When you stop pushing a material and let it relax, the stress should decrease. This is not just an intuition but a thermodynamic expectation: the system moves toward equilibrium, and equilibrium has lower stress than a loaded state. Release the load, watch the stress fall, wait for stillness.
Owens shows that soft glasses — emulsions, foams, gels, dense colloidal suspensions — do something different. Apply a small strain step to a soft glass and release it. The stress does not fall monotonically. It rises. The material becomes more stressed during relaxation than it was at the moment of release.
The mechanism requires no exotic physics. It requires only that the material be heterogeneous in its local strain states. A soft glass is not a uniform medium. It is a patchwork of regions carrying different magnitudes and directions of residual strain from past deformations. Some of these regions are oriented with the applied strain; others against it. Some relax quickly; others slowly.
When a strain step is applied and released, the fast-relaxing regions let go first. If these regions happened to carry strain that opposed the residual background stress, their relaxation removes a counterbalancing force. What remains — the slow-relaxing regions with their directional bias — now dominates, and the net stress rises. The system becomes more stressed not because something is being added but because something was being subtracted, and it stopped.
This persists even after the material has been stressed beyond its yield point. Past deformations leave directional signatures — anisotropic distributions of local strain states — that survive yielding and flow. The soft glass remembers which way it was pushed, and this memory produces stress during relaxation. The material is not recovering toward a neutral state. It is recovering toward its own history.
The implication is not about glasses in particular. It is about any system that carries distributed, heterogeneous states with different relaxation rates. Remove an active balancing force — an intervention, a policy, a relationship — and the system does not relax to the average of its parts. The parts that relax fastest may have been the ones holding the system together. What remains may be more stressed, more directional, more committed to the residual pattern than the whole system appeared to be when all parts were contributing.
Relaxation reveals what the active state was hiding. In a homogeneous system, relaxation is calming. In a heterogeneous system, relaxation is diagnostic — it exposes which components were load-bearing and which were compensating, by showing you what happens when the compensators let go first.