Speakers' use of agency and visual context in spatial descriptions

When speakers describe the world, they typically do so from their own perspective. However, they are able to adopt a different perspective, and sometimes do so even when they are not communicating with someone who has a different perspective from their own. In three experiments, we investigated the factors that might lead speakers to adopt a non-self-perspective. Participants described how objects were located in scenes that contained no other entity, a person, or a symmetrical plant. They were more likely to adopt a non-self-perspective (e.g., using to the left to refer to an object on their own right) if the scene contained a person facing them than a person facing away or a plant, and if it contained a person who could see (and potentially act on) the object than a person who could not, even when that person showed no intention to act. Our results suggest that speakers can put themselves in the shoes of another potential agent and use a simulation of that agent's perspective as the basis for formulating their descriptions.