The specificity of emotion inferences in text comprehension : the role of top-down and bottom-up processes
Thèse de doctorat : Université de Fribourg, 2013.
The aim of this thesis was to investigate the influence of individual differences and the nature of the information presented to readers on emotion inferences in text comprehension, in particular on the specificity of these emotion inferences. Research on emotion inferences has led some researchers to suggest that emotion inferences made during reading are specific (e.g., Gernsbacher et al.,... PlusAjouter à la liste personnelle
- The aim of this thesis was to investigate the influence of individual differences and the nature of the information presented to readers on emotion inferences in text comprehension, in particular on the specificity of these emotion inferences. Research on emotion inferences has led some researchers to suggest that emotion inferences made during reading are specific (e.g., Gernsbacher et al., 1992) whereas others (e.g., Gygax et al., 2007) proposed that readers do not need to elaborate a complex representation of emotions but preferentially infer some components of emotion (like behavioral information), stereotypical of emotion responses. The present thesis investigated the conditions under which readers may go further than the component level and reach complex emotion representations. Results presented in Chapter 3 demonstrate that top-down processes related to readers’ individual differences (i.e., empathy and working memory) or to reading strategies (i.e., simulation, elaboration time) do not fully explain the under-specificity of emotion inferences found in previous research (e.g., Gygax et al., 2003, 2004; Molinari et al., 2009). However, bottom-up processes, examined in the second part of the thesis, in terms of the relevancy of emotion information (i.e., emotion components) conveyed in the narratives, better account for the specificity of emotion inferences. To investigate this issue, an optimal congruent vs. moderate congruent paradigm (as opposed to the habitual match vs. mismatch paradigm) was developed. In this new paradigm, the emotional content of the narratives was manipulated based on the number of emotion components and on their typicality regarding the intended emotion. Chapter 5 presents three experiments suggesting that when the narratives convey highly typical emotion information, readers are very likely to integrate specific emotion inferences into their mental models of the text. When the narratives convey less typical emotion information (but still matching the intended emotion), readers may only map the incoming emotion information onto their representations of the texts. Although highly critical as regard to the materials used in previous studies on the matter (e.g., Gygax and colleagues, 2003, 2004 and 2007), this thesis brings further support for Gygax et al.’s (2007) claim that emotion inferences are elaborated in a constructive manner. Chapter 5’s experiments actually showed that emotion inferences are based on emotion components (as defined in the emotion literature), which are activated and integrated incrementally in readers’ mental representations. Importantly this thesis shows that a shift of paradigm and an interdisciplinary approach were needed to further our understanding of the processes underlying emotion inferences when comprehending text.