Memory in a Droplet: Collections of Brain-Inspired Biomolecular Elements

随着人工智能方法的普及，迫切需要开发替代计算方法，这需要更多的资源来执行它们。替代计算的一种方法直接从大脑本身汲取灵感，通过结合离子传输和活组织中存在的选择性膜屏障来重建传统电路的各个方面。这些方法通过配置内存和处理单元来降低功耗，将内存构建到材料本身的结构中。这个项目建立了由脂质包裹的液滴组成的网络，作为神经元模型，能够根据它们传递的信号调整它们的特性。这种调节跨模型突触交换的能力是神经形态结构的关键组成部分，该项目通过数学预测和实验观察的结合，探索在相互连接的合成细胞网络中使用涌现力学来控制交换。该项目的完成将为能够进行化学计算的动态合成组织提供基础，并能够与传统电子和生化信号相结合。此外，该项目为未来的科学家提供跨学科培训，并通过招募和培训本科生和高中生来加强STEM劳动力的发展。这项研究为基于液滴的神经形态材料提供了一种新的方法，将它们扩展到更大的可重构元素的多维网络。研究这些神经形态隔室网络已经证明了内部液滴状态（如电荷积累）如何被利用为一种记忆形式。所选择的方法认识到，液滴界面双层可以通过软材料的粘弹性/粘塑性透镜来观察，从而实现亚稳态行为。将建立一个新的多物理场模型，通过结合电和化学活动来模拟粘接液滴的力学。然后，该模型将用于探索生物膜集合中的复杂电润湿事件，指导生物分子神经形态材料的设计。模型预测将在实验中得到验证，该模型将逐渐扩展到研究在更大的生物分子元素集合中出现的涌现力学。这种大脑启发材料的新方法提供了长期的突触可塑性和通过电压信号永久保留膜结构变化的能力。该奖项反映了美国国家科学基金会的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

The development of alternative computing approaches is urgently needed as artificial intelligence methods become prevalent, requiring greater resources for their execution. One approach to alternative computing draws inspiration directly from the brain itself, reconstructing aspects of conventional circuitry through combinations of the ionic transport and selective membranous barriers present in living tissues. These approaches require low power consumption through collocating memory and processing units, building memory into the structure of the material itself. This project builds networks of lipid-coated droplets as model neurons capable of adapting their properties in response to the signals they transmit. This ability to tune the exchange across the model synapses is a crucial component of neuromorphic architectures, and this project explores controlling the exchanges using emergent mechanics in networks of interconnected synthetic cells through a combination of mathematical predictions and experimental observations. Completion of the project will provide a foundation for dynamic synthetic tissues capable of chemical computation and able to interface both with conventional electronics and biochemical signals. In addition, the project provides interdisciplinary training for future scientists and enhances the development of the STEM workforce by recruiting and training undergraduate and high school students. This research provides a new approach to droplet-based neuromorphic materials, expanding them to larger multidimensional networks of reconfigurable elements. Investigating these networks of neuromorphic compartments has demonstrated how the internal droplet states (such as accumulation of charge) may be exploited as a form of memory. The selected approach recognizes that droplet interface bilayers may be viewed through the lens of viscoelasticity/viscoplasticity in soft materials, enabling metastable behaviors. A novel multiphysics model will be created for simulating adhesive droplet mechanics by combining the electrical and chemical activities. This model will then be used to explore complex electrowetting events in collections of biological membranes, guiding the design of biomolecular neuromorphic materials. Model predictions will be validated experimentally, and the model will be gradually expanded to study emergent mechanics present in larger collections of biomolecular elements operating in unison. This new approach to the brain-inspired material provides long term synaptic plasticity and the ability to permanently retain changes in the membranous structure through voltage signals.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.