Discovering the mechanics of artificial and real meat

Artificial meat is an eco-friendly alternative to real meat that is marketed to have a similar taste and feel. The mechanical properties of artificial meat significantly influence our perception of taste, but how precisely the mechanics of artificial meat compare to real meat remains insufficiently understood. Here we perform mechanical tension, compression, and shear tests on isotropic artificial meat (Tofurky® Plant-Based Deli Slices), anisotropic artificial meat (Daring™ Chick'n Pieces) and anisotropic real meat (chicken) and analyze the data using constitutive neural networks and automated model discovery. Our study shows that, when deformed by 10%, artificial and real chicken display similar maximum stresses of 21.0 kPa and 21.8 kPa in tension, -7.2 kPa and -16.4 kPa in compression, and 2.4 kPa and 0.9 kPa in shear, while the maximum stresses for tofurky were 28.5 kPa, -38.3 kP, and 5.5 kPa. To discover the mechanics that best explain these data, we consulted two constitutive neural networks of Ogden and Valanis–Landel type. Both networks robustly discover models and parameters to explain the complex nonlinear behavior of artificial and real meat for individual tension, compression, and shear tests, and for all three tests combined. When constrained to the classical neo Hooke, Blatz Ko, and Mooney Rivlin models, both networks discover shear moduli of 94.4 kPa for tofurky, 35.7 kPa for artificial chick'n, and 21.4 kPa for real chicken. Our results suggests that artificial chicken succeeds in reproducing the mechanical properties of real chicken across all loading modes, while tofurky does not, and is about three times stiffer. Strikingly, all three meat products display shear softening and their resistance to shear is about an order of magnitude lower than their resistance to tension and compression. We anticipate our study to inspire more quantitative, mechanistic comparisons of artificial and real meat. Our automated-model-discovery based approach has the potential to inform the design of more authentic meat substitutes with an improved perception of taste, with the ultimate goal to reduce environmental impact, improve animal welfare, and mitigate climate change, while still offering the familiar taste and texture of traditional meat. Our source code, data, and examples are available at https://github.com/LivingMatterLab/CANNs.