Choice Architecture for Human-Computer Interaction

Abstract

People in human–computer interaction have learned a great deal abouthow to persuade and influence users of computing technology. Theyhave much less well-founded knowledge about how to help users choosefor themselves. It's time to correct this imbalance. A first step is toorganize the vast amount of relevant knowledge that has been builtup in psychology and related fields in terms of two comprehensive buteasy-to-remember models: The ASPECT model answers the question“How do people make choices?“ by describing six choice patterns thatchoosers apply alternately or in combination, based on Attributes, Socialinfluence, Policies, Experience, Consequences, and Trial and error.The ARCADE model answers the question “How can we help peoplemake better choices?“ by describing six general high-level strategies forsupporting choice: Access information and experience, Represent thechoice situation, Combine and compute, Advise about processing, Designthe domain, and Evaluate on behalf of the chooser. These strategiescan be implemented with straightforward interaction design, butfor each one there are also specifically relevant technologies. Combiningthese two models, we can understand virtually all existing and possibleapproaches to choice support as the application of one or more of theARCADE strategies to one or more of the ASPECT choice patterns.After introducing the idea of choice architecture for human–computerinteraction and the key ideas of the ASPECT and ARCADEmodels, we discuss each of the Aspect patterns in detail and show howthe high-level ARCADE strategies can be applied to it to yield specifictactics. We then apply the two models in the domains of online communitiesand privacy. Most of our examples concern choices about theuse of computing technology, but the models are equally applicable toeveryday choices made with the help of computing technology.