{"id":874,"date":"2026-01-11T03:30:17","date_gmt":"2026-01-10T18:30:17","guid":{"rendered":"https:\/\/itexplore.org\/jp\/columns\/ai-research-frontiers-reasoning-structure-adaptability\/"},"modified":"2026-01-11T03:30:17","modified_gmt":"2026-01-10T18:30:17","slug":"ai-research-frontiers-reasoning-structure-adaptability","status":"publish","type":"post","link":"https:\/\/itexplore.org\/jp\/columns\/ai-research-frontiers-reasoning-structure-adaptability\/","title":{"rendered":"AI\u7814\u7a76\u306e\u6700\u524d\u7dda\uff1a\u63a8\u8ad6\u3001\u69cb\u9020\u3001\u9069\u5fdc\u6027"},"content":{"rendered":"<p>\u672c\u65e5\u306e\u6ce8\u76eeAI\u30fb\u30c6\u30c3\u30af\u30cb\u30e5\u30fc\u30b9\u3092\u3001\u5c02\u9580\u7684\u306a\u5206\u6790\u3068\u5171\u306b\u304a\u5c4a\u3051\u3057\u307e\u3059\u3002<\/p>\n<div class=\"wp-block-vk-blocks-alert vk_alert alert alert-warning has-alert-icon\">\n<div class=\"vk_alert_icon\">\n<div class=\"vk_alert_icon_icon\"><i class=\"fa-solid fa-triangle-exclamation\" aria-hidden=\"true\"><\/i><\/div>\n<div class=\"vk_alert_icon_text\"><span>Warning<\/span><\/div>\n<\/div>\n<div class=\"vk_alert_content\">\n<p>\u3053\u306e\u8a18\u4e8b\u306fAI\u306b\u3088\u3063\u3066\u81ea\u52d5\u751f\u6210\u30fb\u5206\u6790\u3055\u308c\u305f\u3082\u306e\u3067\u3059\u3002AI\u306e\u6027\u8cea\u4e0a\u3001\u4e8b\u5b9f\u8aa4\u8a8d\u304c\u542b\u307e\u308c\u308b\u53ef\u80fd\u6027\u304c\u3042\u308b\u305f\u3081\u3001\u91cd\u8981\u306a\u5224\u65ad\u3092\u4e0b\u3059\u969b\u306f\u5fc5\u305a\u30ea\u30f3\u30af\u5148\u306e\u4e00\u6b21\u30bd\u30fc\u30b9\u3092\u3054\u78ba\u8a8d\u304f\u3060\u3055\u3044\u3002<\/p>\n<\/div>\n<\/div>\n<div class=\"wp-block-group\" style=\"margin-top:40px;margin-bottom:40px\">\n<h2 class=\"wp-block-heading\">H\u00e1n D\u0101n Xu\u00e9 B\u00f9\uff08\u6a21\u5023\uff09\u304b Q\u012bng Ch\u016b Y\u00fa L\u00e1n\uff08\u7fd2\u719f\uff09\u304b\uff1f\u5927\u898f\u6a21\u8a00\u8a9e\u30e2\u30c7\u30eb\u306b\u304a\u3051\u308b\u63a8\u8ad6\u84b8\u7559\u306e\u8a8d\u77e5\u7684\u8996\u70b9<\/h2>\n<ul>\n<li><strong>\u539f\u984c:<\/strong> H\u00e1n D\u0101n Xu\u00e9 B\u00f9 (Mimicry) or Q\u012bng Ch\u016b Y\u00fa L\u00e1n (Mastery)? A Cognitive Perspective on Reasoning Distillation in Large Language Models<\/li>\n<\/ul>\n<h3 class=\"wp-block-heading\">\u5c02\u9580\u30a2\u30ca\u30ea\u30b9\u30c8\u306e\u5206\u6790<\/h3>\n<p>\u672c\u7814\u7a76\u306f\u3001\u5927\u898f\u6a21\u8a00\u8a9e\u30e2\u30c7\u30eb\uff08LLM\uff09\u306b\u304a\u3051\u308b\u63a8\u8ad6\u84b8\u7559\u306e\u73fe\u72b6\u3092\u3001\u8a8d\u77e5\u79d1\u5b66\u7684\u89b3\u70b9\u304b\u3089\u5206\u6790\u3057\u3066\u3044\u307e\u3059\u3002\u5f37\u5316\u5b66\u7fd2\u3067\u8a13\u7df4\u3055\u308c\u305fLLM\u306f\u4eba\u9593\u306e\u8a8d\u77e5\u30b3\u30b9\u30c8\u306b\u81ea\u7136\u306b\u6cbf\u3063\u305f\u632f\u308b\u821e\u3044\u3092\u793a\u3057\u307e\u3059\u304c\u3001\u6559\u5e2b\u30e2\u30c7\u30eb\u306e\u63a8\u8ad