varies with the user and with each particular interaction. The probabilistic model we present is to be used by decision theoretic pedagogical agents to generate interventions aimed at achieving the best tradeoff between a user's learning and engagement during the interaction with educational games.2

Modeling Students’ Emotions from Cognitive Appraisal in Educational Games

“…In Figure 2, the links between nodes in different time slices indicate that the values of the corresponding variables evolve over time and that their value at time t i influences 1 We currently do not explicitly represent player preferences in our model. 2 Due to space constraints, we omit the description of how the model is used to inform the agent's choice. the value at time t i+1 .…”

“…Furthermore, DDNs provide algorithms that can compute, at any decision point, the agent action with the maximum expected utility, thus providing a decision theoretic basis for the agent behavior. In this paper, we focus on the part of the model that performs predictive assessment (more details on the diagnostic part can be found in [2]). Figure 2 shows two time slices of the DDN that forms our model of student's affect.…”

Modeling User Affect from Causes and Effects

“…Taking student affect into account could be especially beneficial for systems that, like educational (edu-) games, rely heavily on student emotional engagement to be effective. The long-term goal of our research is to devise emotionally intelligent agents for edu-games that model both student affect and learning, and generate adaptive interventions aimed at balancing the two [2]. In this paper, we focus on the model of student affect that we built for one such agent included in an edu-game on number factorization.…”

“…We conclude by discussing future work. Figure 1 shows a high-level representation of two time-slices in our DDN-based framework for modeling user affect [2]. Each time slice represents the system belief over relevant elements of the world after an interaction event of interest, such as a user's action (left slice) or an action from an interface agent (right slice).…”

“…The model is based on a framework that uses Dynamic Decision Network (DDN) to leverage information on both the possible causes and the observable effects of the user's affective reaction [2]. In previous work, we built the model's part that reasons from causes to emotions (predictive model) and found that it can achieve reasonable accuracy [3,20].…”