Working memory components in survey and route spatial text processing

Abstract

Imagine you are a tourist walking around an unfamiliarcity, and you read in your guidebook that somewhere in thecity there is a famous museum of modern art. How do youget there? You stop pedestrians and ask them. Answersmay be different. Someone might say: ‘‘You need to takethe subway, because it’s a long way from here. Go straighton, and then turn right just before the church. Go on for awhile and you’ll see the red subway sign in front ofyou….’’However, someone else might tell you somethingdifferent: ‘‘We are on the south side of the city and themuseum is in the north-east. You need to take the greenline of the subway going north, and then the yellow linegoing east….’’ In both cases, you will be able to construct amental representation of the environment and, hopefully,you will reach the museum. Indeed, giving and asking forspatial information is a common experience in everydaylife.Specialized literature widely acknowledges that theprocessing of spatial descriptions leads to the constructionof mental models with spatial properties isomorphic tothose of the environment described (Bower and Morrow1990; Mani and Johnson-Laird 1982; Morrow et al. 1989;see also Zwaan and Radvansky 1998 for a review ofresearch on the use of situation models in language com-prehension). However, as pointed out by the examplementioned earlier, spatial descriptions vary widely (Deniset al. 1999) and, in particular, they assume different per-spectives, of which the most studied are route and surveyviews (Taylor and Tversky 1992; Tversky 1991). The routeperspective takes the point of view of a person travelingwithin an environment. It uses an intrinsic frame of refer-ence and egocentric terms, such as right, left, front, andback, and has a linear organization, given by the order oflandmarks along the route itself. Instead, the survey per-spective provides an overview of the spatial layout,sometimes having a hierarchical structure, and uses anextrinsic frame of reference and canonical terms such ascompass directions.An issue still being discussed is whether the descrip-tion’s perspective affects the corresponding mental model.Most studies addressing this issue report results consistentwith an influence of the perspective on the mental repre-sentation. For example, in Bosco et al. (1996), Pazzagliaet al. (1994), and Perrig and Kintsch (1985), participantswere faster and more accurate in verifying inferentialspatial statements that adopted the same perspective of thedescription they had previously read. Instead, Taylor andTversky (1992) found that participants were as fast andaccurate in verifying inferential sentences based on thesame or different perspective adopted in the previouslyread text. The authors assumed that spatial knowledge isincorporated in abstract representations that can be viewedor visualized from several different perspectives.However, more recent studies by the same authors havesupported the notion that perspective does indeed affect thecorresponding mental representation. In Schneider andTaylor (1999), the mental models constructed from surveytexts maintained some of the properties of the text’s per-spective. Lee and Tversky (2001, 2005) obtained longerreading times when participants read spatial passagesadopting a perspective different from that of the rest of the