Expeditions: Mind in Vitro — Computing with Living Neurons

计算机系统能用活的神经元建造吗？它们能否实现认知的基本特征，如学习、注意力、好奇心或创造力，这些在生物学中如此普遍，但在现代计算中却难以捉摸？这次探险设想的计算机和机器人是人类设计的，但却是活的。这是可以编程的，但没有指定其行为，而是会出现。这些系统将成长、治愈、学习和探索。通过从数字、刚性建筑到塑料蜂窝基板的转变，速度、精度和精确度被取而代之的是统计的稳健性、极端的并行性、治愈、增长和卓越的能源效率。这些后冯·诺伊曼系统不是布尔逻辑，而是利用数十亿个神经元件的组成动态来产生“开箱即用”的、新出现的计算形式。它们将与肌肉和传感器连接，产生能够探测它们的环境并探索它的有机机器。这项技术将对与信息处理、机器人、健康和医学相关的几乎所有领域产生深远而持久的影响，并对人类知识产生深刻的影响。在耗电的超级计算机上实现的现代深度网络，可能会与苹果大小、靠糖运行的生物计算系统相比相形见绌。神经科学可能会因为全新的行为模型而发生革命性的变化。伦理学研究将被催化。这一遗产将是从人工智能和动物中分离出来的第三种体外智能形式的开始。基于多细胞结构工程的最新进展，通过统计物理对其动力学的洞察，以及信息论的指导，这次探险将发展制造、建模、编程、规模和体现生物处理器的科学和技术。研究将围绕使系统计算和行动的四个方面展开：(1)湿器-在工程平台上整合神经文化；(2)架构-创建可编程的基板以支持有用的计算；(3)编程-开发软件堆栈和编程模型以配置和运行基板；(4)机器人实施-演示多感官处理并探索移动生物机器人中基本认知特征的出现。一个强大的设计伦理计划将渗透到研究的各个方面。这次探险的唤起力量将使来自各种背景和各个水平的学生对计算感到兴奋。这将被用来通过实习、研讨会、研讨会和专门的迷你课程在体外社区中培养思想。在大型公共场所举办的科学艺术展览将使我们能够与广泛和多样化的观众建立联系。最后，完全致力于开放科学是核心，协议、软件、硬件和教育材料都将免费提供。这一奖项反映了NSF的法定使命，并通过使用基金会的智力优势和更广泛的影响审查标准进行评估，被认为值得支持。

Can computing systems be built out of living neurons? Can they achieve basic hallmarks of cognition such as learning, attention, curiosity, or creativity, so pervasive in biology yet elusive in modern computing? This Expedition imagines computers and robots that are human designed but living. That can be programmed, but whose behaviors are not specified, and instead, emerge. These systems will grow, heal, learn and explore. They will open a new space of possibilities yet to be imagined.By shifting from digital, rigid architectures to plastic cellular substrates, speed, accuracy, and exactness are traded off for statistical robustness, extreme parallelism, healing, growth, and superior energy efficiency. Rather than Boolean logic, these post-von Neumann systems will harness the compositional dynamics of billions of neural elements to generate 'out-of-the-box', emergent forms of computations. They will interface with muscles and sensors, to give rise to organic machines able to probe their environment and explore it. This technology will have profound, lasting impact in virtually every field related to information processing, robotics, health and medicine, with deep ramification across human knowledge. Modern deep-networks implemented on power-hungry supercomputers may be dwarfed by bio-computing systems the size of an apple and running on sugar. Neuroscience could be revolutionized, with radically new behavioral models. Ethics research will be catalyzed. The legacy will be the inception of an in-vitro third form of ‘intelligence’, separate from AI and animal.Building on recent advances in the engineering of multi-cellular constructs, insights into their dynamics via statistical physics, and guidance from information theory, this Expedition will develop the science and technology to fabricate, model, program, scale and embody biological processors. Research will unfold across four thrusts, structured around what makes a system compute and act: (1) Wetware – integrate neural cultures on engineered platforms; (2) Architecture – create a programmable substrate to support useful computations; (3) Programming – develop a software stack and a programming model to configure and run the substrate; (4) Robotic embodiment – demonstrate multi-sensory processing and probe the emergence of rudimentary cognitive traits in motile biological robots. A robust in-design ethics program will be pervasive of all aspects of research. The evocative power of this Expedition will excite students from all backgrounds and at all levels about computing. This will be leveraged to grow a Mind in Vitro community, through internships, workshops, seminars, and a dedicated mini-curriculum. Art-of-Science exhibitions in massive public spaces will allow us to connect with a broad and diverse audience. Finally, full commitment to open science is core, and protocols, software, hardware, and educational material will all be made freely available.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.