Making better UX decisions: Reflections on ‘Thinking, fast and slow’ in a UX context
Your mind is lazy. Now, please, don’t be offended. So is mine. It has nothing to do with our personalities, but how the human mind is constructed. I’ve been reading Thinking, fast and slow by Daniel Kahneman and considering how it relates to the practice of user experience.
Kahneman begins by introducing System I and System II thinking. System I processes things quickly and efficiently using what is essentially a heuristic based on available data and programming. When it doesn’t have enough data to recognise and provide an answer, it passes it off to System II for slower, more labour intensive processing. But because our brains are lazy, sometimes System I will just answer an easier question, rather than passing it to System II. And we are particularly bad with statistical information because it’s often easier to think of examples rather than the related statistics.
Consider how we get the governments we do. In school, it might go something like “X has no experience in student government and has promised beer in the fountains, but will they represent me well? (and is beer even possible?)” When System I doesn’t have enough information, it passes the question off to System II to provide an answer. In this example, the question is whether we will be well represented; however, System I may take a shortcut. It might change the question from “Is X a good candidate?” or “Do I agree with their policies?” to answer the question: “Beer? Do I like beer?”
If you think this student government example doesn’t apply, when one party in Ontario promised “buck ($1) a beer” as one of their flagship policies — all the newspapers reported it — you can guess which party won that election.
Election voting is a big complex question that System I can’t effectively answer on its own. And politics rarely works on the people having all the information.
How this relates to UX professional practice
With that in mind, we need to consider how this affects our practice of user experience. If you’ve ever caught yourself or someone else saying, “If I were the user, …” you’ve likely been caught in a situation where System I is answering the wrong question, rather than going for the more labour-intensive System II analysis. “What would I do if I were the user?” is a question for which System I can easily access an answer. It’s easier for System I to answer that question than to pass the question to System II, where it would answer the right question: “What has our research shown about how our users interact with X?”
You may find yourself thinking that you have been doing UX for a long time, surely your System I brain knows more, so it doesn’t take shortcuts. Kahneman says that we can retrain our brain with different beliefs and that this is how an engineer might recognise that a bridge needs some tests to ensure it won’t fall down, or how your doctor might recognise symptoms as being something other than what you found in your Google search. They have trained System I with additional information that helps them make decisions faster. (I once heard someone muse that a Google search only tells you which medical conditions have the best SEO.)
Your experience might have retrained System I to recognise that you have done some research on a specific type of item. As it has access to that information, System I would then also draw on that information to help make a decision. (This likely feeds into our dislike of experts. We are comparing the decisions of our System I brain and assuming that our answer is perfectly adequate. But this is compared with their System I that has been trained with additional information by their System II.)
How do we combat System I making bad decisions?
Kahneman says that although System I sometimes makes bad decisions, for the most part, it generally does a pretty good job. We would drive ourselves crazy if every decision we made had to be passed by System II. And it would be incredibly inefficient. It is a good thing that System I allows that an experienced driver is able to carry on a conversation while driving down a relatively empty motorway. And it’s the same reason that a car passenger often becomes silent when on that same drive comes upon a tailback that requires the driver to pay attention to the task at hand.
There are, I think, two things (so far) that can help us: I) Learn to recognise what questions System I is answering and II) Gain more data to retrain System I to make better decisions.
Learn to recognise the questions System I is actually answering.
When you find yourself making or defending your decisions check in on what questions you or others around you might be answering using System I. Identifying questions like “How would I like it to be if I were the user?” is good to catch quickly. And you may find you have others. It’s also good to clarify what question the other person might be answering. If someone else were making the same sort of statement and it’s something you’d question, that might be a good indicator that you want to question what System I responded with and be sure that the correct question is being answered. And then find ways to redirect the decisions of others so that they are also answering good questions with all the information they need to make a good decision.
Gain to gain more data
An easy way we can gain more data or insights into our users is to participate in UX research. If you don’t have UX researchers on your team, learn to do your own research. Nielsen Norman Group (not paid promotion) has some great courses and articles on how to conduct user research. They also have an extensive collection of reports based on general UX research, that can also help if you are unable to do your own research. If you have researchers on your team, ask to be an observer, so that not only do you have their conclusions, and your own insights into your users as well.
And collect that research. If you don’t already have a research repository that contains your user research, start one. If you do, visit it regularly to look for insights. The more data you have, the more likely System I can be programmed to recognise the patterns that will aid in making good fast decisions.
Understanding how we make decisions is an important first step in making better decisions. I’d highly recommend giving Thinking, fast and slow a read and taking some time to let the content sink in. By examining how we ask and answer questions that help us make decisions, we can improve our practice of UX, which, in turn, will improve the products we design.
Kahneman, D. (2011). Thinking, fast and slow. New York, NY, US: Farrar, Straus and Giroux.