We only perceive, consciously, what we pay attention to. If we are focused on a particular task, we often don’t notice unexpected events – even if they occur right next to the object of our attention.
In their now famous ‘gorilla’ experiment, Daniel Simons and Christopher Chabris of Harvard asked 192 people to watch a video of three people passing a basketball back and forth. Each observer’s task was to count how many passes took place. In the middle of the test, a woman in a gorilla suit walked between the players passing the basketball, stood there for five seconds, and then walked out of the frame.
The researchers then blandly asked the observers if they had noticed anything unusual. Nearly 1/3 of the observers failed to see the gorilla. Period. Didn’t happen. Didn’t exist – at least to their minds. As you can see from a snapshot of the video, the gorilla would seem to be hard to miss.
But one-third of the observers, keenly focused on the task of counting passes, completely ‘missed’ the gorilla – even though the image of the gorilla must have been registered by their eyes. Their brains, however, filtered out that information as they sought to focus on counting the passes.
So what does this have to do with interactive software development? Well, in our products, we often provide feedback in response to actions people have taken. If they are focused on a particular task, the gorilla experiment shows that some people may miss our feedback because of their task focus.
We wrestled with the phenomenon of ‘change blindness’ when designing an interactive balance training game for children with cerebral palsy. The study was a collaboration between Red Hill Studios and the School of Nursing at UCSF under a small business research grant from the National Institutes of Health .
The children were asked to maneuver a virtual paper airplane (in red) through a game course of target planets by shifting their weight on a Nintendo WiiFit board.
Because of the impact of the cerebral palsy, many of these children have difficulty shifting their weight in the proper way – gently moving their hips to one side or the other while standing straight. Instead, they attempt other approaches such as bending over, bending their knees and swinging their arms. Such ‘compensatory movements’ are common in neuromuscular disorders such as cerebral palsy and some strokes. One of our goals was to reduce compensatory movements because they did not help the child improve his or her balance.
We asked the children to hold a Wiimote in the middle of their chests so we could tell when they bent over. Once we had this information, however, we were faced with the ‘gorilla’ problem. Because the children were so focused on shifting their weight and guiding the airplane, would they even notice visual feedback telling them to stand-up straighter?
One of the ways to increase the chance that feedback will be perceived is to connect the feedback to the object of the person’s focus. In this case, the children were focused on the virtual airplane as they guided it through the rings with their movements.
Our feedback approach was actually quite simple: when the child was shifting his/her weight correctly, the airplane was colored green. When they leaned over in a compensatory movement, the airplane would turn red. We also set up a ‘prototype control’ that allowed us to turn the feedback system on and off between different levels of the game.
But would it work? Would the children notice the airplane changing color or would they be so distracted by the difficult task of shifting their weight that our feedback would be filtered out?
The table below shows the experience of one of our subjects, which was similar to many of the other responses.
This subject clearly perceived the feedback and reacted positively. He displayed compensatory movements when the feedback was turned ‘off’. These compensatory movements nearly disappeared when the feedback was turned ‘on’ in Levels 3 and 4. When we turned the feedback back off, he returned to compensatory movements (levels 5 & 6). When we turned the feedback back on again, compensatory movement disappeared once more (levels #7 & 8).
Lessons learned? Two of them.
Lesson #1 – When delivering feedback to people using your software, think about where their attention is focused. Then, incorporate your feedback into the object of their attention. Otherwise, your ‘crucial’ information could be ignored just like the gorilla was.
Lesson #2 – When testing prototypes of a new kind of interface or a complex interactive program, it helps to construct alternate versions that you can easily switch between during testing. This can allow you to collect valuable, quantitative comparisons between different design approaches. We’ll explore this topic further in an upcoming “Co-development of Interactive Software” post. Stay tuned!
For additional information about our research collaboration with the UCSF School of Nursing, please see our research paper published in the Games for Health Journal.
D J Simons, C F Chabris, Perception, 1999, volume 28, pages 1059 – 1074.
Feasibility of Computer-Based Videogame Therapy for Children with Cerebral Palsy
Sandra Radtka, Robert Hone, Charles Brown, Judy Mastick, Marsha E. Melnick, Glenna A. Dowling
Games for Health Journal. August 2013, 2(4): 222-228.