Our example is from Technische Universiteit Eindhoven, where Joep Frens designed a camera with a rich interaction user interface and compared it to conventional cameras. The standard interaction approach in industry is based on a menu on a screen, which can be navigated with buttons. Frens aimed at creating an alternative to this standard approach.

While conventional digital cameras typically have controls based on buttons and menus on screen, rich interaction cameras had tangible controls. For example, with a rich interaction camera, the photographer could take a picture by pushing a trigger and save it by pushing the screen toward the memory card. To delete it, he had to push the screen back to the lens. Frens designed these unconventional forms, interactions, and functions so that the photographer could read the possibilities for action and function from the form (Figure 4.1).

Building on ecological psychology and literature on tangible interaction, Frens created a series of hypotheses for each camera variation to be able to compare user experience. Frens’ main hypothesis was that a rich interaction camera is more intuitive to use than a conventional camera. Another hypothesis was that people think it is more beautiful. The cameras were stimuli in his study; measures for things like use and beauty came from Marc Hassenzahl, a German design psychologist. ⁷

Frens’ design process was driven by a wish to find alternatives for the prevalent industrial interaction paradigm for cameras. His inspiration came from his knowledge of trends in interaction design and from his background research, not from user studies. Research questions, hypotheses, and the rich interaction framework came after the first designs, and they were based on the insights gained during the design process (Figure 4.2 and Figure 4.3).

Frens built several camera prototypes out of cardboard and tested these with students. Having a good idea of how to build a rich interaction camera, he built the body from aluminum. He also built several control modules that could be fitted to the body. The first module was with conventional controls; the second with light controls; the third with mixed controls; and the fourth with rich controls, all within the same form language. These designs formed a scale. At one end was an interface using conventional menus and buttons. At the other end was a radically reworked camera with tangible controls only.

He went on to test these cameras with 24 students of architecture in a laboratory setting. Each student received instructions, viewed the camera, and took photographs with it. ⁹ This study was repeated four times, once for each camera variation. ¹⁰ User experience was evaluated with a questionnaire. Afterward, the participants compared the cameras and completed a closing questionnaire.

From a cognitive psychology perspective one would expect that the rich interaction camera would do worse than a traditional one, as it breaks the user interface conventions of cameras. However, it did not, and it was appreciated by many of the participants. The rich interaction camera did not fare better than other cameras in aesthetic and practical terms. Still, Frens was able to say that his design was successful. At a more fine-grained level, the results were also positive. For example, saving images was found to be pleasing with the rich interaction camera, which participants also found more beautiful than other cameras.

Studying things in a laboratory means that something is taken out from its natural environment and brought into a controlled area where it can be subjected to experimentation. Almost anything can be studied in the laboratory: armies, design, chemical reactions, rich interaction, and so forth.

The trouble with studying a phenomenon in the real world is that usually many things shape it. This makes it difficult to find what causes something one sees; there are typically several possible explanations, and it is impossible to rule any of them out with a high degree of certainty. Research becomes an exercise in “what about if.…”

Studying a phenomenon in a laboratory helps with this problem. The laboratory gives the researchers an opportunity to focus on one thing at a time. Most typically, this “thing” is a relationship, such as the relationship of rich interaction and user experience in Joep Frens’ study. The laboratory also helps researchers study alternative explanations and competing hypotheses; doing this is far more difficult in natural settings. After researchers have eliminated alternative explanations, they are able to confidently say things about how rich interaction improves user experience in camera design. It is possible that the results are wrong, but this is highly unlikely.

Working in a laboratory has many other benefits. For example, a laboratory can be equipped with instruments that help make detailed and accurate observations and measurements.

As the laboratory environment and experiments are typically documented in detail, it is also possible to replicate the study in other laboratories. This rules out errors coming from the setting and its research culture. This also applies to issues like “observer effects” — the researcher giving people cues about his intentions. If people understand a researcher’s intentions, they may change their behavior to please or to confuse the researcher. There has to be a proper strategy to deal with these effects. ¹² Most threats to validity are beyond this book. For example, to see how things like user experience develop over time, researchers often study the same people several times in so-called time series analysis. However, over time, people in the study learn about the study and change their behavior. There are ways to minimize such threats, but this complicates research design. ¹³

The crux of any laboratory study is experimentation. ¹⁴ The researcher manipulates the thing of interest in the lab to learn how people react to it while holding other things constant. Typically, he assumes a new design will improve things like user experience when compared to older designs. In research language, the null hypothesis predicting no change is rejected. Having established the basic relationship, researchers study other explanations to see whether they somehow modify results. Ideally, researchers vary one additional variable at a time to see whether it alters the basic relationship. ¹⁵

For example, Frens studied user experience first by varying his camera designs and learned that a rich interaction interface improved user experience. He could have gone further; for example, he could have studied men and women separately to see whether gender somehow was relevant in explaining the link between rich interaction and user experience. However, he chose not to do these additional analyses, keeping his focus on the basic relationship only (Figure 4.5).

