1. Introduction                                                                                                5. Discussion

                2. A Simple Decision-Making Task                                                                6. Conclusion

                3. Method                                                                                                      Biographies

                4. Results                                                                                                      References

                1. Introduction                                                                                                5. Discussion

                2. A Simple Decision-Making Task                                                                6. Conclusion

                3. Method                                                                                                      Biographies

                4. Results                                                                                                      References

5.   Discussion


 

The key design goals of our multimodal display were validated in the experiment. First, the decision-making task had a general success rate of about 41%, within the desired range of 40-60%, necessary to allow a sufficient number of both correct and incorrect trials (Williams, Nesbitt, Eidels and Elliott 2011). However, there was considerable variation between players, with four of the 48 players averaging below a 25% success rate and another four achieving higher than 70% of correct responses when all their trials were considered. Figure 7 shows the distribution of win-percentage across players (the line marks the mean correct rate of 41%).

 

Figure 7: Percentage of wins for players.

Second, the application of increasing sound cue to prevent timeouts was also successful. When we analyzed all trials together and then player-by-player, we found a significant reduction in timeouts in the increasing sound condition compared to the no sound condition. This reduction in timeouts did not seem to come at the expense of players responding more quickly in the task. Indeed, when we examined all trials together, we found a significant increase in reaction time, so players actually slowed their response time during the increasing sound condition, as compared to the no sound condition. Overall, they also recorded more wins and fewer losses in the increasing sound condition than the no sound condition. This was consistent with the game design, as we expect the task to become easier to answer successfully as players wait longer.

 

These results are mitigated by the fact that when we compared the data by averaging player by player outcomes, the difference in timeouts was still significant. The increase in response time and wins in the increasing sound condition, however, was no longer evident. This suggests that there was some bias introduced into the overall results by the performance of individuals in the experiment. Regardless, we can be confident that the players’ speed and accuracy in the multimodal task did not reduce to offset the reduction in timeouts. 

 

The most surprising result was that when we compared the overall trials, we also found a significant difference between the response time and number of wins in the constant sound condition, as compared to the no sound condition. Again, this is mitigated by the fact that this significance was not evident when we compared the average results over the 48 players. Regardless, this is a surprising result and worth further discussion.

 

 The increasing sound signal was specifically designed to facilitate players in waiting longer before deciding without risking a timeout. The constant sound was introduced as a control, yet, somewhat surprisingly, players seemed to wait longer before responding in this condition as well. This could imply they also receive some timing information from this constant sound signal. One explanation for these results is that players have an internal mechanism for measuring time that is activated by a constant sound signal. Indeed, such a model has been proposed, describing an internal pacemaker sending a regular series of pulses to some kind of counter mechanism (Zakay and Block 1995).

 

In this model, the internal clock mechanism can also be calibrated by external events. In the Buckets game this calibration might occur in accordance with the visual updates or the game timeouts. The model has also been used to consider how such a clock could have an impact on cognitive functions, such as the decision-making task (Block 2003). Because of the need to track time in the constant condition, this model predicts that we would see slightly longer response times, yet lower success rates. While this is the case in this experiment, these differences were not found to be significantly different in the constant sound condition, as compared to the increasing sound condition. Regardless, this interesting result is probably worth further study.

 

In terms of decision-making tasks in game designs, we have demonstrated the usefulness of gathering empirical data to test player performance on game-like tasks. We have also demonstrated how simple informative sound displays can provide useful information in a perceptual visual challenge.