Our findings are well predicted by the #LAMINART model (Francis et al. 2017), which employs #recurrent segmentation processes unfolding over time to separate objects into distinct representation layers 🧵7/9