Pleasure and displeasure are qualities that are imposed onto sensory experiences. This hedonic value of sensations motivates human behavior – it guides us towards a given stimulus or informs us that it is better to avoid it. The influence of hedonic processing on human decision-making has its basis in two distinct components – “liking” and “wanting” which determine the attractiveness and motivational properties of the stimulus. Within the context of visual processing, the hedonic value of an image or object is dictated by the elicited aesthetic experience.
Visual aesthetic experience is a sum of perceptual, cognitive, and emotional processes. First, perceptual processing is based on intrinsic, formal properties of visual objects. Physical attributes (luminance, colors, contrast, lines edges, texture, shading, image complexity) are detected and analyzed by the human visual system. This low-level processing occurs in a rapid, automatic manner and it is modulated by the attentional mechanisms. These perceptual mechanisms of beauty appear to be universal among humans and relatively stable over time.
Cognitive processing is based on contextual information, such as depicted content, its meaning, cultural and individual circumstances. This mode of processing of aesthetic experience is generally slower than the perceptual one and varies between people. While the influence of contextual information on the overall experience of visual pleasure can be minimized, sensual processing – based on perceptual mechanisms – is indispensable for the induction of the aesthetic experience.
Eventually, perceptual and contextual information are merged leading to the final interpretation of experience and – often – inducing emotional response. Even though emotions induced by the image may modulate an overall aesthetic experience, they are neither a necessary nor sufficient prerequisite to generate it.
The neural structures underlying visual aesthetic experience are associated with the brain reward system containing pleasure centers. They mediate the “liking” and “wanting” sensations – the same ones that make us crave a piece of chocolate or physical touch of our loved ones. The hedonic value of visual experiences is believed to be – to some extent – an inherent property of the stimulus, which may be universally recognized as pleasant. These general mechanisms of hedonic processing are applicable to a broad variety of visual object categories, classes and contexts. Most people execute multiple unconscious aesthetic evaluations every single day – we make judgements about window display designs we pass by on our way to work, advertisements we see on the streets, products presented in an online clothing store, and thumbnails we are exposed to whenever we access our Netflix account.
The “hedonic score” of an image is a new, trans-category parameter that Kellify identifies in order to help brands drive users towards their products and services. As it is computed based on intrinsic, formal properties of the image, it determines universal sensations elicited in the viewer. These compositional rules for creating sensory-pleasing, captivating images seem to be beyond the reach of human cognition making it hardly possible to precisely define what renders an image stimulating. For that very reason, Artificial Intelligence capacities come to our help. Deep Learning algorithms learn to recognize and extract the critical features and patterns of images scoring highly on the scale of hedonic value, resulting in the development of a highly scalable approach that can be applied to various scenarios across different categories of industries.