6\u30d7\u30ed\u30bb\u30b9\u3092\u6a21\u5023\u3055\u305b\u308b\u6559\u5e2b\u3042\u308a\u30d5\u30a1\u30a4\u30f3\u30c1\u30e5\u30fc\u30cb\u30f3\u30b0\uff08SFT\uff09\u306b\u3088\u308b\u84b8\u7559\u3067\u306f\u3001\u3053\u306e\u8a8d\u77e5\u69cb\u9020\u304c\u4f1d\u9054\u3055\u308c\u306a\u3044\u3053\u3068\u3092\u660e\u3089\u304b\u306b\u3057\u307e\u3057\u305f\u300214\u306e\u30e2\u30c7\u30eb\u3092\u7528\u3044\u305f\u5b9f\u9a13\u3067\u306f\u3001\u300cH\u00e1n D\u0101n Xu\u00e9 B\u00f9\u300d\uff08\u8868\u9762\u7684\u306a\u6a21\u5023\uff09\u4eee\u8aac\u3092\u691c\u8a3c\u3057\u3001\u84b8\u7559\u304c\u300c\u6a5f\u80fd\u7684\u6574\u5408\u6027\u5d29\u58ca\u300d\u3092\u5f15\u304d\u8d77\u3053\u3059\u3053\u3068\u3092\u767a\u898b\u3057\u307e\u3057\u305f\u3002\u6559\u5e2b\u30e2\u30c7\u30eb\u304c\u4eba\u9593\u306e\u96e3\u6613\u5ea6\u30b9\u30b1\u30fc\u30ea\u30f3\u30b0\u3092\u53cd\u6620\u3059\u308b\u306e\u306b\u5bfe\u3057\u3001\u84b8\u7559\u3055\u308c\u305f\u30e2\u30c7\u30eb\u306f\u8457\u3057\u304f\u305d\u306e\u6574\u5408\u6027\u3092\u4f4e\u4e0b\u3055\u305b\u3001\u3057\u3070\u3057\u3070\u84b8\u7559\u524d\u306e\u30d9\u30fc\u30b9\u30e9\u30a4\u30f3\u3092\u4e0b\u56de\u308b\u6027\u80fd\u3092\u793a\u3057\u307e\u3057\u305f\uff08\u300c\u8ca0\u306e\u8ee2\u79fb\u300d\uff09\u3002\u5206\u6790\u306b\u3088\u308b\u3068\u3001SFT\u306f\u3001\u63a8\u8ad6\u306e\u8a00\u8a9e\u7684\u5f62\u5f0f\uff08\u5197\u9577\u6027\uff09\u3092\u5100\u5f0f\u7684\u306b\u6a21\u5023\u3059\u308b\u3082\u306e\u306e\u3001\u6559\u5e2b\u306e\u52d5\u7684\u306a\u30ea\u30bd\u30fc\u30b9\u914d\u5206\u30dd\u30ea\u30b7\u30fc\u3092\u5185\u9762\u5316\u3057\u306a\u3044\u300c\u30ab\u30fc\u30b4\u30ab\u30eb\u30c8\u300d\u52b9\u679c\u3092\u8a98\u767a\u3057\u307e\u3059\u3002\u7d50\u679c\u3068\u3057\u3066\u3001\u63a8\u8ad6\u84b8\u7559\u306f\u8a08\u7b97\u30b3\u30b9\u30c8\u3068\u8a8d\u77e5\u9700\u8981\u3092\u5207\u308a\u96e2\u3057\u3001\u4eba\u9593\u306e\u3088\u3046\u306a\u8a8d\u77e5\u306f\u53d7\u52d5\u7684\u306a\u6a21\u5023\u3067\u306f\u306a\u304f\u3001\u80fd\u52d5\u7684\u306a\u5f37\u5316\u5b66\u7fd2\u306e\u5275\u767a\u7684\u306a\u7279\u6027\u3067\u3042\u308b\u3053\u3068\u3092\u793a\u5506\u3057\u3066\u3044\u307e\u3059\u3002<\/p>\n<p>\ud83d\udc49 <strong><a href=\"https:\/\/arxiv.org\/abs\/2601.05019\" target=\"_blank\" rel=\"noopener\">arXiv \u3067\u8a18\u4e8b\u5168\u6587\u3092\u8aad\u3080<\/a><\/strong><\/p>\n<ul>\n<li><strong>\u8981\u70b9:<\/strong> Supervised Fine-Tuning (SFT) for reasoning distillation in LLMs leads to a 'Functional Alignment Collapse,' where models mimic the form but not the cognitive process of reasoning, resulting in negative transfer and decoupling computational cost from cognitive demand.