Research is successful when the basic relationship exists and the most important competing explanations are ruled out. Thus, if a rich interaction camera functions as expected and there are no serious alternative explanations, the theory about rich interaction ought to be accepted until a better theory comes along. If the first rich interaction camera of its sort is already about as good as conventional cameras, even though cognitive theory would predict otherwise, something must have gone right in the design process.

Selecting what to study and what to leave out is ideally a matter of theory, but equally often, it is also a matter of judgment. Many things influence human behavior, and it is impossible to study everything carefully. What is included is a theoretical question; for example, when Philip Ross selected people for studying his lamp designs, he selected university students with a similar value system in mind (see Chapter 8). ¹⁶ This limited his ability to generalize but also made his analyses easier. He would have gained little from knowing how people with different medical conditions would have reacted to his lamps. This might be an interesting question for another study but not for his.

There are also methodic ways to make research designs simpler. The most popular technique is randomized trial, in which researchers take one group of typically randomly selected people. Then they repeat the study with another randomly selected group. One of the groups is given a “treatment,” for example, they have to use a rich interaction camera. The other group gets a placebo, for example, a conventional camera. Researchers measure things like user experience before and after the treatment in both groups, with the expectation that satisfaction has increased more in the treatment group than in the control group. Randomization does not eliminate variables like gender or horoscope sign, but in large enough groups the impact of such variables evens out.

In constructive design research, the epitome of analysis is an expression such as a prototype. It crystallizes theoretical work, and becomes a hypothesis to be tested in the laboratory. For Stephan Wensveen, it was the alarm clock he built. For Ianus Keller, it was a cabinet for interacting with visual material. For Andres Lucero, it was a space he called the “funky-design-space” meant to support moodboard construction and browsing. ¹⁷ For Joep Frens, his camera variations are considered physical hypotheses.

However, adding the design phase to research also adds an important new ingredient to the soup — the designer’s skill and intuition.

Stappers recently described some of the complexities involved in treating prototypes as hypotheses. In the design process, a prototype integrates many types of information. Theory is one component in the prototype, but not the only one. For example, ecological psychology tells us to build tangible controls that use human sensory motor skills to accomplish tasks like taking photos or setting the time for the alarm. However, it says little about many of the sensual issues so important for designers, including shape, colors, sound, feel, surfacing, and so on. Ideas for these come from the designer and the design process. As Stappers said:

Elsewhere, Stappers listed some uses of prototypes. They can be used to test a theory, in which case they become embodiments of theory or “physical hypothesis.”¹⁹ However, they also confront theories: researchers, whose prototypes cannot hide in abstractions, have to face many types of complexities that working designers face. Similarly, they confront the world. When building a prototype, the researcher has to face the opinions of other people. Furthermore, they serve as demonstrations, provocations, and criticisms, especially to outsiders who have not seen their development from within (Figure 4.7). ²⁰

As Stappers points out, prototyping is more than theory testing, it is also a design act. A design process may be inspired by theory, but it goes beyond it. A prototype is an embodiment of design practice, but it also goes beyond theory. For this reason, design prototypes are also tests of design, not just theory. Indeed, one of the most attractive things in research in Eindhoven has been the quality of craftsmanship. These designs can be evaluated as design statements. They are good enough to please a professional designer aesthetically, structurally, and conceptually.

Research sets some requirements for prototypes at odds with doing good design. Researchers almost invariably aim at simplification; for example, people bring in many types of aesthetic opinions to the laboratory and are barely aware of most of them. The way to control this is to eliminate clutter by keeping design simple. When the subjects’ mind does not wander, changes in their behavior can be attributed to the designs. As Overbeeke wrote with his colleagues:

This is where there is tension. As Stappers noted, research prototypes are not pure expressions of theory; they also embody design values. The more they do, the more difficult it becomes to say with confidence that the theory that inspired design actually works. The secret of success, quite simply, may be design.

This is a catch-22. On the one hand, the more seriously researchers take design, the more difficult it becomes to draw unambiguous theoretical conclusions. On the other hand, when the theoretical frame and the aims of the study guide prototyping, ²³ a good amount of design relevance is sacrificed. Ultimately, the way in which prototyping is done is a matter of the researcher’s personal criteria for quality and taste. Most design researchers think design quality is more important than theoretical purity, but opinions differ.

Most design researchers, however, find it easy to agree that research prototypes differ from industrial prototypes. As Joep Frens noted, his cameras are finished enough for research but not production ready.

As Frens related, prototypes are done to see where theoretically informed design leads. Issues like durability, electric safety, and the quality of computer code are in the background. In the foreground are things that serve knowledge creation; it is better to leave concern for production to industry.

When things are taken from society to a laboratory, many things are decontextualized; however, this comes with a price. A laboratory is a very special place, and things that happen in the laboratory may not happen in society or may happen in a different way, as conditions are different. Do results of laboratory studies tell anything about real world?

