A Californian study compares animal-based and plant-based meat products using mechanical simulations of chewing, paired with an artificial neural network to identify plant-based textures most similar to their animal counterparts.
Current meat production is failing to meet the needs of a growing global population. According to the “World Population Prospects 2024“ by the UN Department of Economic and Social Affairs, by the 2080s, the global population is expected to reach 10.3 billion, compared to 8.2 billion in 2024. Traditional meat production not only struggles to keep up with demand but also poses significant risks to human health and the environment.
The majority of meat consumed globally comes from intensive farming practices. These operations are responsible for nearly half the carbon emissions of major fossil fuel industries and drive large-scale deforestation – such as in the Amazon rainforest – to create pastures and grow feed crops for billions of animals [source: Greenpeace].
Adding to this, the widespread use of antibiotics in animal farming raises alarm. Globally, «the average annual antimicrobial consumption per kilogram of animal meat is estimated at 172 mg for pigs, 148 mg for chickens, and 45 mg for cattle» [source: “Global trends in antimicrobial use in food animals” – Agricultural Sciences].
This overuse not only compromises animal welfare but also contributes to the rise of antimicrobial resistance, a growing threat to public health.
On the human health front, excessive red meat consumption has been linked to cardiovascular diseases and colorectal cancer. Combined with the environmental impacts, these concerns underscore the urgent need to reduce animal protein in favour of plant-based alternatives, which provide fibre, vitamins, and essential minerals crucial for mental and physical well-being [source: “Meat consumption: Which are the current global risks? A review of recent (2010–2020) evidence” – Food Research International).
TAKEAWAYS
Alternative to animal protein: meat made from vegetables and legumes
Commonly referred to as “fake meat”, these food products mimic the appearance of animal-based meat but are made from a variety of plants (primarily chickpeas, lentils, peas, mushrooms, and soy), sometimes supplemented with grains and spices.
First produced in California around 15 years ago, today “fake meat” is found worldwide, with international brands and local producers using vegetables and legumes to create plant-based burgers, fillets, chicken, sausages, and kebabs.
Several studies have quantified the environmental impact of plant-based meat, which is significantly lower than that of animal-based meat.
A 2023 study led by the German Institute of Food Technologies (“Meat substitutes: Resource demands and environmental footprints” – Resources, Conservation and Recycling) highlights that while animal-based food contributes 57% to global greenhouse gas emissions from food production, plant-based alternatives account for just 29%.
And yet, despite such findings, the global presence of fake meat on our tables remains relatively modest.
«Although more people are curious about flexible diets and intentionally reduce their consumption of animal meat, most do not eliminate it entirely from their diet», note the authors of “Finding flexitarians: Current studies on meat eaters and meat reducers” (Trends in Food Science & Technology).
Data from a US consumer survey, published in “A Survey of Consumer Perceptions of Plant-Based and Clean Meat in the USA, India, and China” (Frontiers Sustainable Food Systems), revealed that – excluding vegetarians and vegans – only one-third of participants were “very” or “extremely likely” to purchase plant-based meat products. Why this resistance?
Plant-based meat and the issue of taste and texture
A more sustainable and healthier food system, one richer in plant-based proteins (though not exclusively reliant on them), is a feasible alternative in all developed countries. However, it does not fully convince consumers to give up their favourite foods – such as meat – nor does it seem to disrupt deeply ingrained eating habits.
Psychological mechanisms aside (the “meat paradox” refers to a cognitive dissonance whereby people love animals yet continue eating them, as the steak on the plate is dissociated from the animal it came from), there is, quite simply, an issue of taste. It is hard to replicate the texture and flavour of animal meat – rich in adipose tissue and therefore more flavourful – in plant-based products like chickpea or tofu-based meat alternatives.
This concept enters the realm of sensory experience with food, a critical factor in product selection. It cannot be ignored, especially when discussing fake meat.
Sensory experience has historically been the focus of many studies on food texture. However, the findings vary greatly, influenced by participants’ personal preferences, cultural backgrounds, individual sensory abilities, and the lack of standardised data in the field.
