Visual recognition seems to happen easily and automatically — yet if we see a small, transparent truncated conical object how do we know that it might be something to use when consuming liquid? For holding flowers? Recognising objects, animals, and people is a challenging and complex process that is considered to be one of the “hard problems” of computational science. Such visual semantic abilities can break down highly specific ways following brain injury. Moreover they can break down in people who still understand what something is when it is called by its name or when they hear the characteristic sound it makes. My talk will describe some patients whose problems in visual recognition may help us to understand the way that the brain solves the challenge of understanding the world.