Hit a bag of rice hard enough and it gets softer.
This is not how granular materials are supposed to behave. Sand hardens under fast loading. Most studied granular materials do the same or hold steady. In the literature on jammed granular matter, rate-strengthening is the default. You push faster, the material pushes back harder. It's so consistent that the exceptions barely get studied.
Rice is the exception. Mingchao Liu and colleagues at the University of Birmingham published a paper in Matter showing that rice granules exhibit pronounced rate softening: yield stress drops as loading rate increases. The mechanism turns on grain geometry. Rice grains are elongated, with a high aspect ratio compared to the roughly spherical grains of sand or glass beads. When you load them slowly, friction between those oblong surfaces builds up and the network of force chains that carries load through the packing holds firm. But at high strain rates, friction between the grain surfaces drops sharply. The force chains that hold the packing together collapse. The load-bearing architecture fails while the grains themselves stay intact.
Force chains are worth pausing on. Granular materials don't distribute load evenly like a fluid or a solid. Instead, a small fraction of grains bear most of the stress, forming chain-like paths through the packing. The rest of the grains are along for the ride. The strength of a jammed granular material isn't really about the strength of individual grains. It's about the stability of these chains. And chain stability depends on friction between grains. When that friction is rate-dependent, so is everything built on top of it.
The paper could have stopped there. It didn't.
Liu's team took rice (rate-softening) and sand (rate-hardening) and packed them into opposite chambers of a bi-beam structure. Under slow loading, the rice side is stiffer than the sand side. The neutral axis shifts toward rice, and the beam buckles in that direction. Under fast loading, the rice side softens while the sand side hardens. The neutral axis moves the other way. The beam buckles in the opposite direction. Same structure, same materials, opposite behavior. Selected by speed alone.
No sensors. No electronics. No active control. Two granular materials with opposite signs, composed into a switch.
"Rather than treating this phenomenon as curiosity, we turned it into a design principle," Liu said. "Fast loads trigger one behaviour, slow loads another." The team extended this to a dual-unit design that amplifies the rate dependence: contact reinforcement under gentle movement, separation under sudden impact.
Metamaterial design usually works by engineering microstructure. You arrange unit cells in specific geometries to produce properties that the base material doesn't have on its own. Auxetic metamaterials expand laterally when stretched. Acoustic metamaterials block sound at specific frequencies. The design intelligence lives in the geometry.
What Liu's team did is different. The design intelligence lives in the sign. They didn't engineer a novel geometry. They took two ordinary materials whose rate responses point in opposite directions and let the opposition itself become the mechanism. The geometry is a simple bi-beam. All the interesting behavior comes from the fact that rice and sand disagree about what to do when loading speed changes.
This distinction matters. Traditional composites blend properties. An alloy averages the characteristics of its components. A fiber-reinforced polymer combines the tensile strength of the fiber with the formability of the matrix. The design logic is additive. You're looking for materials whose properties sum to something better.
Sign opposition produces a qualitatively different outcome. You don't get a blend. You get a switch. A threshold emerges from the sign mismatch itself, the speed at which one material's softening overtakes the other's hardening. Below that threshold, one behavior. Above it, the opposite. The control parameter provides the switching point for free. No logic circuit required.
Here's what I keep coming back to. Materials science has spent decades searching for materials with the right properties. The right stiffness, the right damping, the right thermal expansion coefficient. The search assumes the property is the unit of design. But what this paper demonstrates is that the sign of a property's rate dependence can be the unit of design instead. You don't find a material that does what you want. You find two materials that disagree about what to do when conditions change, and you let the disagreement do the work.
The principle extends beyond granular matter. Any system where sub-units have opposite-sign sensitivity to a shared control parameter can be composed into a switch. Rate-dependent modulus is one example. Temperature-dependent expansion is another. Voltage-dependent resistance in a transistor is, if you squint, the same pattern: one region conducts better with voltage, another conducts worse, and the composition of those two behaviors at a junction is what makes a switch.
The specific physics changes. The compositional logic doesn't. Two elements with opposite signs, a shared control variable, a threshold. That's a switch.
The smallest surprise in the paper is the one about rice. The largest is that nobody tried this composition before. Rate-hardening and rate-softening materials have been studied for decades, in separate literatures, by separate communities. The idea of putting one next to the other and letting loading speed select the behavior was sitting in plain sight. It just needed someone to stop looking for the right material and start composing the wrong ones.