<\/li>\n<li><strong>\u8457\u8005:<\/strong> Yueqing Hu, Xinyang Peng, Shuting Peng, Hanqi Wang, Tianhong Wang<\/li>\n<\/ul>\n<blockquote class=\"wp-block-quote\">\n<p>This research analyzes the current state of reasoning distillation in Large Language Models (LLMs) from a cognitive science perspective. While LLMs trained via reinforcement learning exhibit behavior naturally aligned with human cognitive costs, the study reveals that distillation through Supervised Fine-Tuning (SFT), which trains student models to mimic the reasoning process of teacher models, fails to transmit this cognitive structure. Experiments with 14 models tested the 'H\u00e1n D\u0101n Xu\u00e9 B\u00f9' (Superficial Mimicry) hypothesis, finding that distillation induces a 'Functional Alignment Collapse.' Teacher models mirror human difficulty scaling, whereas distilled students significantly degrade this alignment, often underperforming their pre-distillation baselines ('Negative Transfer'). The analysis suggests SFT induces a 'Cargo Cult' effect, where students ritualistically replicate the linguistic form of reasoning (verbosity) without internalizing the teacher's dynamic resource allocation policy. Consequently, reasoning distillation decouples computational cost from cognitive demand, indicating that human-like cognition is an emergent property of active reinforcement rather than passive imitation.<\/p>\n<\/blockquote>\n<\/div>\n<div class=\"wp-block-group\" style=\"margin-top:40px;margin-bottom:40px\">\n<h2 class=\"wp-block-heading\">\u52d5\u7684\u5927\u898f\u6a21\u6982\u5ff5\u30e2\u30c7\u30eb\uff1a\u9069\u5fdc\u7684\u610f\u5473\u7a7a\u9593\u306b\u304a\u3051\u308b\u6f5c\u5728\u7684\u63a8\u8ad6<\/h2>\n<ul>\n<li><strong>\u539f\u984c:<\/strong> Dynamic Large Concept Models: Latent Reasoning in an Adaptive Semantic Space<\/li>\n<\/ul>\n<h3 class=\"wp-block-heading\">\u5c02\u9580\u30a2\u30ca\u30ea\u30b9\u30c8\u306e\u5206\u6790<\/h3>\n<p>\u5927\u898f\u6a21\u8a00\u8a9e\u30e2\u30c7\u30eb\uff08LLM\uff09\u306f\u3001\u8a00\u8a9e\u304c\u6301\u3064\u975e\u5747\u4e00\u306a\u60c5\u5831\u5bc6\u5ea6\u306b\u3082\u304b\u304b\u308f\u3089\u305a\u3001\u5168\u3066\u306e\u30c8\u30fc\u30af\u30f3\u306b\u5747\u4e00\u306a\u8a08\u7b97\u3092\u9069\u7528\u3057\u3066\u3044\u307e\u3059\u3002\u3053\u306e\u30c8\u30fc\u30af\u30f3\u5747\u4e00\u6027\u306f\u3001\u5c40\u6240\u7684\u306b\u4e88\u6e2c\u53ef\u80fd\u306a\u7bc4\u56f2\u3067\u80fd\u529b\u3092\u6d6a\u8cbb\u3059\u308b\u4e00\u65b9\u3067\u3001\u610f\u5473\u7684\u306b\u91cd\u8981\u306a\u9077\u79fb\u306b\u306f\u8a08\u7b97\u80fd\u529b\u3092\u5272\u308a\u5f53\u3066\u4e0d\u8db3\u3057\u307e\u3059\u3002\u672c\u7814\u7a76\u3067\u306f\u3001\u6f5c\u5728\u8868\u73fe\u304b\u3089\u610f\u5473\u5883\u754c\u3092\u5b66\u7fd2\u3057\u3001\u63a8\u8ad6\u304c\u3088\u308a\u52b9\u7387\u7684\u306a\u5727\u7e2e\u3055\u308c\u305f\u6982\u5ff5\u7a7a\u9593\u3078\u3068\u8a08\u7b97\u3092\u30b7\u30d5\u30c8\u3055\u305b\u308b\u968e\u5c64\u7684\u8a00\u8a9e\u30e2\u30c7\u30ea\u30f3\u30b0\u30d5\u30ec\u30fc\u30e0\u30ef\u30fc\u30af\u3067\u3042\u308b\u300c\u52d5\u7684\u5927\u898f\u6a21\u6982\u5ff5\u30e2\u30c7\u30eb\uff08