Metamaterials that learn to change shape

For living organisms from cells to animals, learning to change shape is a fundamental strategy of their adaptation and evolution. By contrast, human-made materials can have rich shape-changing space but crucially lack the ability to learn. Here, we build metamaterials that can learn complex shape-changing responses by using a contrastive learning scheme. By being shown examples of the target shape changes they should achieve, our metamaterials are able to learn those shape changes by progressively updating internal learning degrees of freedom---local stiffnesses. Unlike traditional materials that are designed once and for all, our metamaterials have the ability to forget and relearn new shape changes, to learn multiple shape changes that deviate from non-reciprocity relations, and to learn multistable shape changes. Our findings establish metamaterials as an exciting platform for physical learning and in turn open avenues for the use of physical learning to devise adaptive materials and robots.