One Focus or Many? Modeling Attentional Distributions During Spatial Term Use

Abstract

Are humans able to split their attentional focus? This master's thesis tries to answer this question by proposing several modifications to the Attentional Vector Sum (AVS) model (Regier & Carlson, 2001). The AVS model is a computational cognitive model of spatial language use that assumes visual attention. Carlson, Regier, Lopez, and Corrigan (2006) have developed a modification to the AVS model that integrates effects of world knowledge (functionality of spatially related objects) into the AVS model. This modified model assumes that people are able to split their visual spatial attention. However, it is debated whether this assumption holds true (e.g., Jans, Peters, & De Weerd, 2010). Thus, this thesis investigates the assumption in the domain of spatial language use by proposing and assessing alternative model modifications that do not assume split attention. Based on available empirical data, the results favor a uni-focal distribution of attention over a multi-focal attentional distribution. At the same time, the results cast doubt on the proper modeling of functional aspects of spatial language use, as the AVS model (not considering functionality) is performing surprisingly well on most data sets. (See https://doi.org/10.1007/978-3-319-11215-2_6 for a condensed version of this work.)