DLCM\uff09\u300d\u3092\u63d0\u6848\u3057\u307e\u3059\u3002DLCM\u306f\u3001\u5b9a\u7fa9\u6e08\u307f\u306e\u8a00\u8a9e\u5358\u4f4d\u306b\u4f9d\u5b58\u305b\u305a\u306b\u3001\u30a8\u30f3\u30c9\u30c4\u30fc\u30a8\u30f3\u30c9\u3067\u53ef\u5909\u9577\u306e\u6982\u5ff5\u3092\u767a\u898b\u3057\u307e\u3059\u3002\u968e\u5c64\u7684\u5727\u7e2e\u306f\u30b9\u30b1\u30fc\u30ea\u30f3\u30b0\u6319\u52d5\u3092\u6839\u672c\u7684\u306b\u5909\u5316\u3055\u305b\u307e\u3059\u3002\u672c\u7814\u7a76\u3067\u306f\u3001\u30c8\u30fc\u30af\u30f3\u30ec\u30d9\u30eb\u306e\u5bb9\u91cf\u3001\u6982\u5ff5\u30ec\u30d9\u30eb\u306e\u63a8\u8ad6\u5bb9\u91cf\u3001\u304a\u3088\u3073\u5727\u7e2e\u7387\u3092\u5206\u96e2\u3057\u3001\u56fa\u5b9aFLOPs\u4e0b\u3067\u306e\u539f\u5247\u7684\u306a\u8a08\u7b97\u5272\u308a\u5f53\u3066\u3092\u53ef\u80fd\u306b\u3059\u308b\u3001\u521d\u306e\u300c\u5727\u7e2e\u3092\u8003\u616e\u3057\u305f\u30b9\u30b1\u30fc\u30ea\u30f3\u30b0\u5247\u300d\u3092\u5c0e\u5165\u3057\u307e\u3059\u3002\u3053\u306e\u4e0d\u5747\u4e00\u306a\u30a2\u30fc\u30ad\u30c6\u30af\u30c1\u30e3\u3092\u5b89\u5b9a\u3057\u3066\u8a13\u7df4\u3059\u308b\u305f\u3081\u306b\u3001\u5e45\u3084\u5727\u7e2e\u30ec\u30b8\u30fc\u30e0\u9593\u3067\u30bc\u30ed\u30b7\u30e7\u30c3\u30c8\u306e\u30cf\u30a4\u30d1\u30fc\u30d1\u30e9\u30e1\u30fc\u30bf\u8ee2\u9001\u3092\u30b5\u30dd\u30fc\u30c8\u3059\u308b\u300c\u5206\u96e2\u3055\u308c\u305f\u03bcP\u30d1\u30e9\u30e1\u30fc\u30bf\u5316\u300d\u3082\u958b\u767a\u3057\u307e\u3057\u305f\u3002\u5b9f\u7528\u7684\u306a\u8a2d\u5b9a\uff08R=4\u3001\u6982\u5ff5\u3042\u305f\u308a\u5e73\u57474\u30c8\u30fc\u30af\u30f3\u306b\u76f8\u5f53\uff09\u3067\u306f\u3001DLCM\u306f\u63a8\u8ad6\u8a08\u7b97\u306e\u7d043\u5206\u306e1\u3092\u3088\u308a\u9ad8\u5bb9\u91cf\u306e\u63a8\u8ad6\u30d0\u30c3\u30af\u30dc\u30fc\u30f3\u306b\u518d\u5272\u308a\u5f53\u3066\u3057\u3001\u540c\u7b49\u306e\u63a8\u8ad6FLOPs\u4e0b\u306712\u306e\u30bc\u30ed\u30b7\u30e7\u30c3\u30c8\u30d9\u30f3\u30c1\u30de\u30fc\u30af\u5168\u4f53\u3067\u5e73\u5747+2.69%\u306e\u6539\u5584\u3092\u9054\u6210\u3057\u307e\u3057\u305f\u3002<\/p>\n<p>\ud83d\udc49 <strong><a href=\"https:\/\/arxiv.org\/abs\/2512.24617\" target=\"_blank\" rel=\"noopener\">arXiv \u3067\u8a18\u4e8b\u5168\u6587\u3092\u8aad\u3080<\/a><\/strong><\/p>\n<ul>\n<li><strong>\u8981\u70b9:<\/strong> Dynamic Large Concept Models (DLCM) offer a hierarchical framework that reallocates computation from uniform token processing to a compressed concept space, improving reasoning efficiency and achieving better performance on zero-shot benchmarks by introducing a compression-aware scaling law.