07 November, 2009

Visual Decision Making

User interface experts are often suspicious of the role of visual aesthetics in user interfaces—and of designers who insist that graphic emotive impact and careful attention to a site’s visual framework really contribute to measurable success. Underneath the arguments, I see a fundamental culture clash. In academia, text (and lots of it) is the only way serious people make serious arguments, and very polished presentations are often seen as prima facie that the presenter may be hiding a weak argument with graphic frou-frou.

Many eyetracking studies conclude that large graphics and graphic elements attract few “gaze fixations,” but can we really conclude that large, aesthetic, tone-setting graphics have no lasting effect on the user’s attitudes toward a site? To put it bluntly, are designers who create visually compelling sites simply wasting time and treasure on graphic indulgences that obstruct efficient e-commerce and communication?

Perhaps not. Another body of web user experience research shows that website users are powerfully influenced by aesthetics, and that positive perceptions of order, beauty, novelty, and creativity increase the user’s confidence in a site’s trustworthiness and usability. Recent design writing and interface research illustrate how visual design and user research can work together to create better user experiences on the web: experiences that balance the practicalities of navigation with aesthetic interfaces that delight the eye and brain. In short: there’s lots of evidence that beauty enhances usability.

What can we learn from user experience analysis tools such as eyetracking? Eyetracking is great for analyzing and understanding how users see, interpret, and use information. However, I disagree with usability researcher Jakob Nielsen, who asserts that, since his research subjects (apparently) pay little attention to large graphics on web pages, we can infer that those graphics have little influence on users — and that “useless” or “superfluous” images form “obstacle courses” for users. Nielsen isn’t talking about catalog images or other images closely related to merchandize or tasks, but about the images and other graphic content that designers use to create a site’s aesthetic ambiance.

I have no doubt that Nielsen and other researchers accurately report what their eyetracking study participants do in response to task-centric research on websites. Context is important here: In such studies, participants have a set of specific tasks to accomplish, and thus their gaze tends to focus on navigation links, titles, labels, and interface controls such as buttons and form fields. Expressive or visual tone-setting graphics are rarely useful in such tasks, and it’s not surprising that users performing these tasks (apparently) ignore most page graphics, as indicated by the infrequent gazes directed at large images.

To reconcile the differing views of the proper role of visual aesthetics, we need to understand how the brain processes images and responds to what we see.

It happens in an instant

Thanks to the work of the early 20th Century Gestalt to many studies since—we know that the brain’s response to images is extremely complex, and in many cases nearly instantaneous. The process seems semi-magical and therefore untrustworthy. How could something so complex happen so fast? How can we trust the results of a process we don’t thoroughly understand? Research confirms that users make aesthetic decisions about the overall visual impression of web pages in as little as 50 milliseconds (1/20th of a second). These instant visceral reactions to web pages happen in virtually all users, are consistent over visit length, and strongly influence the user’s sense of trust in the information. In short, users have made fundamental, consistent, and lasting aesthetic decisions about the credibility and authority of sites before major eyetracking events begin.

In intensely visual fields such as art history and diagnostic radiology, this kind of sophisticated, complex, near-instant Gestalt visual judgment is well-known and respected, although the exact neural mechanisms at work are not well understood. In Blink Malcolm Gladwell tells the story of the Getty Museum’s famous kouros, nominally ancient Greek statue now widely regarded as a modern forgery. Although the provenance and mineralogy tests seemed to provide definitive proof of the statue’s age, independent art history experts were almost universally negative in their first visceral reactions to the statue. Thomas Hoving’s immediate reaction to the statue’s unveiling was “fresh,” hardly the right word for a sculpture that had supposedly been buried for two thousand years. At first, the experts had only their gut visual reactions as proof, but their skepticism caused the Getty to reconsider, and the evidence supporting the statue’s age and provenance fell apart on closer inspection.

Although Jakob Nielsen has been one of the most vocal skeptics of highly graphical sites, his business partner, Donald Norman, has best articulated the ways in which sophisticated visual design not only influences users, but also contributes to better usability.

Why attractive things work better

In psychology, emotional reactions to stimuli are called affective responses. Affective responses happen very fast, and are governed in an automatic, unconscious way by the lower centers of the brain that also govern basic instincts (food, fear, sex, breathing, blinking, etc.). Think of affective responses as the brain’s bottom-up reaction to what you see and feel. Cognitive responses are your brain’s slower, top-down, more considered responses. They’re governed by your personal cultural views, learning, experiences, and personal preferences that you are aware of and can easily articulate. Affective reactions assign value to your experiences; cognitive reactions assign meaning to what you see and use.

