The human brain is an extraordinary organ that far surpasses any computer in terms of power and complexity. With an estimated 86 billion neurons and up to a quadrillion synapses, the brain has an unparalleled capacity to process information at incredible speeds. Unlike modern computing devices with physically separated processor and memory units, the brain’s efficiency lies in the fact that neurons can serve as both processors and memory devices. Many efforts have been made to develop computer systems that emulate the brain’s capabilities. However, a groundbreaking new development called Brainoware takes this concept further by integrating real human brain tissue with electronics. Led by engineer Feng Guo of Indiana University Bloomington, a team successfully fed Brainoware tasks such as speech recognition and nonlinear equation prediction, demonstrating an important first step towards a new kind of computer architecture. While the researchers followed ethical guidelines in developing Brainoware, it is crucial to consider the numerous neuroethical issues associated with biocomputing systems incorporating human neural tissue, as noted by researchers from Johns Hopkins University in a related commentary in Nature Electronics.
Simulating the activity of the human brain in an artificial system has been an ongoing challenge. In 2013, Riken’s K Computer attempted to mimic the brain’s activity using 82,944 processors and a petabyte of main memory. However, it took a staggering 40 minutes to simulate just one second of the activity of 1.73 billion neurons interconnected by 10.4 trillion synapses, representing only a small percentage of the brain’s capacity. Scientists and engineers have since been working on alternative approaches to replicate the brain’s capabilities through hardware and algorithms. Neuromorphic computing, which aims to mimic the brain’s structure and functionality, has shown promise. However, it is energy-intensive and training artificial neural networks can be time-consuming. This led Guo and his colleagues to explore a different approach by using real human brain tissue grown in a lab.
Guo’s team developed Brainoware by growing human pluripotent stem cells into brain organoids, 3D structures that mimic the brain’s organization. These mini-brains consist of tissue arrangements without conscious thought or emotion but provide valuable insights into brain development and functioning. Brainoware involves connecting these brain organoids to an array of high-density microelectrodes, utilizing a type of artificial neural network called reservoir computing. Information is transported into the organoid through electrical stimulation, which then processes the information within the reservoir. Brainoware produces its calculations in the form of neural activity and utilizes normal computer hardware for input and output layers. These layers require training to effectively work with the organoid, with the output layer reading the neural data and making classifications or predictions based on the input.
To demonstrate Brainoware’s capabilities, the researchers presented it with 240 audio clips of Japanese vowel sounds from eight male speakers and asked it to identify a specific individual’s voice. After just two days of training, Brainoware achieved an accuracy rate of 78% in identifying the speaker. The team also tasked Brainoware with predicting a Hénon map, a chaotic dynamical system. After four days of unsupervised learning, Brainoware showed better accuracy in predicting the map compared to an artificial neural network without a long short-term memory unit. Although slightly less accurate than neural networks with a long short-term memory unit that underwent 50 training epochs, Brainoware achieved similar results within a significantly shorter training time. The researchers highlight the high plasticity and adaptability of organoids, emphasizing the potential of Brainoware for adaptive reservoir computing.
While Brainoware shows remarkable potential, there are significant limitations to address, such as maintaining the health and vitality of the brain organoids and managing the power consumption of peripheral equipment. Nonetheless, with careful consideration of the ethical implications, Brainoware has implications beyond computing. It offers insights into the mechanisms of learning, neural development, and the cognitive implications of neurodegenerative diseases. Although it may take decades before fully functional biocomputing systems can be realized, the research conducted with Brainoware can pave the way for foundational discoveries in understanding the complexities of the human brain. As Lena Smirnova, Brian Caffo, and Erik C. Johnson from Johns Hopkins University caution, it is crucial for the scientific community to deeply examine the neuroethical issues involved in advancing biocomputing systems incorporating human neural tissue.
Brainoware represents a pioneering advancement in the field of biocomputing by integrating real human brain tissue with electronics. With its ability to process information and perform tasks, Brainoware takes a significant step towards emulating the brain’s extraordinary capabilities. While there are challenges and ethical considerations to address, the potential for enhanced understanding of the human brain and the development of more sophisticated computing systems is immense. The journey towards fully functional biocomputing systems may be long and complex, but the insights gained from research like Brainoware will undoubtedly shape the future of computing and neuroscience.
Leave a Reply