There are several ways to answer this question, and sometimes this question is not relevant. Researchers may want to show that a certain outcome is possible by building upon it, and there is no need to produce definitive proof beyond the construct. This is called “existence proof.”²⁵ This proof is well known in mathematics and is common in engineering but almost non-existent in empirical research. Sometimes, generalization happens to theory, and this is typical in many natural sciences — a piece of pure tin melts at the same temperature whether it is in Chicago or Patagonia.

Usually the jump from the laboratory to the real world builds on statistics. Many things in ergonomics may be universal enough for theoretical analysis, but it is more difficult to argue this in, say, aesthetics or design. ²⁶ Researchers can calculate statistics like averages and deviations for those people they studied. They cannot, however, use these figures only as estimates of what happens in larger populations: it is always possible to err. The basic rule is that as sample size increases, confidence in estimates increases.

Finally, proof goes beyond one study. As Lakatos argued, there is no instant rationality in research. ²⁷ Generalizing from individual studies is risky. However, if a program repeatedly leads to interesting results, it should be taken seriously. It was only after hundreds of studies that the world came to believe that asbestos and tobacco caused life-threatening illnesses. This logic also applies in constructive design research (Figure 4.8).

In actual research practice, these proofs coexist. Again, Frens provided a good example. In terms of an existence proof, his camera shows that it is possible to build rich interaction cameras. In terms of generalizing to theory, his research framework can be applied in many different circumstances. In statistical terms, his empirical results apply to people with a high level of aesthetic abilities and a great deal of experience in photography. Finally, by now dozens of designs coming out from the Netherlands show that ecological psychology can be fruitfully applied to designing interaction. The burden of proof does not lie on Frens only.

There is a margin of error in even the best of studies; however, after reading Frens, we know more about rich interaction in smart products than before. Researchers are skeptics and do not easily accept anything, but when they accept something, they stand behind it. Proving that a good study is wrong requires a careful study that shows in detail what was wrong.

Recent work in Eindhoven has taken a significant step away from its basis in ecological psychology. Earlier work built on J.J. Gibson’s ecological psychology and aimed at formulating conditional laws and constructing mathematical models of actual interaction. ²⁸ Since Kees Overbeeke’s inauguration in 2007, however, research has turned to phenomenological and pragmatic philosophy. ²⁹ Ecological psychology worked marvelously well when creating systems that can be used with one’s body instead of cognition, but it gave few tools to see how things like social interaction and culture shape conduct. It remained limited in its ability to conceptualize reflection, thinking, and discourse, and these are just the things in the center of the most recent work coming out of the program.

For people trained in philosophy, this may sound risky. Phenomenology makes few claims about explaining human conduct. However, there is a tradition of experimental psychological phenomenology, and Maurice Merleau-Ponty, the key philosopher of psychological phenomenology, built many of his theories on reinterpretations of clinical and experimental studies. ³⁰ Also, there are ways to combine careful measurement and phenomenologically informed theory. ³¹ William James, one of the founding fathers of the pragmatists, had a background in medicine, making him no stranger to experimental research. Here the line between philosophical thinking and experimental research is fine, but as long as a researcher keeps in mind that experiments are aids to imagination, he is on safe ground. ³²

The push came from Paris where philosopher Charles Lenay and his colleagues recently studied how we perceive other people’s perception with our bodies. It is one thing to be alone somewhere and to see things; it is another thing to be there when others see me seeing things. I have to take into account other viewpoints and how they change as time goes by. ³³ With his colleagues, Lenay studied Tactile Vision Sensory Substitution, a system developed by Paul Bach y Rita. This system transforms images captured by a camera into three-dimensional “tactile image,” which is applied to the skin or tongue. It is meant to give blind people a three-dimensional perception of their surroundings. For Lenay and his colleagues, this commercial and prosthetic failure provided an opportunity to study how technology mediates perception. They showed what people are able to say when they interact with another person through the system. We give a body image to others, and others respond to this image. We can recognize this response. We know we do not deal with a machine.

This may sound like hairsplitting, but it has important implications for understanding many technologies that are specifically built on social assumptions. People do think about the image they give others because they know that others act on this image and that way affects them. Many “presence” technologies build on similar assumptions, and such considerations certainly shape behavior on social networks on the Web. This question is relevant in traditional design as well. For example, we dress differently for Midtown than we do for the neighborhood bar in Brooklyn because we know people will look at us differently in those places. It is much like looking into a mirror: you become more self-conscious.

This work has recently been picked up in Eindhoven. For example, in her master’s thesis, Eva Deckers weaved a carpet that responds to movement of the hand, giving it the ability to perceive and react to people’s perception. In her doctoral work, she focused on situations with multiple users. ³⁴ Work like this is bringing the Eindhoven program to a junction. These studies promise a better understanding of many kinds of interactive technologies but also make research more difficult. Importantly, this new direction creates links to twentieth century Continental thinking in the humanities and the social sciences. Recent work is also bringing work in Eindhoven closer to interpretive sociology, ethnomethodology, and discourse analysis. This shift has certainly paved the way for more sophisticated research programs that may fill the promise of providing knowledge of “la condition humaine” — what makes humans tick. ³⁵