Psychological Scientists in the Public Sector

Applications Drive Theory

Stuart Card, Xerox PARC

Stuart Card Nothing drives theory better than a good applied problem. At Xerox PARC, I'm currently working with Peter Pirolli on a theory of how people access information - "information foraging theory" - and how we can invent and model forms of external cognition, such as dynamic information visualization displays, that graphically-agile computers make possible. We use many behavioral science methodologies, from field observations, to experiment, to eye-tracking, to formal cognitive modeling.

My interdisciplinary background has been directly relevant to my work. As an undergraduate, I was a physics major at Oberlin College. Later, I briefly was the director of Oberlin's computer center. After hearing Herbert Simon give a talk, I decided to go to Carnegie Mellon University to study with Allen Newell and Simon in the Systems and Communications program, part of which was based in psychology. When that program ended, I defined my own program in computer science and psychology. From there, I went to Xerox's Palo Alto Research Center, where I have been ever since, (although I also have been an adjunct professor at Stanford).

My colleague Tom Moran and I, along with Newell as a consultant, have been given enormous leeway to explore theories and studies in the area of human-machine interaction. Among other things, our behavioral studies were responsible for the commercial introduction of the mouse. It was our theory of mouse movement, not the empirical results, that finally beat down the objections of a hostile group of engineers toward the introduction of such an odd device.

In developing theories and models, I draw both on my psychological training and my training in physics. The most useful part of my psychological training was probably in cognitive modeling - how to do protocol analyses, and how to make computational models, since those dealt with mechanisms underlying behavior. Neuroanatomy and physiology were also useful, because of the contact with the structures of the brain. My introduction to the principle of bounded rationality was key.

Essential Differences
There are some essential differences between academic research and industrial research. My experiences may not be universal, but the differences that I have found include:

  1. There is greater access to applied problems. In a university, applied problems also affect the agenda for research but much more generically and indirectly and at a time lag through government agency funding priorities. It's a good place to discover new classes of problems and be the first to work on them.
  2. Progress is often made at the interstices of disciplines, where methods developed in one discipline can be applied to problems in another. That's difficult to do in universities, because the structure is by discipline, but it's easier in industry, where the management structure for a project can be put together more flexibly and more rapidly.
  3. We are fully funded for a base program without the need to write grant proposals. There is no need to worry where summer salary comes from. Also, I believe travel and capital needs for research tend to be met more easily. Government contracts add to this base.
  4. There is more pressure for tangible results in industry. What the corporation cares about are prototypes of things that could be products and especially the transfer of those to operating companies. It also cares about quality scientific work, but the idea is that eventually that work leads to something. But this is exactly the environment I was looking for to have a pressure on psychological theory for producing practical results.
  5. There is no automatic or tacit acceptance of the paradigms of psychology, since most industrial colleagues are not psychologists. Instead, the value and validity of these paradigms has to be established in argument. My closest colleagues are from computer science, electrical engineering, and theoretical physics. We are all evaluated against each other on the same scale (by a non-psychologist). That creates more pressure for the theories to put out results that can be appreciated across sciences.
  6. In industry, a scientist is in a natural position to be an entrepreneur. Ideas that seem to have merit can be licensed, transferred to an operating division, or used to start a new company. Of course, with effort, this can also be done from a university. But industry is set up for the purpose. Doing the deals to move ideas into the real world can be fun, and this also provides opportunities for demonstrating practical results from psychological theory. We have had a hand in launching three new businesses using our research results in the last few years.

One final point: A good way to disseminate a theory is to embed it in an artifact. In an industrial research laboratory, the coin of the realm is prototypes. For that reason, my group not only does behavioral research, but also designs and builds prototypes for novel forms of human machine interaction. That way, we make sure the ideas will get into application form.

American Psychological Society
October 2000
Vol. 13, No. 8

©2001 American Psychological Society
Current location: APS | News & Research | APS Observer
About APS   |    Press Kit   |    Links   |    Observer   |    Member Services   |    Human Capital Initiative