{"id":12222,"date":"2025-07-15T15:34:44","date_gmt":"2025-07-15T13:34:44","guid":{"rendered":"https:\/\/it-u.at\/?p=12222"},"modified":"2025-07-16T07:14:10","modified_gmt":"2025-07-16T05:14:10","slug":"ai-meets-neuroscience-itus-research-highlighted-in-quanta-magazine","status":"publish","type":"post","link":"https:\/\/it-u.at\/en\/news\/ai-meets-neuroscience-itus-research-highlighted-in-quanta-magazine\/","title":{"rendered":"AI Meets Neuroscience: IT:U\u2019s Research Highlighted in Quanta Magazine"},"content":{"rendered":"\n

What if artificial intelligence (AI) could learn more like the brain? A recent feature in Quanta Magazine explores this exact question,  spotlighting research co-led by IT:U Founding Professor J<\/a>ie Mei<\/a> <\/a>and Professor Srikanth Ramaswamy<\/a> (from Newcastle University and the Massachusetts Institute of Technology (MIT)). Their work brings neuroscience and AI closer together by modelling biological processes, such as  neuromodulation, in artificial neural networks.<\/em><\/p>\n\n\n\n

Neuromodulators and Learning<\/strong><\/h4>\n\n\n\n

Neuromodulators are the brain’s “master switches”, which enable adaptation to ever-changing environments and task demands. Neuromodulators like dopamine, acetylcholine, and norepinephrine operate across temporal scales, from seconds to hours, fundamentally altering the computational properties of neural circuits. Building on this neurobiological foundation, Professors Mei and Ramaswamy have pioneered research into how the computational principles of neuromodulation can be systematically integrated into artificial neural networks. The implementation involves embedding artificial neuromodulatory signals that can adjust network parameters, learning rates, connection weights, and activation thresholds, based on performance feedback and environmental context. The results demonstrate AI systems that learn more flexibly, exhibiting several key advantages. These systems can switch between focused, exploitation-driven learning and broad, exploratory learning, mirroring the brain’s remarkable ability to balance stability and plasticity.<\/p>\n\n\n\n

Implementing Brain-inspired Rules in AI models<\/strong><\/strong> <\/h4>\n\n\n\n

The duo\u2019s co-led studies show that applying biological principles \u2014 such as different neuron types, their firing properties and neuromodulatory mechanisms \u2014 improves both the performance and efficiency of AI systems. Their research is cited in Quanta Magazine as part of a broader shift in AI thinking: one that moves away from rigid imitation and toward a deeper appreciation for the complexity of biological intelligence.<\/p>\n\n\n\n

\n

“Systematically incorporating more biological principles into AI systems could address current limitations in reasoning and adaptability.\u201d<\/em><\/p>\n– Prof. Srikanth Ramaswamy<\/cite><\/blockquote>\n\n\n\n

Computational Neuroscience at IT:U<\/strong><\/h4>\n\n\n\n

IT:U\u2019s Computational Neuroscience Research Group, led by Prof. Jie Mei, focuses on bridging neuroscience, artificial intelligence, and systems thinking. As part of IT:U\u2019s networked university model, the group collaborates closely with leading researchers and institutions worldwide, such as Prof. Srikanth Ramaswamy, to advance interdisciplinary research that transforms how machine learning, cognition, and human understanding are explored and applied. In addition, IT:U Prof. Nina Hubig<\/a> also contributes to this mission with her work in Explainable AI, supporting efforts to build more transparent and interpretable neural models.<\/p>\n\n\n\n

Beyond research publications, Prof. Jie Mei and Prof. Ramaswamy have made notable contributions to outreach and collaborative project activities such as:<\/p>\n\n\n\n