Imagine a world where neuroscientists from across the globe unite to tackle one of the brain's most intriguing mysteries: predictive processing. But here's the catch—this isn't your typical research endeavor. It's a bold, community-driven experiment that challenges the very way we study the brain. And this is the part most people miss: it's not just about the science; it's about revolutionizing how we collaborate.
In a groundbreaking initiative, over 50 scientists joined forces to pinpoint the critical missing pieces needed to rigorously test predictive processing theories. Traditionally, neuroscience research operates in silos, with individual labs pursuing their own agendas. This approach has yielded remarkable discoveries—from synaptic learning breakthroughs to insights into how the cortex interprets visual scenes. Yet, the brain's complexity demands a more integrated strategy. The challenge? Our current system, while productive, often incentivizes fragmentation rather than collaboration.
Enter OpenScope, a platform pioneered by Christof Koch, myself, and colleagues at the Allen Institute. Inspired by large-scale observatories in astronomy and physics, OpenScope facilitates shared, high-throughput neurophysiology experiments. Scientists worldwide propose projects, and OpenScope records neural activity from thousands of neurons across the mouse brain using advanced techniques like Neuropixels and two-photon calcium imaging. The datasets are then shared with research teams for analysis before being made publicly available.
Initially, OpenScope focused on independent projects selected through a double-blind review process. However, a pattern emerged: many proposals centered on predictive processing. With guidance from a stellar steering committee—Adrienne Fairhall, Satrajit Ghosh, Mackenzie Mathis, Konrad Körding, Joel Zylberberg, and Nick Steinmetz—and support from the U.S. National Institutes of Health, we pivoted. We decided to rally multiple labs around a single, community-designed experiment to unify disparate theories and data on predictive processing.
But here's where it gets controversial: Can a radically open, collaborative approach truly bridge the gaps in our understanding of the brain? We started by proposing a workshop at the 2024 Cognitive Computational Neuroscience conference in Boston. Realizing a workshop alone wouldn't suffice, we created a shared Google Doc months in advance. This document became the epicenter of a global conversation, where scientists reviewed predictive processing literature, debated theories, and identified missing data. Inspired by Ray Dalio’s principles, we granted full editing access to everyone, fostering trust and respectful dialogue.
The results were astonishing. Over a year, the document evolved into a comprehensive review article, with over 1,900 comments and contributions from a diverse community. By October 2024, we’d outgrown Google Docs and migrated to GitHub, continuing our discussions in a public forum. Weekly meetings, now recorded and shared on YouTube, keep the community engaged across time zones. Pilot experiments began at the Allen Institute in April 2025, with datasets analyzed collaboratively and shared on platforms like DANDI.
The review, now on arXiv, highlights a fascinating insight: predictive processing might not rely on a single mechanism but rather a ‘bag of models,’ with the brain dynamically switching strategies based on context. This challenges traditional assumptions and opens new avenues for exploration. Our ‘science in the open’ approach has proven so effective that participants are eager to apply it to other neuroscience questions.
Here’s the thought-provoking question for you: Could this model of open, collaborative science become the new standard in neuroscience? Or does the traditional siloed approach still hold irreplaceable value? Share your thoughts in the comments—let’s spark a conversation that could shape the future of brain research.
