A brain-computer interface (BCI) infers our actions (e.g. a movement), intentions (e.g. preparation for a movement) and psychological states (e.g. emotion, attention) by interpreting our brain signals. It uses the inferences it makes to manipulate a computer. Although BCIs have long been used exclusively to support disabled people (e.g. through brain-controlled wheelchairs, spellers), with the emerging low-cost and portable hardware, they have started to be considered for a variety of human-computer interaction (HCI) applications for non-disabled people as well. Among these, games have been receiving the interest of researchers and practitioners from both the BCI and HCI communities. In BCI research, games have long been used solely to demonstrate the performance of signal processing and analysis methods. Therefore, they have been evaluated only for their performance (e.g. recognition accuracy, information transfer rate). However, games are not meant to satisfy our practical needs. They satisfy our hedonic needs. They challenge us, let us make our fantasies true, evoke our memories, and so on. We look for these experiences while playing games. Thus, rather than the performance of the controller used, the user experience (UX) of the game is essential. UX of a game is a consequence of the player’s internal state, the game characteristics and the context. Evaluating such a complex phenomenon is non-trivial. Often, UX is measured in terms of other, measurable concepts such as flow, immersion, presence, social behaviour and so on. Methods to evaluate UX also vary and include questionary, inter- viewing, and observation analysis. Evaluating UX of BCI games is even harder because UX evaluations may be biased due to the low recognition performance of the BCI. But this should not keep us from investigating UX of BCI games and identifying the good and bad practices, indepen- dent of performance. Because, ignoring UX while trying to improve performance might lead to games that are perfectly functional, but not enjoyed or played by anyone. In this work, we investigated how the BCI control can influence the UX of a computer game. We considered a futuristic scenario in which BCI functioned as perfectly as other modalities. To simulate this scenario, we proposed and followed an approach called equalised comparative evaluation (ECE). For this, we equalised the perceived performance of BCI and several other modalities. We did not simply introduce artificial errors on the modalities as this could reduce player effectance and, thus, enjoyment. Instead, we manipulated the challenge of the tasks the players performed. Then, we evaluated and compared the UX while playing with BCI and with the other modalities. Our work consisted of three studies in each of which we evaluated different UX related concepts and used different data collection methods. In all the studies, participants played an experimental multimodal game that we had developed, called Mind the Sheep! (MTS!). They controlled 3 dogs using different modalities in order to herd 10 sheep across a meadow. The goal was to pen all the sheep as quickly as possible. In Study 1, we showed the effectiveness of our ECE approach. Pairs of participants played a collaborative, multi-player version of MTS! once using a BCI that relied on the steady-state visually evoked potential (SSVEP), once by simple mouse pointing and clicking (non-ECE ap- proach) and once using a visuomotor control mechanism that was as challenging to control as BCI (ECE approach). We relied on observation analysis, interviewing and questionnaires to evaluate UX in terms of social interaction. We found that challenging control dampened collabo- rative social interaction but it improved emotional social interaction. In Study 2, participants played single-player MTS! once using BCI and once using the visuomotor control mechanism we used in Study 1. They indicated their UX in terms of affect and immersion using questionnaires. We found that the BCI selection method was more immersive and that the participants were more indulgent towards BCI control. One question that arose from our findings was whether the positive UX of BCI control was due to a novelty effect. This was what we investigated in Study 3. In Study 3, we compared UX of BCI control to that of automatic speech recogniser (ASR) control, under the assumption that both ASR and BCI were novel game input modalities. Participants played single- player MTS! once with BCI, once with ASR and once with the option of switching between the two. Using questionnaires, they rated their expectations, engagement and workload levels as well as perceived game/controller quality. We also conducted interviews and analysed game logs. The participants rated BCI control higher in hedonic quality but lower in pragmatic quality than ASR control. The challenge and novelty of BCI influenced their modality switching behaviour. The contributions of our work are manifold. The ECE approach we proposed allows evaluating UX of BCI games (or applications in general) independent of their performance and investigating the unique capabilities of BCI. The UX evaluation results demonstrate the ways the challenge, cognitive involvement and novelty offered by BCI can influence the UX of a game. Our discussion on the preferable (and non-preferable) measures and methods for evaluating the UX of BCI games provides guidelines to other researchers. Furthermore, the BCI and physiological computing (PC) frameworks we proposed allows developers to situate their applications among other BCI or PC systems.
|Award date||21 Sep 2012|
|Place of Publication||Enschede|
|Publication status||Published - 21 Sep 2012|
- HMI-CI: Computational Intelligence
- HMI-MI: MULTIMODAL INTERACTIONS
- HMI-HF: Human Factors