Experiments in serious game design : a cognitive approach

Abstract

Serious games sometimes are not as efficacious as they should be when it comes to learning, but the reasons for this are unclear; something that is brought on by a lack of rigorous scientific research into what constitutes good game design. We created a serious game with which players can learn the triage procedure, called Code Red Triage, and found that, at least in the short run, it was less efficacious than a static PowerPoint presentation with the same information. Players also reported higher cognitive load in the game condition, which was negatively correlated with learning gains, and likely due to the player having to actively select and organize the information. Mayer’s cognitive theory of multimedia learning states that learning from multimedia involves the selection of relevant from irrelevant information, the organization of this information into coherent knowledge structures, and the integration of these structures with prior knowledge structures. In accordance with this theory, we proposed four experiments that were designed to improve these cognitive processes while playing a serious game. In the first experiment, we contrasted the use of auditory and visual cues with a control group to aid in the selection of relevant from irrelevant material. Players in the auditory cueing condition showed significantly lower learning gains from gameplay than the other two groups. Participants in the visual cueing condition learned more than the auditory cueing condition, but not the control condition. However, a significant effect of prior game experience was found in the visual cueing condition, in favor of those with prior game experience. In the second experiment we investigated the best order in which to present the complexity of the game, on two variables: the presentation of problems and the presentation of options with which to overcome these problems. Problem complexity was operationalized as either spaced or massed presentation of victim cases, and option complexity as either just-in-time presentation or just-in-case presentation. No effect of any of the conditions was found on learning gains, and therefore effectiveness of organization, but the just-in-case option presentation with a massed victim presentation was enjoyed significantly more than other conditions, likely due to the player feeling more autonomous and competent. In the third experiment, we tested whether a serious game can be made more efficient if the game adapts the presentation of victim cases to the performance of the player. In this case, remaining victims belonging to a certain complexity level were deleted if the player scored high enough on a respective victim. This was shown to significantly decrease the amount of time for an equal amount of learning to occur. In the final experiment, surprising events were added around the time an update of the mental model was needed, in order to stimulate the activation of, and integration with a prior mental model. This was shown to lead to superior knowledge structures in the experimental condition compared with the control group.