Contributed Talk Sessions | Poster Sessions | All Posters | Search Papers

Poster Session B: Wednesday, August 13, 1:00 – 4:00 pm, de Brug & E‑Hall

MoralNet: Visual representations of moral intuitions in artificial and biological neural networks

Tim Lauer¹, Karolina Drożdż, Sarah Marie Müller², Frederic R. Hopp²; ¹ZPID, ²Leibniz Institute for Psychology

Presenter: Tim Lauer

Humans quickly detect moral situations in visual surroundings, suggesting that moral perception is attuned to features of the sensory environment. Yet, few computational models describe how combinations of stimulus features evoke perceptions of different moral situations. Here, we develop a convolutional neural network that decodes images into 10 distinct moral categories. We train and cross-validate the model on socio-moral images and show that image content is sufficient to predict the category of human morality ratings. In a fMRI study, we demonstrate that patterns of human visual cortex activity encode moral category–related model output and can decode multiple categories of moral scenes. Our preliminary results suggest that category-specific visual features can be reliably mapped to distinct moral intuitions, and they are coded in distributed representations within the human visual system.

Topic Area: Reward, Value & Social Decision Making

Extended Abstract: Full Text PDF