Psychological Scientists in the Private Sector
More Time for Research
Peter Pirolli, Xerox PARC
Most academics are baffled when I tell them I switched from a tenured position in academia to a job in industrial research mainly because of my desire for greater academic freedom to pursue new areas of fundamental research. This switch is baffling to them because it doesn't fit with the strange value system shared by those inside the ivory tower: that any career outside of academia is the mark of failure, a fall from the self-directed pursuit of enlightenment into the drudgery of mundane nine-to-five work.
This value system does not seem to exist in the outside world. For instance, Silicon Valley is permeated with stories of success - bordering on the kind of worship heaped on rock stars - involving people who exhibit individualism, intellect, insight, determination, hard work, and entrepreneurial spirit. None of these stories seem to involve academics except those who left the university to pursue a dream, the same ones who are probably marked as failures by their colleagues and graduate students.
My point isn't to praise celebrity worship; it is that one should always question the value system that one is buying into. After all, who is it you want to please? My most successful graduate students were those who found careers that really fit their dreams and ambitions. On the other hand, I know colleagues who are utterly miserable in academia because its standards are so at odds with their own personal goals.
I should explain how my psychological training and experience led me to my current position. Throughout my career, my goal has been to understand complex learning and information processing. Basically, my current job is to do research on human-computer interaction (HCI). Sometimes this involves trying to understand and improve technological innovations in user interfaces, like the invention of new information visualizations. Other times, this involves trying to develop a scientific base of understanding from which new technologies might emerge, such as our current work on information foraging theory, which deals with how humans interact with complex information ecologies like the World Wide Web. In some cases, psychological insights lead directly to new inventions that get patented.
My undergraduate degrees are in anthropology (mostly linguistics and cultural anthropology) and psychology (mostly cognitive and neuroscience). It turns out that psychology and anthropology are among the central disciplines in the field of human-computer interaction. The ability to conduct field work in real-world settings is highly valued in the world of HCI. Unfortunately, field techniques are not commonly taught to psychology students. This is unfortunate because cognition in real-world settings is so highly evolved compared to what is studied in the typical laboratory experiment. Generally, cognitive psychologists are trained as "experimentalists" but not as "naturalists," and cognition in the wild is full of wonderful phenomena that have yet to be studied. Also, as an undergraduate, I was fortunate to learn about ethology and behavioral ecology, which eventually seeded my current theoretical development of information foraging theory.
I got my PhD in cognitive psychology from Carnegie Mellon University, a unique place where people like Allen Newell, Herbert A. Simon, and John R. Anderson brought together fundamental insights about computation and cognition, and especially the importance of understanding how the complex structures of tasks and the environment are reflected in complex behavior. Along with this theoretical perspective came training in task analysis, the analysis of verbal protocols, and writing computational simulation models. These methods also turn out to be central to a lot of work in HCI. For instance, the GOMS1 analysis of user interfaces uses variations on verbal protocol analysis, task analysis, and an approximate computational model of cognition.
I was also fortunate to have been steered into taking upper division and graduate courses in computer science, and Bayesian probability and statistics. I cannot overemphasize how important it is to have a solid base of computer science, probability, statistics, and mathematics in the world of industry. These are the universal languages of the scientists and engineers that surround me.
Fresh out of graduate school, I took a position at the University of California, Berkeley, in the School of Education, and as a member of the Cognitive Science Institute. A few years later, I got tenure, but I found that I was required to spend an inordinate amount of time on tasks that had nothing to do with research or teaching students. A time sampling study of my activities revealed that I was spending less than 10 percent of my time on research, and that time had to be squeezed out at the margins, i.e., evenings and weekends. One might say I only had academic freedom 10 percent of the time, and that was practically on my own time. Particularly bothersome was having to deal with committee meetings and the associated tasks of reading and writing endless internal reports and memos. In my opinion, academic politics and administration are just a huge waste of time.
When a research position opened at Xerox PARC, I took a leave of absence to check it out. I discovered I had more time to pursue research problems that were in the interstices among disciplines. Two current examples are research on the use of information visualizations and information foraging theory, both of which I am doing with Stuart Card and others in our group.
Our psychological research on information visualizations is at the intersection of visual attention, visual search, and problem solving. There are pretty good theories of visual search and attention, and pretty good ones about problem solving, but there is a huge chasm between the two; that's where we get to do some fundamental research.
Our research on information foraging theory was motivated by our interest in developing systems for finding information, making sense of it, and using it, especially on the World Wide Web. Revisiting my interests in behavioral ecology, I found a way to merge optimal foraging theory (mathematical models that explain food seeking behavior) and computational cognitive models. Out of a practical problem and a mix of disciplines, we forged a mathematical-computational theory that deals with how people shape their moment-by-moment information-seeking behavior to the flux of information in the environment.
Another great thing about working in an industrial research lab is the greater degree of collaboration and the greater variety in collaborators. Universities and their funding agencies are structured to reward individual work rather than collaborations, but collaborations are really necessary to tackle complex research problems. Consequently, these institutions systemically hinder interdisciplinary research, a territory that holds many gold mines. Academics may be free to delve into that territory, but it requires working against the system.
I have had the opportunity to work with physicists (on models of probability distributions of length of surfing paths), and most often with computer scientists. These latter collaborations range from helping develop technologies for browsing, search engines, and caching/prefetching, to formulating interface principles and designs, to evaluation studies of prototypes. It is really important that a psychologist in industry have an impact on design or engineering. One has to generate "intellectual property." It is very easy to fall into the role of being just an evaluator or critic - a psychologist who runs studies on existing systems - but that is not a sustainable position. In industrial research, fundamental knowledge has to have some path to practical use.
When I was in the midst of deciding to go to Xerox PARC, all of my academic colleagues warned me that I would no longer have graduate students to do my work (yes, this is pretty much how they put it). Contrary to myth, there are many opportunities for graduate students to intern in industrial research, and I have been very fortunate to have worked with some exceptionally bright and hard-working students and post-docs. We have used external funding (the Office of Naval Research) to support some of these students, and PARC itself funds a variety of undergraduate and graduate internships.
Working in a multidisciplinary area like HCI, and working on both fundamental and applied research does increase the amount of research that one has to track. Researchers at PARC are encouraged to publish, participate in the same professional circles as our academic colleagues, and act as editors and reviewers. I subscribe to eight psychology or cognitive science journals, four computer science journals, and half dozen more across other areas. I closely track two dozen more through an electronic service, and I use the Web. I also find myself tracking the business world more closely through trade magazines. In addition to going to conferences and keeping up with my academic colleagues, I now also find myself talking to people who have commercial interests who want to partner with our group in some way. This is also a great way to find out what's really going on out in the world and what new problems are arising. Although my bread-and-butter research takes place in the laboratory, I am always enamored with the intricate complexities of how people in the real world engage their environments through ever-evolving technology.
1An engineering model of human performance, GOMS (Goals, Operators, Methods and Selection rules) was introduced by Stuart Card, Tom Moran and Allen Newell in the early 1980s. It has been shown to be a valuable tool in real-world HCI design, having been verified with extensive empirical field research. See the October, 2000 issue of the Observer for Card's views on working in the private sector.