<\/li>\n<li><strong>\u8457\u8005:<\/strong> Editorial Staff<\/li>\n<\/ul>\n<blockquote class=\"wp-block-quote\">\n<p>Large Language Models (LLMs) apply uniform computation to all tokens, despite language exhibiting highly non-uniform information density. This token-uniform regime wastes capacity on locally predictable spans while under-allocating computation to semantically critical transitions. The paper proposes 'Dynamic Large Concept Models (DLCM),' a hierarchical language modeling framework that learns semantic boundaries from latent representations and shifts computation from tokens to a compressed concept space where reasoning is more efficient. DLCM discovers variable-length concepts end-to-end without relying on predefined linguistic units. Hierarchical compression fundamentally changes scaling behavior. The study introduces the first 'compression-aware scaling law,' which disentangles token-level capacity, concept-level reasoning capacity, and compression ratio, enabling principled compute allocation under fixed FLOPs. To stably train this heterogeneous architecture, a 'decoupled \u03bcP parametrization' is developed to support zero-shot hyperparameter transfer across widths and compression regimes. In a practical setting (R=4, corresponding to an average of four tokens per concept), DLCM reallocates roughly one-third of inference compute into a higher-capacity reasoning backbone, achieving a +2.69% average improvement across 12 zero-shot benchmarks under matched inference FLOPs.<\/p>\n<\/blockquote>\n<\/div>\n<div class=\"wp-block-group\" style=\"margin-top:40px;margin-bottom:40px\">\n<h2 class=\"wp-block-heading\">\u8133\u306e\u3088\u3046\u306a\u91cd\u307f\u4ed8\u3051\u30fb\u6307\u5411\u6027\u30cb\u30e5\u30fc\u30ed\u30f3\u30cd\u30c3\u30c8\u30ef\u30fc\u30af\u306e\u51fa\u73fe\u306e\u305f\u3081\u306e\u5e7e\u4f55\u5b66\u7684\u767a\u9054\u539f\u7406<\/h2>\n<ul>\n<li><strong>\u539f\u984c:<\/strong> Geometric developmental principles for the emergence of brain-like weighted and directed neuronal networks<\/li>\n<\/ul>\n<h3 class=\"wp-block-heading\">\u5c02\u9580\u30a2\u30ca\u30ea\u30b9\u30c8\u306e\u5206\u6790<\/h3>\n<p>\u672c\u7814\u7a76\u306f\u3001\u795e\u7d4c\u767a\u9054\u306b\u304a\u3051\u308b\u5e7e\u4f55\u5b66\u7684\u5236\u7d04\u304c\u3001\u9032\u5316\u7684\u8ddd\u96e2\u3092\u8d85\u3048\u3066\u89b3\u5bdf\u3055\u308c\u308b\u795e\u7d4c\u56de\u8def\u306e\u4fdd\u5b58\u3055\u308c\u305f\u69cb\u9020\u539f\u7406\u3092\u3069\u306e\u3088\u3046\u306b\u8aac\u660e\u3067\u304d\u308b\u304b\u3092\u8abf\u67fb\u3057\u3066\u3044\u307e\u3059\u3002\u5358\u4e00\u30cb\u30e5\u30fc\u30ed\u30f3\u89e3\u50cf\u5ea6\u306e\u30b3\u30cd\u30af\u30c8\u30fc\u30e0\u30925\u7a2e\uff08C. Elegans\u3001Platynereis\u3001Drosophila M.\u3001\u30bc\u30d6\u30e9\u30d5\u30a3\u30c3\u30b7\u30e5\u3001\u30de\u30a6\u30b9\uff09\u306b\u308f\u305f\u3063\u3066\u5206\u6790\u3057\u305f\u7d50\u679c\u3001\u8ddd\u96e2\u4f9d\u5b58\u6027\u63a5\u7d9a\u306e\u307f\u3067\u306f\u30b9\u30e2\u30fc\u30eb\u30ef\u30fc\u30eb\u30c9\u30cd\u30c3\u30c8\u30ef\u30fc\u30af\u306f\u751f\u6210\u3055\u308c\u308b\u3082\u306e\u306e\u3001\u91cd\u307f\u4ed8\u304d\u5206\u5e03\u306e\u91cd\u5c3e\u5206\u5e03\u306f\u751f\u6210\u3055\u308