Situational Ambiguity and Ageing : Navigating the ambiguous world of underspecified situations in current and later life.

Abstract

Ambiguity detection, understanding if a text has more than one valid interpretation, in mental models, is not well understood. We developed a novel paradigm for investigating if, how, and when readers of texts that describe a situation detect ambiguity. We also investigated how texts which describe spatial situations are processed differently from texts that describe non-spatial situations and how ageing affects processing. My doctoral thesis presents a novel paradigm for investigating situational models, from paragraphs, which better resemble “real life”, by embedding premises into paragraphs. We found that ambiguity is hard for readers to detect but suggest that encouraging, or discouraging certain reasoning strategies we improved rates of ambiguity detection. We speculate that using certain reasoning strategies allows participants to create a more complete model from the premises which is more conducive to ambiguity detection. We did this by changing how the reasoner’s understanding of a model is probed during questioning. Specifically, we found that probing the “internal” part of the model significantly increased ambiguity detection. In line with prior research (e.g., Light, 1988; Radvansky et al.,1990) we found that that despite slower reading (Myerson et al., 1990) and processing (Copeland & Radvansky, 2007), older participants are just as capable as younger readers at creating mental models and detecting ambiguity. We suggest that older participants may use more holistic reasoning strategies, e.g.encoding the entire model rather than the two premise parts. Which has been suggested previously as a compensatory mechanism for lower working memory capacity (Copeland & Radvansky, 2007). Our findings lead us to conclude that situational ambiguity is harder to detect for both older and younger people and that this is highly dependent on strategies, as probing different different parts of the model yield different rates of ambiguity detection.