Прямое сравнение изображений: границы применимости «дискретной» и «многомерной» моделей восприятия эмоциональных экспрессий

Abstract

Studying perception of emotional expressions suggest, supposed to analyze effectiveness of discrimination between images or the severity of specific emotional experiences. Neither paradigm allows us to fully explicate the observer’s subjective ideas about the relationships between different facial expressions, and to reconstruct the whole subjective space of perceived emotional expressions. In the experiment, a direct assessment was made of the similarity of images of strongly and weakly expressed “basic” emotional ex- pressions. In addition to actually evaluating the similarities, the study participants in free form gave the rationale for their decision. It was found that when performing direct comparison of images of emotional expressions presented for an unlimited time, observers mainly focus on assessing the emotional state in terms of “basic” emotions or equivalent to them. As a rule, strongly expressed emotional expressions are evaluated as “completely dissimilar” to each other. However, in the case of similar semantics of the expressed emotions, the expressions are evaluated as more similar, despite the fact that the observer distinguishes them well enough. Reconstruction of multidimensional space according to pairwise comparison indicates that the “basic” emotional expressions are not independent from each other constructs. The relationships between them correspond to the semantic space described by the Core Affect model. Thus, explicitly the participants in the experiment in explaining the degree of similarity describe the compared images in terms of “discrete” emotions; at the same time, the implicit structure of similarity corresponds to a two-dimensional semantic space with dimensions “pleasant — unpleasant” and “activation — deactivation”.