as the environment becomes safer we manufacture new threats. To study how concepts change when they become less common, we brought volunteers into our laboratory and gave them a simple task — to look at a series of computer-generated faces and decide which ones seem “threatening.” The faces had been carefully designed by researchers to range from very intimidating to very harmless. As we showed people fewer and fewer threatening faces over time, we found that they expanded their definition of “threatening” to include a wider range of faces. When they ran out of threatening faces to find, they started calling faces threatening that they used to call harmless. Rather than being a consistent category, what people considered “threats” depended on how many threats they had seen lately.