Representation of spatial descriptions in working memory

Abstract

Two experiments examined subjects’ ability to make drawings of spatial layouts based on verbal descriptions. There were two sorts of descriptions that depended solely on the order of the sentences within them: (1) referentially continuous descriptions that can be immediately integrated into a unified representation, and (2) referentially discontinuous descriptions that cannot, because their second sentence makes no reference to what has gone before and can only be integrated subsequently. The first experiment demonstrated that it was harder to encode the descriptions while carrying out a visuo-spatial tracking task. The second experiment showed that having to remember six digits selected at random also had an adverse effect on performance, but it affected discontinuous descriptions more than continuous descriptions. It produced more errors and longer reading times than either a control condition or a condition in which the subjects had to articulate six digits continuously. The pattern of results suggests that both continuous and discontinuous descriptions rely on visuo-spatial working memory, but discontinuous descriptions tax verbal working memory to a greater extent than do continuous descriptions. It seems that discontinuous descriptions tend to be encoded in superficial linguistic representations once the discontinuity is encountered.