It’s something we all notice but probably never think much about. When something in our surroundings changes, we take note. In fact, our safety depends on us noticing changes, such as when a green light changes to red, or when a warning sign is posted to keep us from walking into an open ditch.
Scientists at Queen Mary University of London have been studying how people perceive changes and what causes “change blindness”, which is a well documented situation when people fail to notice changes that seem like they should be obvious. In one of the first studies to use computer intelligence to help study human intelligence, researchers in London used computer-based models to help them predict what kinds of changes people are most likely to notice. The researchers developed and tested mathematical systems that are based on human biology. They hope these tools will help in the development of computer vision systems for robots and other applications that will allow the robots to notice interesting changes in their environments.
During the study, research subjects were asked to compare different versions of a series of pictures. Some pictures had elements added or removed. Others had colors changed or added. The changes were based on what the researchers believed were attention-grabbing properties of the items in the pictures. This study broke new ground by using computer programs to decide what to change in each version of the pictures. In past studies, humans had made the decisions on what to change. The Queen Mary University group believed removing the human factor in making the decisions also removed a human bias that could alter the results of the study.
As researchers expected, the computer model was able to predict when “change blindness” would occur. Because color plays such an important role in everyday life, researchers believed changes in color would be the easiest changes for people to notice. So they were surprised to learn that people are actually more likely to notice when an object is added or taken away from the surroundings, rather than when the color is changed.