Support the show to get full episodes, full archive, and join the Discord community. Today I'm in conversation with Rajesh Rao, a distinguished professor of computer science and engineering at the University of Washington, where he also co-directs the Center for Neurotechnology. Back in 1999, Raj and Dana Ballard published what became quite a famous paper, which proposed how predictive coding might be implemented in brains. What is predictive coding, you may be wondering? It's roughly the idea that your brain is constantly predicting incoming sensory signals, and it generates that prediction as a top-down signal that meets the bottom-up sensory signals. Then the brain computes a difference between the prediction and the actual sensory input, and that difference is sent back up to the "top" where the brain then updates its internal model to make better future predictions. So that was 25 years ago, and it was focused on how the brain handles sensory information. But Raj just recently published an update to the predictive coding framework, one that incorporates actions and perception, suggests how it might be implemented in the cortex - specifically which cortical layers do what - something he calls "Active predictive coding." So we discuss that new proposal, we also talk about his engineering work on brain-computer interface technologies, like BrainNet, which basically connects two brains together, and like neural co-processors, which use an artificial neural network as a prosthetic that can do things like enhance memories, optimize learning, and help restore brain function after strokes, for example. Finally, we discuss Raj's interest and work on deciphering an ancient Indian text, the mysterious Indus script. Raj's website. Related papers A sensory–motor theory of the neocortex. Brain co-processors: using AI to restore and augment brain function. Towards neural co-processors for the brain: combining decoding and encoding in brain–computer interfaces. BrainNet: A Multi-Person Brain-to-Brain Interface for Direct Collaboration Between Brains. 0:00 - Intro 7:40 - Predictive coding origins 16:14 - Early appreciation of recurrence 17:08 - Prediction as a general theory of the brain 18:38 - Rao and Ballard 1999 26:32 - Prediction as a general theory of the brain 33:24 - Perception vs action 33:28 - Active predictive coding 45:04 - Evolving to augment our brains 53:03 - BrainNet 57:12 - Neural co-processors 1:11:19 - Decoding the Indus Script 1:20:18 - Transformer models relation to active predictive coding

Join Grace Hwang, a Program director at NIH specializing in neuroscience and AI, and Joe Monaco, a scientific program manager at NIH, as they dive deep into the future of NeuroAI. They discuss the BRAIN Initiative's successes, the integration of AI with neuroscience, and the significance of neurodynamical computing. Grace and Joe explore innovative concepts like digital twins and their implications for neurosurgery. They also address the need for interdisciplinary collaboration to tackle the complexities of brain research and the ethical dimensions of these advancements.

Hessam Akhlaghpour, a postdoctoral researcher at Rockefeller University, delves into the theoretical realm of molecular computation and its intersections with neuroscience. He unveils fascinating insights on RNA's potential as a universal computational medium. The discussion spans the evolutionary significance of this capability and how combinatory logic plays a role in biological systems. Hessam reflects on how his journey through neuroscience reshaped his understanding of memory and computation, challenging traditional views and igniting new ideas in computational theories.

In this intriguing discussion, Tony Zador, head of the Zador lab at Cold Spring Harbor Laboratory, shares his insights on the synergy between neuroscience and artificial intelligence. He argues that biological principles can significantly improve AI efficiency, particularly through understanding animal behavior. The conversation dives into the evolution of NeuroAI, the pitfalls of current AI models, and the parallels between genetic coding and neural networks. Zador highlights the importance of incorporating developmental learning stages from humans and animals to create more adaptable AI systems.

In this insightful discussion, Karen Adolph, a professor at NYU and head of the Infant Action Lab, shares her groundbreaking research on how infants learn to move and think. Alongside her partner Mark Blumberg, they challenge traditional views of the motor cortex, revealing its role in processing sensory information rather than just motor functions. They dive into the importance of real-world observations in understanding development, the influence of ecological psychology, and how insights from infants can inform advances in artificial intelligence and robotics.

Cristina Savin, a neuroscientist studying learning through recurrent neural networks, and Tim Vogels, who explores synaptic plasticity using AI, join the conversation. They discuss the transformative impact of deep learning on neuroscience research and the balance between innovation and traditional scientific inquiry. The duo reflects on the challenges of staying diverse in methodologies while utilizing AI tools. They also humorously address the academic pressures of productivity and work-life balance, emphasizing the importance of interdisciplinary collaboration and broad reading in research.

Mikkel Lepperød, an organizer of the NeuroAI workshop, joins neuroscientists Ken Harris and Andreas Tolias to explore AI's influence on neuroscience. They delve into the intersection of neural modeling and AI, discussing the balance between predictive accuracy and interpretability. The conversation highlights the role of deep learning in understanding cognition, the potential pitfalls of AI in research, and the philosophical implications of modern models. They also share insights on validating scientific ideas and the evolving productivity landscape in academia.

Vijay Namboodiri, who runs the Nam Lab at UCSF, teams up with Ali Mohebi, an assistant professor at UW-Madison, to dive deep into the intricacies of dopamine. They challenge the classic narrative of dopamine's role in reward prediction, proposing a retrospective view that redefines how we understand causal relationships. Their discussions cover sign tracking versus goal tracking in learning, the implications for addiction, and the need for new models that integrate temporal differences. They also touch on how our past experiences inform current decisions, shaping our understanding of learning.

Kim Stachenfeld, a Senior Research Scientist at Google DeepMind and a researcher at Columbia's Center for Theoretical Neuroscience, dives into the captivating world of neuroscience and AI. She discusses the critical role of neural networks in emulating human cognition and their applications in understanding the brain. Kim explores the nuances of reinforcement learning, the intersection of academia and industry, and insights into memory and intelligence. She also challenges traditional model hierarchies, emphasizing the need for predictive and interpretable models in AI.

Àlex Gómez-Marín, a theoretical physicist turned neuroscientist, heads The Behavior of Organisms Laboratory in Alicante, Spain. He explores the "edges of consciousness," discussing experiences like hallucinogens and near-death episodes. Àlex shares profound insights from his own near-death experience, challenging conventional scientific approaches and embracing a pluralistic view of consciousness. He highlights the importance of open-mindedness in understanding anomalies and discusses the philosophical implications of consciousness in today's technological age.