Visual liking as sensory valuation

How the brain computes the hedonic value of sensory objects remains a foundational question in neuroscience. In the visual domain, this question has often been approached through the search for stimulus properties that reliably predict liking, such as symmetry, curvature, order, complexity, and color. Such work has shown that perceptual features contribute to visual preference. Recent evidence, however, strongly suggests that hedonic value cannot be explained by feature encoding alone. Thus, behavioral studies have revealed substantial individual differences in feature sensitivity, effects of expertise and expectations, and strong modulation by semantic knowledge, perceptual context, and task conditions. Neuroimaging studies likewise show that activity tracking objective stimulus properties in early visual cortex does not reliably predict hedonic judgments, which depend more closely on subjective and higher-level representations of those properties. We argue that these findings mark an important shift in the study of visual liking: from asking which features people prefer to asking how the brain computes hedonic value from sensory input. We explain this shift from the perspective of sensory valuation, which conceptualizes visual liking as a flexible computational process that integrates perceptual representations with semantic knowledge, task models, internal state, and reward-related valuation mechanisms. On this view, visual liking is not a direct readout of stimulus features, but the computed value of an interpreted sensory object in a specific organismic and situational context.