The JTEC panel did not expect to answer all of these questions or have opportunities to discuss many of these issues in depth during its brief visit to Japan. However, the panel was able to visit a very diverse collection of research labs to obtain an overall impression of the types of HCI research in progress and develop a feel for the general directions of research and the areas of interest.
Although there is an increase in HCI research activities at Japanese universities, most research takes place in industrial labs. Government support for university-based HCI research is growing, and the top universities are receiving significant amounts of these government funds. As an example, one university was reported to have received the equivalent of $50 million, and another was provided with funds to equip a large HCI research center.
Strong corporate investment in HCI research and development is expected to continue. At most sites visited, the Japanese reported investments in R&D of 7-8% of sales. The best corporate labs are very similar to corporate research labs in the United States but are not experiencing the same downsizing as their U.S. counterparts. Overall, the Japanese labs visited had a similar feel to labs in the United States. For example, NTT is quite similar in a number of respects to Bell Labs or Bellcore, and ATR is similar to MCC in its earlier days.
The JTEC team saw projects that were similar to projects underway in the United States. At most sites the team saw a standard suite of demonstration projects. They were comparable in focus, approach, and execution to projects that visitors would see in most U.S. HCI labs. In addition, the hardware and the software development environments that the team saw were also quite similar to those employed in the United States.
While the character, focus, and general style of research was similar to current U.S. work, there was considerably more activity in the areas of speech recognition, applications of fuzzy logic, and facial animation than there is in U.S. labs. The former no doubt is explained by the difficult interface problem that the Japanese language presents interface designers, especially in a world market in which Roman alphabet interfaces are so dominant. One can conjecture that the success of fuzzy logic in control applications in a host of Japanese products, and the general importance of developing more adaptive interfaces, are reasons for the quite extensive Japanese fuzzy logic research activities. The much greater interest in facial animation may arise from the same source as the considerable recent interest in Japan in what is termed kansei engineering: it deals with all of the noncognitive aspects of human experience with a view of how they might be represented in a computer.
Work in Japan also parallels U.S. research activities in investing more in hardware than in empirical investigations or software development environments. For example, as in the United States, there are very expensive video-on-demand trials underway. The JTEC panel visited one project that started in July 1994 and is to cost ¥10 billion (about $100 million) over a three-year trial. Like similar U.S. projects, this effort could benefit from incorporating a more user-centered design approach and more carefully planned empirical evaluations. On the other hand, while it is not yet as strong as in the United States, there is certainly a growing focus in Japan on cognitive science research and an appreciation of its import for human-computer interaction. Cognitive science activities are particularly strong at a few universities such as Chukyo, Hokkaido, and Keio, and at several industrial research labs, especially NTT.
A final overall impression is of the importance of general corporate philosophies and the key role given to human-computer interaction as a component of those philosophies. Table 2.1 includes some examples.