c\u306a\u3044\u3053\u3068\u304c\u793a\u3055\u308c\u307e\u3057\u305f\u3002\u30b7\u30ca\u30d7\u30b9\u304c\u795e\u7d4c\u7a81\u8d77\u306b\u6cbf\u3063\u3066\u7a7a\u9593\u7684\u306b\u30af\u30e9\u30b9\u30bf\u30fc\u5316\u3059\u308b\u3053\u3068\u306b\u8d77\u56e0\u3059\u308b\u91cd\u307f\u512a\u5148\u4ed8\u7740\u3092\u7d44\u307f\u8fbc\u3080\u3053\u3068\u3067\u3001\u30b9\u30e2\u30fc\u30eb\u30ef\u30fc\u30eb\u30c9\u30c8\u30dd\u30ed\u30b8\u30fc\u3092\u7dad\u6301\u3057\u306a\u304c\u3089\u91cd\u307f\u5206\u5e03\u306e\u91cd\u5c3e\u5206\u5e03\u3092\u518d\u73fe\u3057\u307e\u3057\u305f\u3002\u6a39\u72b6\u7a81\u8d77\u3068\u8ef8\u7d22\u306e\u5206\u5c90\u7bc4\u56f2\u306b\u95a2\u9023\u3059\u308b\u6b21\u6570\u512a\u5148\u4ed8\u7740\u3092\u52a0\u3048\u308b\u3053\u3068\u3067\u3001\u6b21\u6570\u5206\u5e03\u306e\u91cd\u5c3e\u5206\u5e03\u306e\u751f\u6210\u304c\u53ef\u80fd\u306b\u306a\u308a\u307e\u3057\u305f\u3002\u4f53\u7cfb\u7684\u306a\u30d1\u30e9\u30e1\u30fc\u30bf\u63a2\u7d22\u3092\u901a\u3058\u3066\u3001\u8ddd\u96e2\u4f9d\u5b58\u6027\u3001\u91cd\u307f\u512a\u5148\u4ed8\u7740\u3001\u6b21\u6570\u512a\u5148\u4ed8\u7740\u306e\u7d44\u307f\u5408\u308f\u305b\u304c\u3001\u7d4c\u9a13\u7684\u306a\u8133\u30cd\u30c3\u30c8\u30ef\u30fc\u30af\u306e\u5168\u3066\u306e\u7279\u6027\u3092\u518d\u73fe\u3059\u308b\u306e\u306b\u5341\u5206\u3067\u3042\u308b\u3053\u3068\u3092\u5b9f\u8a3c\u3057\u307e\u3057\u305f\u3002\u3053\u308c\u3089\u306e\u7d50\u679c\u306f\u3001\u795e\u7d4c\u767a\u9054\u4e2d\u306e\u6d3b\u52d5\u975e\u4f9d\u5b58\u7684\u306a\u5e7e\u4f55\u5b66\u7684\u5236\u7d04\u304c\u3001\u9032\u5316\u7684\u8ddd\u96e2\u3092\u8d85\u3048\u3066\u89b3\u5bdf\u3055\u308c\u308b\u4fdd\u5b58\u3055\u308c\u305f\u69cb\u9020\u539f\u7406\u3092\u8aac\u660e\u3067\u304d\u308b\u3053\u3068\u3092\u793a\u5506\u3057\u3066\u304a\u308a\u3001\u795e\u7d4c\u56de\u8def\u30a2\u30bb\u30f3\u30d6\u30ea\u3092\u652f\u914d\u3059\u308b\u666e\u904d\u7684\u306a\u30e1\u30ab\u30cb\u30ba\u30e0\u3092\u793a\u5506\u3057\u3066\u3044\u307e\u3059\u3002<\/p>\n<p>\ud83d\udc49 <strong><a href=\"https:\/\/arxiv.org\/abs\/2601.05021\" target=\"_blank\" rel=\"noopener\">arXiv \u3067\u8a18\u4e8b\u5168\u6587\u3092\u8aad\u3080<\/a><\/strong><\/p>\n<ul>\n<li><strong>\u8981\u70b9:<\/strong> Activity-independent geometric constraints during neural development, specifically distance dependence, weight-preferential attachment, and degree-preferential attachment, are sufficient to explain the emergence of conserved, brain-like network architectures across diverse species.