How To Break the Information Age Trance of 'Continuous Partial Attention'
Computer scientist Andreas Bulling wants to save attention spans from technology.
Hand wringing about the "distraction economy" and the general decline in humans' attention-giving abilities is by now a cliche. We get it: user interfaces are everywhere, competing for just a few seconds of our braintime to probably sell us something, and this is bad. We think better when we can actually think about one thing.
We're not winning the attention war it would seem, but we also haven't lost it. In a piece published in the January edition of Computer, Andreas Bulling, a human-computer interface (HCI) researcher at the Max Planck Institute for Informatics, argues that winning the sustained attention of technology users (everyone) is perhaps the most pressing and difficult challenge in the entire HCI field. Whoever finally cracks the attention problem will likely wind up a very rich person.
By 2020, there will be approximately 9.7 billion displays on Earth serving up over 5,000 ad-views per day per urban-dwelling person, a 25 percent increase from 2015. This figure doesn't even include displays found on appliances or in cars. The situation, according to Bulling and other HCI researchers, is one of "continuous partial attention."
"Sustained attention is increasingly being replaced by continuous partial attention: the act of paying simultaneous attention to multiple sources of information but only at a superficial level," Bulling writes. "In economics, this well-known phenomenon has led to the 'attention economy' theory that acknowledges both the scarcity and superficiality of consumer attention and, consequently, the importance of managing it."
Can it even be done? Can we have attention in a warzone of distraction? It's at least worth researching, he argues.
"Given that users' sustained attention can be interrupted during all explicit and implicit interactions with computing systems," Bulling says, "the HCI community should strive to develop computational methods to estimate and analyze the visual attention of a potentially large number of users, unobtrusively and continuously over long periods of time in their everyday life, as well as user interfaces that leverage attention information."
"Sustained attention is increasingly being replaced by continuous partial attention."
As Bulling explains, attentive user interfaces currently span a spectrum from very subtle to obtrusive. The prior demands relatively little mental effort while the latter requires much more; less information throughput (the amount of information that can be processed by a user) vs. more information throughput.
The "pervasive attentive user interfaces" imagined by Bulling might utilize what he calls "attention accounts." This is basically a running balance of available user attention. If I were to focus for a few minutes on this chat box in the next tab over, the result would be an attention withdrawal. The system would then know that I now have less attention—defined here as the act of concentrating on discrete units of information while ignoring other information—to give and so would adjust itself accordingly.
"Instead of interrupting the user whenever new information becomes available," he explains, "future interfaces could trade information importance with users' current interruptibility level and time the delivery of information appropriately—for example for a period of low cognitive load, free attentional capacity, or even boredom."
These interfaces should featuring six crucial qualities, according to Bulling: accuracy, unobtrusiveness, scalability, durability, seamlessness, and context awareness. A contextually aware interface might, for example, combine eye tracking technology with GPS positioning and inertial awareness sensors.
Some serious, fundamental challenges remain before the attention problem can be solved. Attention is still difficult to measure and model.
"[A] key challenge is to contextualize attentive behavior by considering users' current overall situation, activities, and goals," Bulling says. "This requires extending existing user models with models of the environment, available systems, and interactions among multiple users. Although attention is a core aspect of user modeling, it has yet to be examined in unconstrained everyday settings."
So, this is why we need to get to work researching and developing new computational methods for analyzing and estimating user attention. It won't be easy, but whoever solves the problem will have solved a seemingly intractable modern ailment.