Context straightens representational trajectories, but only in continual prediction

Hosseini et al. study Gemma-2-27B residual-stream trajectories across four task families using trajectory curvature (the angle between consecutive per-token transition vectors), participation-ratio effective dimensionality, and PC1/PC2 elongation [hosseini-etal-2026-context-structure-geometry] Trajectories straighten (curvature falls) as context grows in continual-prediction settings, in natural language and grid-world tasks (long- vs short-context t=-7.46, d=0.75; latent grid t=-11.62, d=1.16) [hosseini-etal-2026-context-structure-geometry] Straightening is inconsistent or absent in structured-prediction tasks: few-shot learning straightens only in the transition phase (F=152.7) and not during answer generation (F=0.67, p=0.72) [hosseini-etal-2026-context-structure-geometry] Straightening correlates with behavioral output-logit gaps in grid-world tasks (r=0.99), but the paper performs no causal intervention [hosseini-etal-2026-context-structure-geometry]