Purchase this article with an account.
Daniel Barratt; Investigating the link between visual attention and emotion: A TVA-based computational approach. Journal of Vision 2009;9(8):199. doi: https://doi.org/10.1167/9.8.199.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Although visual attention and emotion appear to be radically different processes, they are closely related in an adaptive, functional sense. The visual world presents a vast amount of information (over ten million bits) in any given second. However, human visual resources (processing and storage capacities) are limited and therefore must be directed to the most significant parts of the visual field. An important measure of significance is emotional relevance: i.e., relevance to the observer's goals and concerns.
This paper will investigate the link between visual attention and emotion by presenting a computational model based on the results from a range of empirical studies. The computational framework is provided by the Theory of Visual Attention (Bundesen, 1990; Bundesen, Habekost, and Kyllingsbæk, 2005). TVA proposes that visual attention comprises two waves of processing. First, each object in the visual field is assessed for “pertinence” values and given a corresponding attentional weight. Second, each object is categorised according to certain “bias” values; those objects/categories which have the greatest weight and bias are most likely to win the race for entry into visual short-term memory. In order to link visual attention to emotion, both schematic and photographic faces were used as emotional stimuli. The first experimental paradigms used accuracy as the main response measure (identification of one or more letter targets with face distractors), whereas the second paradigms measured reaction time (variations of the Eriksen and Eriksen flanker task). In TVA terms, the results suggest that negative emotional stimuli have high pertinence values and attentional weights, in comparison with either positive or neutral stimuli. However, significant effects are only observable if the number of stimuli exceeds VSTM capacity of around four items (accuracy paradigms), or if targets and distractors come from the same stimulus class and bias values are set constant (RT paradigms).
This PDF is available to Subscribers Only