Positive Feedback Traps New Ideas

Key Points:

• A new Northwestern University study reveals that Hebbian learning, which strengthens connections between neurons or social contacts that interact repeatedly, can paradoxically trap new information in tight feedback loops, preventing its spread.
• Positive reinforcement, by strengthening existing connections, creates "death spirals" similar to ant mills where activity circulates without reaching new areas, while negative reinforcement weakens connections and promotes the exploration of new network paths.
• The research introduces a novel theoretical framework showing that learning-driven changes in network connections significantly impact how activity spreads across various systems, including social networks, brain signaling, infection transmission, and animal behavior.
• The findings suggest that efficient spreading of ideas or signals depends on weakening old connections to avoid echo chambers and encourage broader dissemination, challenging traditional models that assume static relationships.
• Published in Communications Physics, the study highlights the universal dynamics of activity propagation influenced by learning and plans to test these effects in real-world networks to better understand complex adaptive systems.