Affective and cognitive responses to visual stimuli are governed by a three-stage process in the brain, at visceral, behavioral, and reflective processing levels:

The three-stage process governed by affective and cognitive responses to visual stimuli.

The visceral (“gut”) processing level reacts quickly to appearances. It’s the visceral reaction to web pages that researchers measure when they detect reaction times as fast as 50 milliseconds. It’s crucial to understand that these instant good/bad visceral-level affective responses are largely unconscious: it can take seconds or minutes to become consciously aware of your first, visceral reaction to a stimulus—particularly a stimulus as complex as a web page.

Behavioral-level processing involves the more familiar aspects of usability: it responds to the feel of using the site, the functionality, the understandability of the structure and navigation, and the overall physical performance of the site. At this level, users are consciously aware of their attitudes toward the behavior of the system, and their reactions (pleasure, for example, or frustration) play out over seconds and minutes as users interact with a site. It’s at this behavioral level that techniques such as eyetracking are most powerful and trustworthy, because they offer detailed moment-by-moment evidence of what users consciously decided to look at and do to fulfill a given task.

Reflective processing of reactions is the most complex level, and typically involves a user’s personal sense of a site’s beauty, meaning, cultural context, and immediate usefulness. Reflective processing often triggers memories and encourages pragmatic judgments about the overall aesthetic worth and value of what a user sees. Eyetracking and traffic logs are irrelevant at this level, but user interviews can give you insight into your user’s reflective judgments.

Visceral (affective) reactions can take a relatively long time to bubble up through layers of processing to enter conscious awareness at the behavioral or reflective levels, but that doesn’t mean that affective reactions don’t immediately influence thought. In fact, it’s the instant, pre-conscious pleasure of seeing a well designed page that makes you predisposed to find a beautiful design easy to use—an effect that lingers long after the slower, conscious behavioral and reflective levels of processing kick in and make you aware of how you feel about what you see.

Classical and expressive aesthetics in web design

To explore viewer reactions to site design, user experience researchers need basic aesthetic models that simplify a complex subject, but nonetheless do offer some guidance to designers. These models are often divided into “classical aesthetics” versus “expressive aesthetics.” Users bring context, expectations, and intentions with them, and the extent to which a site’s visual design matches those expectations will strongly influence perceptions of the quality and reliability of information or site interactions.

Classical aesthetics

Classical aesthetics stress orderliness and clarity in design, and use familiar web and print conventions. Research shows that classical aesthetics strongly correlate with perceived usability. This model is most successful when users expect an e-commerce or content site to provide large amounts of information in a well-organized way, with a clear visual hierarchy and conventional headers, subheads, captions, and navigation—think newspapers, sophisticated magazines or e-commerce sites, or reference sites such as Wikipedia. The pleasure and trust that classical aesthetics generate seems to be centered in the higher and slower reflective level of processing. However, classically beautiful sites also generate instant positive visceral reactions as well, and this creates a lasting sense that a site is easy to use.

Expressive aesthetics

Expressive aesthetics emphasize the originality, creativity, and visual richness of the site design. Think about the kind of site that typically wins design awards from Print or Communication Arts. Successful expressive designs generate immediate, positive visceral reactions in most users. It is less clear that this immediate expressive response influences longer-term positive judgments as strongly as successful classical designs, particularly in perceived ease of use. Understanding your audience is particularly crucial in expressive designs: a “successful” design approach for foreign policy wonks may be less successful with fans of the Jonas Brothers.

Expressive and classical aesthetics seem to correlate equally with the general positive or negative attractiveness of a site, but user expectations and context influence longer-term responses. Users expect the aesthetic model to complement the purposes of the site, for example, and thus text-rich content sites seem to benefit more from the classical aesthetic approach. Carefully built designs contribute strongly to the sense that a site is trustworthy and credible, no matter which aesthetic model they use.

Using data in the context of design (and not vice versa)

The visual aesthetics that frame and define content are much more than simply a “skin” that we can apply or discard without consequence. Users react in fast, profound, and lasting ways to the aesthetics of what they see and use, and research shows that the sophisticated visual content presentation influences user perceptions of usability, trust, and confidence in the web content they view. Those user judgments begin within 50 milliseconds of seeing the first page of your site.

Smart graphic design is always some balance of current expressive trends, information architecture, classical layout aesthetics, and detailed research on user preferences and motivations. You should never ignore solid user experience data, but mountains of data won’t auto-magically build you a successful site. Design is a synthetic activity. It can be informed by the results of analysis, but the tools of analysis don’t create beautiful designs. Douglas Bowman’s recent experience as a graphic designer among Google’s engineers who want “data” behind every aesthetic decision is instructive: data and user feedback are always important components of good design, but they are never its sole ingredients.