大腦如何辨識氣味？中國醫大江安世校長研究團隊揭開神經網路運作奧祕
日期：2026-10-22
資料來源：研究亮點｜江安世校長
分享：
大腦如何辨識氣味？中國醫大江安世校長研究團隊揭開神經網路運作奧祕
為什麼我們能夠迅速辨識食物香氣，甚至從不同氣味中判斷周遭環境？這些看似自然的能力，背後其實藏著精密的神經網路運作機制。
中國醫藥大學江安世校長參與領導的跨機構研究團隊，在國際期刊《Science Advances》發表重要腦科學研究成果，揭開大腦如何透過特殊神經連結辨識及處理不同氣味。
研究團隊以果蠅大腦為研究對象，結合高解析度神經連結圖譜、電腦模擬與神經活動影像分析，發現大腦處理氣味的神經網路並非完全隨機連結，而是具有精密分工的「混合式神經網路」。
這種特殊結構讓不同神經細胞各司其職，有些擅長辨識食物氣味，有些則整合多種氣味訊息，使大腦能兼顧感知敏感度與辨識準確性。
這項發現不僅深化人類對大腦感知、學習與記憶機制的理解，也為未來神經連結體研究及人工智慧神經網路設計提供新的科學啟發。
Research Team Uncovers the Secrets of Neural Networks
How can we instantly recognize the aroma of food or distinguish different smells in our surroundings? Behind these seemingly effortless abilities lies a remarkably sophisticated network of neural connections in the brain.
A collaborative research team co-led by President Ann-Shyn Chiang of China Medical University (CMU) has published an important neuroscience study in the prestigious journal Science Advances, revealing how specialized neural connections enable the brain to recognize and process different odors.
Using the fruit fly brain as a model, the researchers combined high-resolution brain connectomics, computational modeling, and neural activity imaging to investigate how olfactory information is processed. Their findings revealed that the brain's odor-processing circuits are not entirely random but instead form sophisticated hybrid neural networks with specialized functions.
Within these networks, different neurons perform distinct roles. Some are specialized in recognizing food-related odors, while others integrate information from multiple odor signals, allowing the brain to balance sensory sensitivity with accurate odor recognition.
This discovery advances our understanding of how neural circuits support sensory perception, learning, and memory. It also provides valuable insights for future research in brain connectomics and artificial intelligence, potentially inspiring new approaches to designing more efficient artificial neural networks.
原文網址: https://www.biorxiv.org/content/10.1101/2023.12.06.570338v1.abstract
相關圖片：
回前頁