Using users’ physiological responses for the estimation of websites’ aesthetic judgments

The aesthetic appearance of websites can influence the perception of their usability, reliability, and trustworthiness. Several studies investigated the relationship between single aesthetic features and explicit aesthetic judgments, demonstrating the existence of an attribution bias. However, only a limited amount of studies focused on the interaction between multiple visual properties and have considered not only explicit ratings, but also implicit judgments. In this work, we employ a novel approach, based on the analysis of physiological signals (implicit measures) and the application of machine learning and neural network models to predict users’ perceived aesthetic pleasure from the empirical analysis of web pages’ advanced visual properties (e.g. symmetry, visual complexity, colorfulness, ratio between visual and textual areas). Young adults (N=59, 33 females, Mean age = 21.52 years) assessed the aesthetic appeal of websites and emotional pictures while their physiological activity was recorded. Results using recursive partitioning and generalized linear models demonstrate the possibility of predicting the average aesthetic rating of a website using both explicit (behavioral ratings) and implicit measures(physiological activities).