Emergent Intelligence Under Structural Learning Pressures

We propose that general intelligence emerges from four simultaneous learning pressures:
prediction, spatial consistency, object persistence, and transformation compression. Unlike
current approaches that rely on architectural priors or task-specific engineering, our frame
work explains how cognitive capabilities like segmentation, abstraction, and rule inference
arise naturally from constraint satisfaction. Each pressure alone produces characteristic
failure modes; only their joint enforcement yields intelligent behavior. This provides a prin
cipled foundation for systems that learn structure rather than encode it, with falsifiable
predictions about how manipulating individual pressures affects cognitive emergence.