<\/li>\n<li><strong>\u8457\u8005:<\/strong> Aitor Morales-Gregorio, Anno C. Kurth, Karol\u00edna Korvasov\u00e1<\/li>\n<\/ul>\n<blockquote class=\"wp-block-quote\">\n<p>This study investigates the geometric developmental principles for the emergence of brain-like weighted and directed neuronal networks. By analyzing single-neuron resolution connectomes across five species (C. Elegans, Platynereis, Drosophila M., zebrafish, and mouse), the research shows that distance-dependent connectivity alone produces small-world networks but fails to generate heavy-tailed weight distributions. Incorporating weight-preferential attachment, arising from spatial clustering of synapses along neurites, reproduces heavy-tailed weight distributions while maintaining small-world topology. Adding degree-preferential attachment, linked to the extent of dendritic and axonal arborization, enables the generation of heavy-tailed degree distributions. Through systematic parameter exploration, the combination of distance dependence, weight-preferential attachment, and degree-preferential attachment is demonstrated to be sufficient to reproduce all characteristic properties of empirical brain networks. These findings suggest that activity-independent geometric constraints during neural development can account for the conserved architectural principles observed across evolutionarily distant species, indicating universal mechanisms governing neural circuit assembly.<\/p>\n<\/blockquote>\n<\/div>\n","protected":false},"excerpt":{"rendered":"<p>LLM\u306e\u63a8\u8ad6\u84b8\u7559\u3001\u52d5\u7684\u306a\u6982\u5ff5\u30e2\u30c7\u30eb\u3001\u8133\u306e\u69cb\u9020\u539f\u7406\u306b\u95a2\u3059\u308b\u6700\u65b0\u306e\u7814\u7a76\u3092\u8981\u7d04\u3002AI\u306e\u9032\u5316\u3068\u8a8d\u77e5\u79d1\u5b66\u3078\u306e\u6d1e\u5bdf\u3092\u63d0\u4f9b\u3057\u307e\u3059\u3002<\/p>\n","protected":false},"author":1,"featured_media":854,"comment_status":"","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"vkexunit_cta_each_option":"","footnotes":""},"categories":[3],"tags":[8,64,61,62,15,63,65],"class_list":{"0":"post-874","1":"post","2":"type-post","3":"status-publish","4":"format-standard","5":"has-post-thumbnail","6":"hentry","7":"category-columns","8":"tag-ai","9":"tag-64","10":"tag-61","11":"tag-62","13":"tag-63","14":"tag-65"},"_links":{"self":[{"href":"https:\/\/itexplore.org\/jp\/wp-json\/wp\/v2\/posts\/874","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/itexplore.org\/jp\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/itexplore.org\/jp\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/itexplore.org\/jp\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/itexplore.org\/jp\/wp-json\/wp\/v2\/comments?post=874"}],"version-history":[{"count":0,"href":"https:\/\/itexplore.org\/jp\/wp-json\/wp\/v2\/posts\/874\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/itexplore.org\/jp\/wp-json\/wp\/v2\/media\/854"}],"wp:attachment":[{"href":"https:\/\/itexplore.org\/jp\/wp-json\/wp\/v2\/media?parent=874"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/itexplore.org\/jp\/wp-json\/wp\/v2\/categories?post=874"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/itexplore.org\/jp\/wp-json\/wp\/v2\/tags?post=874"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}