If there is one take-home lesson from perception research, it is that humans wildly distort and simplify sensory input when attending to the world around them. In the past few years, research into change blindness has shown some stunning failures of our ability to identify changes in our surroundings - the principle of perceptual constancy, while generally valuable, sometimes prevents us from seeing changes that occur. A recent article in the New York Times gives an overview of the phenomenon, but I am curious how it might actually enhance user interfaces.
The risks of change blindness are obvious - if a system supervisor is blind to the appearance of a warning sign/light/sound, the consequences could be severe. It is important to be aware of the potential for this failure to occur and therefore design warning systems that invade awareness as aggressively as possible, but that is a simplistic and intuitive application, one that hardly needs reiterating in contemporary design. The more difficult, but potentially more valuable question is how this ‘blindness’ can actually enhance usability. Essentially, perceived change is information, and the goal of communicative design is to present appropriate information effectively. What does it mean, then, to utilize unperceived change to enhance usability? I have a few ideas, but I expect there are many others that could really enhance this undeveloped aspect of interface design.
One way to utilize change blindness might be when updating GUIs to include new functions in software. Although many people hope for radical enhancements to visual layout and function with each new version of software, once the visual interface has been tweaked to be highly usable (or it conforms to broad user interface guidelines such as Apple’s), it is often most effective to avoid changes so that users can continue to utilize the efficiency of habitual interaction with an interface. When new features are added, however, they may need space on the graphical interface - the visual presentation must change. The principles of change blindness suggest how to add visual elements to an interface without interfering with the previous version’s functionality. Users will be blind to the change when interacting “normally” with the program, but become aware of the new feature when they actively search for it. It is a way to add functionality gracefully. The radical changes Microsoft made when putting together the Office “Fluent” interface exemplifies the risks of breaking with convention, and the resulting loss of productivity. Hopefully, the short-term reduction in usability will be overcome by adaptation to new conventions 
, although the remaining visual complexity will continue to make almost any future addition difficult to find.
Although change blindness research has focused on visual perception, what about perception of function? I would be interested to see how ‘blind’ people are to changes in the way a thing operates - which extends into being aware of when a thing begins to fail. For complex systems, it is usually assumed that expert users are most effective at identifying potential problems, but is this related to the claim that novices are most susceptible to change blindness? If we can adjust visual interfaces to reduce change blindness, can we use similar principles to adjust the function of systems to reduce change blindness? For example, can car engines be developed in such a way that the early signs of failure - slight changes in normal vibration, particular smells, and other behaviors - are more salient and meaningful? This suggests different solutions than just creating a new sensor connected to a warning light on the dashboard. Acoustic and tactile feedback are key parts of the equation that until recently have been actively diminished in the name of ‘comfort’.
Change blindness, although superficially a failure of our perceptual systems, may in fact make some tasks easier - it is the result of selective attention, which may not always be helpful, but ultimately enhances our ability to maintain focus in the presence of distraction without any additional cognitive effort.