Predicting Image Quality and Aesthetics from EEG

When we look at a photograph, our brain simultaneously processes its technical quality (exposure, contrast, sharpness, noise) and aesthetic appeal (composition, mood, beauty). Image quality assessment (IQA) and image aesthetic assessment (IAA) are computational equivalents of these two judgements. This project will investigate whether neural responses to images encode both these perceptual dimensions and whether they share common neural mechanisms. Previous work suggests that IQA and IAA rely on overlapping perceptual features but differ in their level of abstraction and cognitive processing, with technical quality being more closely related to low-level visual factors and aesthetics to higher-level interpretation.

Questions we aim to answer:

1. Do EEG responses correlate with image quality and aesthetic scores?
2. Can we build models that predict IQA or IAA from EEG features?
3. Do images with low IQA and IAA scores evoke convergent neural signatures in early EEG components (P1, N1) but divergent signatures in later components (P300, LPP), reflecting shared perceptual mechanisms but different higher-order evaluation? (advanced)

During the hackathon, we will calculate image quality and aesthetic metrics for visual stimuli used in the EEG dataset. We will then explore how these scores relate to EEG features by analysing correlations, creating exploratory predictive models, and visualising neural signatures associated with different levels of technical quality and aesthetic appeal. A secondary objective that relies on the results of the previous two is exploratory validation of the “early vs late processing” hypothesis: whether technical quality aligns more strongly with early sensory components and aesthetics with later cognitive components.