If we are truly asking plant-based meat to mimic the structural characteristics of animal meat to the extent of offering the same sensory experience, the challenge lies in designing products based on an analysis of the mechanical properties of both types of meat. This requires laboratory comparison and a methodology capable of developing computational models to define the attributes of the “best” plant-based meat and then replicating them outside the lab.
From this premise stems an innovative study conducted by a research group at the Department of Mechanical Engineering at Stanford University. Their findings are detailed in “The mechanical and sensory signature of plant-based and animal meat” (npj Science of Food, November 2024). Let’s take a closer look.
Animal meat vs plant-based meat: comparing mechanical properties
Animal meat, the research team explains, consists of muscle, connective tissue, fat, and water, and has an anisotropic microstructure, meaning its properties vary under different conditions. In contrast, plant-based meat, made from vegetables, legumes, soy, and grains, is generally isotropic, meaning its structure remains uniform, «exhibiting the same properties in all directions».
The comparison focused on processed animal meat products (three) and plant-based products (five), including turkey breast, animal sausage, and hot dogs in the first group; tofu-based turkey, plant-based sausage, plant-based hot dog, extra-firm tofu, and compact tofu in the second group. These were subjected to standardised tension, compression, and shear tests, followed by the development of machine learning models trained with the parameters of each product’s mechanical properties.
Specifically, Stanford researchers conducted a total of 157 mechanical tests and 288 simulations – via an artificial neural network – of plant-based meat models closely resembling (in mechanical properties) animal meat.
«Our aim was to investigate how closely plant-based meat products could match the mechanical attributes of animal meat, not only in terms of texture and compression profiling but also across the full three-dimensional spectrum of the products analysed». Let’s delve deeper.
Plant-based meat: AI maps the properties that bring it closer to animal meat
Aware that their work has the potential to drive the development of plant-based meat products increasingly similar to animal meat, the team conducted mechanical tests designed to explore the three-dimensional spectrum of all samples.
They achieved this by placing pieces of meat into a machine programmed to stretch, compress, and shear them, three modes intended to simulate human chewing.
Exploring the three-dimensional properties of meats is essential to understanding the subtle differences in texture that influence the sensory experience associated with their consumption.
Following the simulation of chewing through the machine, the authors first used machine learning algorithms to analyse and classify the large volume of data generated by the tests. Subsequently, they developed an artificial neural network that, based on this analysis, learned the mappings of the structural and mechanical characteristics of those plant-based products that most successfully aligned with all three tests (tension, compression, and shearing). The ultimate aim was to replicate these findings to design better plant-based meat.
By combining mechanical tests with artificial intelligence techniques, the team was able to rigorously describe the texture of meats with remarkable accuracy, aligning closely – as we’ll see later – with the evaluations provided by human testers.
Thanks to this methodology, the research group also discovered that some plant-based meat products already on the market are, in fact, confirming the texture of the animal meats they aim to imitate.
Test results
To validate the equations formulated by the neural network and demonstrate their ability to map the perception of plant-based meat texture, the team conducted a survey involving human testers. Participants were asked to taste samples of the eight products and evaluate them on a five-point scale across the following categories: soft, hard, crumbly, chewy, viscous, elastic, sticky, fibrous, fatty, and moist.
«In the mechanical tests – the researchers highlight – plant-based hot dogs and sausages performed very similarly – in the tension, compression, and shearing tests – to their animal counterparts. Plant-based turkey, on the other hand, was found to be twice as stiff as animal turkey, while tofu was much softer than meat-based products».
To the authors’ surprise, human testers also evaluated the hot dogs and sausages in a way that closely aligned with the mechanical tests and the mappings generated by the neural networks.
«What’s truly fascinating is that the rankings created by the testers are almost identical to the rankings produced by the machine. This suggests that, in the future, machines and artificial intelligence techniques could be used to conduct further quantitative and reproducible tests».
Glimpses of Futures
Refining the mechanical properties of plant-based protein products to make their texture (a variable that influences taste) as similar as possible to that of animal-based protein products is a rapidly growing field of research. In the years to come, it could accelerate the development of better plant-based meats, making them more appealing even to the most steadfast traditional meat consumers.
