Fabian Jogl co-authored the paper “Graph Representational Learning: When Does More Expressivity Hurt Generalization?”, accepted at the 14th International Conference on Learning Representations (ICLR 2026) in São Paulo. The work introduces a family of premetrics capturing structural similarity between graphs and derives generalization bounds depending on training-test graph distance, model complexity, and training set size. The results show that more expressive GNNs may generalize worse unless complexity is balanced by sufficient data, offering theoretical insights supported by empirical results.