Since haptic is a relatively new research area, there have been not many studies with real subjects, but the number is growing. Gillespie and others (1998) tested their “virtual teacher” with 24 participants, and found that most participants profited from it. Klatzky and others (1993), through their experiments on haptic exploration, found that touch occurred frequently, and was initiated faster for objects’ material judgment tasks, while touch is much less used for objects’ shape judgment tasks. Langrana and others (1997) tested the Rutgers Master II with 32 subjects to find that the experimental group with force feedback training had slightly better performance than the control group. Colwell and others (1998) tested Impulse Engine 300 with 22 subjects to find that a haptic interface has considerable potential for blind computer users. Rosenberg (1994) experimented with 20 subjects to find that a delay as high as 100 ms can exist between the presentation of haptic and audio sensations in display of a virtual rigid surface (wall) before users notice any perceptual distortion. Jeong (2001) tested ordering tasks in auditory and haptic displays with 23 subjects and found that subjects performed better with haptic only displays than with auditory only displays or with auditory/haptic combined displays. Jeong and Gluck (2003) applied haptic display to geographic information system and tested with 51 subjects to confirm the possibility of using haptic display complementing visual display.

Emery and others (2003) tested multimodal interface including haptic with 29 older adults to find the all participants performed well under auditory- haptic bimodal feedback. Jacko and others (2003) tested multimodal interface with 29 normal vision older adults and 30 visually impaired older adults to find that for some cases, non-visual (auditory or haptic) feedback forms demonstrated significant performance gains over the visual feedback form. Jeong and others (2004) proposed the interactive system which combines an immersive virtual environment with human-scale haptic interface.

Adding force feedback effect is a way of multimodality. Multimodality is beneficial when one mode is not enough or not available. For example, for the blind or visually impaired, vision is not available. Even for the general population, when light is dim or nonexistent, vision is not enough. In those cases, other modes can replace or augment vision.

Theoretically speaking, cognitive load theory (Sweller 1993; Sweller 1994) and dual-coding theory (Paivio 1986) provide a framework for investigating multimodality. Cognitive load theory assumes that people possess a limited working or short-term memory, and for effective information processing, users need to reduce all unnecessary cognitive loads. Based on the fact that each mode has its own working memory, dual-coding theory states that information can be encoded in different modes each with its own working memory. According to the dual-coding theory, visually presented information is processed in visual working memory, whereas auditory information is processed in auditory working memory (Mayer and Moreno 1998). Thus, effective working memory capacity can be enlarged by using multiple channels, so the cognitive load associated with split attention can be reduced by presenting information in dual rather than unitary mode (Mousavi, Low, and Sweller 1995).

Students from classes in School of Information Studies at University of Wisconsin - Milwaukee participated in the experiment voluntarily. A pre-screening interview asked whether they were frequent video game player with more than 10 years of experience. This screening eliminates several “non-gamers,” primarily non-traditional returning students. 42 qualified players’ records were included in this study.

A PlayStation 2 game console was used with a video game titled Gran Turismo 3. Gran Turismo 3 was a relatively simple driving game not requiring so many button-actions like other sports or role playing games. In Gran Turismo 3, for every session, Arcade Mode, Single Race, and “Easy” Level were selected. Then, Swiss Alps (Area A) for the track, Rally Cars (Dirt Car) and IMPREZA Rally Car Prototype for the car, and Automatic transmission were selected.

A Dual Shock 2 game pad, which was included in PlayStation 2 game console set, and MadCatz’s MC2 MicroCon Racing Weel and Pedals were used as game controller in the experiment.

Participants were asked to play the video game with the given setting of Level, Track and Car Type 10 times in total. The completion time of each game was recorded. After the game session, participants were to ask to answer a post-questionnaire to confirm their reactions to the force feedback.

They played 6 games with the wheel and 4 games with the game pad. Since a game pad is thought to be more accustomed, fewer games with the game pad were thought to be reasonable due to the time constraint for the experiment. Each game controller had two options in terms of force feedback effect: a half with force feedback effect, and the other half without it. As a result, the following 2 by 2 matrix was formed for independent variables. The dependent variable was the completion time of each condition. To minimize a potential learning effect, each independent variable was randomly assigned by Latin Square method.

For each combination, the fastest completion time was compiled for statistical analysis with SPSS. A multivariate test showed that there is a significant interaction between two factors (force and controller type). Figure 1 shows the interaction by crossing two lines.

While there is little difference between “with force feedback” and “without force feedback” for performance, there is a significant difference between those two controllers. The reason for the worse performance of the wheel is thought that its rather smaller size than a real wheel caused not a good control condition. An interesting finding was that with the given game pad, the performance without force feedback is better than without it, while with the given wheel, the performance with force feedback is better. It implies that the effect of force feedback depends on other environments. The much better performance with a game pad may suggest that future automobiles might consider the change of the wheel into a pad-like controller due to its easy handling.

Research showed that users’ perceptions or preferences are not always consistent with their performances. Often even if they do not have better performance, still they like certain conditions. This phenomenon has been confirmed in this study. It was found that regardless the controller type and with or without force feedback, even if the participants thought force feedback effect helped their performances, in fact, it did not help, and vice versa. Similarly, as seen in Figure 2, participants who responded the force feedback effect was annoying had better performances, especially with the game pad, than those who said no.

However, force feedback seemed to give some advantage to those who thought more fun with force feedback in video game. As seen in Figure 3, the participants who replied that force feedback added more fun, had better performances than those who replied no difference.