To anticipate potential future scenarios, let us now – using the STEPS framework – explore the social, technological, economic, political, and sustainability impacts that advancements in this field of study might have.
S – SOCIAL: the work presented moves towards a healthier, more efficient, and sustainable global food system. Over the next three decades, the quest for the tastiest plant-based meat could lead to the total elimination of foods produced through animal suffering and farming practices that contribute to the climate crisis. The findings recorded by the scientists at Stanford University – it is worth noting – demonstrate that the data-driven approach is promising in envisioning a method to drive the development of plant-based meat products that are ever more succulent and similar to animal meat. In a future scenario, aided by the progression of the data-driven process described, it is easy to imagine the automated generation – via an AI system – of recipes for plant-based meat menus with predefined characteristics.
T – TECHNOLOGICAL: the Californian study demonstrates that artificial intelligence – more specifically, machine learning – can provide a powerful tool to understand and map correspondences between different food products and their structures. In the context of plant-based meat design, this technique could, in the future, also be used to map the relationship between a desired level of texture and taste and the ingredients needed to create plant-based meat with those precise characteristics. However, a significant limitation to this potential innovation lies in the lack of large datasets needed to train AI models for plant-based meat. This gap stems from the reluctance of research centres and companies in the sector to share information on the subject. «How can we design the perfect plant-based steak without sufficient data?» the team asks. To set an example, they continue testing the mechanical properties of animal- and plant-based meat products and are making their findings publicly available through an open-access database.
E – ECONOMIC: from an economic perspective, in the future, plant-based meat products that increasingly match animal meat in terms of texture and taste – and are therefore more appealing to consumers currently reluctant to make the switch – could act as a driving force for plant-based markets that are currently facing a steep decline. For instance, in the United States, sales of plant-based meat and seafood have dropped by 13% over the past two years [source: “Plant-based: meat, seafood, eggs, and dairy. 2023 State of the Industry Report” – Good Food Institute USA]. The situation differs in the European Union, where the Good Food Institute Europe reported sales of €5.8 billion for the 2020–2022 period, with growth rates exceeding 30% for certain products such as plant-based milk and meat, whose consumption rose by 38% and 35%, respectively. Given these figures, what will become of the animal meat market? It’s hard to predict, especially as we approach the five-year countdown to the completion of the UN’s 2030 Agenda for Sustainable Development, which prioritises the protection of the planet’s natural resources and climate for future generations. When considering intensive farming – the emblem of a meat industry that is neither humane nor sustainable – the gap between the United Nations’ goals on one side and the interests of a sector under global scrutiny on the other becomes starkly apparent.
P – POLITICAL: the Stanford University study brings renewed focus to one of the longstanding debates between the cruelty-free food sector and the animal meat industry. Governments, influenced by political orientation and vested interests, often take differing stances. For example, in November 2023, Italy passed legislation banning both the production and sale of cultivated meat as well as the use of terms like “steak,” “hamburger,” and “sausage” on packaging for plant-based meat products, with heavy penalties for violations. The reasoning cited was the concern that consumers might be misled by such labels. However, on 4 October 2024, the EU Court of Justice ruled that «plant-based foods may continue to be sold and marketed using terms traditionally associated with animal meat, provided their composition is clearly indicated on the packaging». In a future scenario where the methodology employed by the Stanford researchers accelerates the growth of the plant-based food sector, oversight by a regulatory body equivalent to the EU Court of Justice would help safeguard the interests of all stakeholders. This would include advocates of low-carbon, ethically produced foods that respect animal rights.
S – SUSTAINABILITY: as previously highlighted, a world where the global dietary paradigm shifts significantly – or even entirely – away from meat produced by intensive farming to foods derived from vegetables and legumes (with the addition of grains and/or tofu), would embody a harmonious convergence of food sustainability, environmental sustainability, and animal ethics. These plant-based alternatives, rich in nutrients and fibre, and equally as flavourful as animal meat, represent the potential foundation for a sustainable and ethical future.
