Research Psychology at Microsoft

I have always had an interest in science, and my passion for brain research and visual information processing eventually led me to a PhD in cognitive psychology. As part of the PhD process, I was introduced to computers and, hence, the programming of experiments, the analysis and modeling of data, and even the presentation of relatively simple graphics on Tektronix displays. I was excited by the (then) new area of research, where psychological issues merged with computer metaphors and models — the field of human-computer interaction (HCI). However, I was not entirely sure that I wanted to leave academia. As luck would have it, a “two-body” problem landed me in a job at Bell Communications Research right out of graduate school. My job title was “member of technical staff,” and my responsibilities included writing the specification for a user interface to packet-switched networking software.

Considering that I was trained formally as a cognitive psychologist looking at attention and automatic processing, this was a stretch from my area of expertise. I had no understanding of HCI standards or guidelines or even for how my background in psychology would eventually play such a crucial role in my career. I can still remember the first three weeks on the job at Bellcore, reading an intimidating stack of books on HCI from the extensive library we had on campus. Soon I learned that the analytical skills, in addition to the good communication and programming skills, that I had picked up through my graduate training (thanks to Rich Shiffrin!) were tools to help my career take off in the area of iterative design and evaluation of complex computer systems for human users.

Psychologists have a toolkit of paradigms for answering questions that often come up in design discussions, and we know how to run experiments that can quickly bring crucial data home to design teams, which would otherwise make less well-diagnosed decisions based primarily on brainstorming. Psychologists also have an objective eye and utilize user data, not design intuition or their own habits, to guide complex computer system design. These skills, if communicated well and performed in a timely manner, quickly become incorporated into the design process, and the research psychologist can see their research efforts delivered as hardware and software products used by millions around the world. This was the case for me, as I traversed job opportunities from Bellcore to NASA-Johnson Space Center (via Lockheed), Compaq Computer Corporation and finally Microsoft.

After managing a group of usability engineers in Microsoft’s Interactive Media Group for over two years, I decided that it was time to return to my true passion — research. In 1997, I joined the User Interface Research group (now called Adaptive Systems and Interaction) at Microsoft Research, and I have remained happy and engaged in my research with my colleagues to this day. I have had the benefit of working on myriad research projects that keep me on the cutting edge of technology in artificial intelligence and information visualization. My role is to provide the “usefulness and usable” check on emerging technology and to refine our research projects in the face of user data iteratively over time. In the following few paragraphs, I outline a couple of recent examples of my research.

How can consumers handle the large number of digital photos they collect online? Can automatically grouping a user’s personal photographs into clusters improve the speed and/or ease of browsing those photographs? Together with John Platt, a senior researcher in the Signal Processing group, we addressed this question by first building and testing an image browser based on clustering. In response to issues highlighted by an initial user study, we created an improved clustering-based image browser. The new browser creates an automatically generated overview of a set of photographs. The browser was tested on users’ own photographs against three other browsers: a hierarchical folder browser, a flat detail view with no automatically-generated overview and the original design. Searching for images with the improved browser produced greater user satisfaction than the other browsers without sacrificing performance. This result shows that automatic clustering of personal photographs is effective — it requires no organization effort by the user and yet facilitates efficient and satisfying searches. Some form of automatic photo clustering will eventually be shipped in Microsoft consumer software.

In the area of hardware and jumbo display design, we previously reported the benefits of large displays for females navigating in 3-D virtual worlds (Tan, Robertson & Czerwinski, 2001). We extended that work with two studies to replicate and understand the gender/display finding under tighter control conditions. Specifically, these studies explored the hypothesis that a wider field of view (FOV) enhances integration of early, piecemeal, cognitive map information when navigating in both dense and sparse virtual environments. We compared the effects of FOV and display size on both male and female users. The first study replicated the gender-selective benefit from a wider FOV coupled with a large display for females. The second study suggested that a wider FOV was useful to females in sparse but especially densely populated worlds. The findings have serious educational and training implications for females who use simulated or virtual environments as their tools and will be presented at the premiere conference on HCI, the Association for Computing Machinery's (ACM) Special Interest Group on Computer-Human Interaction (SIGCHI) 2002.

These examples give just a glimpse of the variety of research opportunities that I have before me every day at Microsoft Research. While I still maintain ties to academia through an adjunct faculty position at the University of Washington, I have never regretted moving into the exciting field of HCI. More information about my projects can be found at Microsoft's website, and Microsoft opportunities can be found on their jobs site. 

(Originally published in the January/February 2002 issue of Psychological Science Agenda, the newsletter of the APA Science Directorate.)