[{"data":1,"prerenderedAt":205},["ShallowReactive",2],{"DlFXI4Eibt_Bn9lrEZz1TYbHCWFZj3IvqwHQSEW-Exc":3,"wW4AxEnpc2Y0Q4D5wyx32EYBwqip4ThmVBG-nIjrgkk":194},{"code":4,"msg":5,"data":6},0,"",{"category":7,"tag":11,"hot":39,"new":78,"banner":118,"data":143,"cache":193},[8,9,10],"Agent","OpenAI","LLM",[12,14,17,20,23,25,27,30,33,36],{"title":8,"total":13},39,{"title":15,"total":16},"Google",44,{"title":18,"total":19},"Nvidia",13,{"title":21,"total":22},"Claude",11,{"title":9,"total":24},35,{"title":10,"total":26},85,{"title":28,"total":29},"DeepSeek",9,{"title":31,"total":32},"OCR",1,{"title":34,"total":35},"Chat",7,{"title":37,"total":38},"Generator",116,[40,48,55,64,71],{"id":41,"publish_date":42,"is_original":4,"collection":5,"cover_url":43,"cover_url_1_1":44,"title":45,"summary":46,"author":47},557,"2022-04-29","article_res/cover/7a9b1375ed9bb298154981bae42b794d.jpeg","article_res/cover/afa281dd52bc0454e6735daa8e6b0706.jpeg","Translation and summary of Messari Report [2.8 Kristin Smith, Blockchain Association and Katie Haun, a16z]","We need unity and speed right now.","Translation",{"id":49,"publish_date":50,"is_original":4,"collection":5,"cover_url":51,"cover_url_1_1":52,"title":53,"summary":54,"author":47},531,"2022-05-25","article_res/cover/e8362057f8fa189594c60afdfaaeb6e5.jpeg","article_res/cover/8ea08d0d6fa7eee6b57ed4ec61b61ad6.jpeg","Decentralized Society: Finding Web3’s Soul / Decentralized Society: Finding the Soul of Web3 -7","Decentralization through Pluralism When analyzing ecosystems, it's desirable to measure how decentralized it is.",{"id":56,"publish_date":57,"is_original":32,"collection":58,"cover_url":59,"cover_url_1_1":60,"title":61,"summary":62,"author":63},127,"2024-11-14","#Google #AI Game #World Model #AI Story","article_res/cover/0233a875b7ec2debf59779e311547569.jpeg","article_res/cover/6ffddb6ae4914b3c699493311aa9f198.jpeg","Google Launches \"Unbounded\": A Generative Infinite Character Life Simulation Game","Unbounded: A Generative Infinite Game of Character Life Simulation","Renee's Entrepreneurial Journey",{"id":13,"publish_date":65,"is_original":32,"collection":66,"cover_url":67,"cover_url_1_1":68,"title":69,"summary":70,"author":63},"2025-02-14","#Deep Dive into LLMs #Andrej Karpathy #LLM #Tool Use #Hallucination","article_res/cover/11e858ad6b74dfa80f923d549b62855c.jpeg","article_res/cover/615e1b320f1fc163edc1d2d154a6de33.jpeg","Andrej Karpathy's in-depth explanation of LLM (Part 4): Hallucinations","hallucinations, tool use, knowledge/working memory",{"id":72,"publish_date":73,"is_original":4,"collection":5,"cover_url":74,"cover_url_1_1":75,"title":76,"summary":77,"author":47},579,"2022-04-07","article_res/cover/39387376ba28447af1eb40576b9df215.jpeg","article_res/cover/02727ede8551ed49901d0abe6d6305b7.jpeg","Messari Report Translation and Summary 【1-7 Surviving the Winter】","I’d be more cautious here: 10 year and 10 hour thinking only.",[79,87,95,103,111],{"id":80,"publish_date":81,"is_original":32,"collection":82,"cover_url":83,"cover_url_1_1":84,"title":85,"summary":86,"author":63},627,"2025-03-20","#AI Avatar #AI Video Generation","article_res/cover/d95481358f73924989f8c4ee9c75d1c8.jpeg","article_res/cover/b74bc0fab01f8b6a6aa87696c0c3ed8b.jpeg","DisPose: Generating Animated Videos by Driving Video with Reference Images","DisPose is a controllable human image animation method that enhances video generation.",{"id":88,"publish_date":89,"is_original":32,"collection":90,"cover_url":91,"cover_url_1_1":92,"title":93,"summary":94,"author":63},626,"2025-03-21","#Deep Dive into LLMs #LLM #RL #Andrej Karpathy #AlphaGo","article_res/cover/446553a5c8f8f2f07d97b20eaee84e56.jpeg","article_res/cover/e6c2823409c9b34624064b9acbaca6f1.jpeg","AlphaGo and the Power of Reinforcement Learning - Andrej Karpathy's Deep Dive on LLMs (Part 9)","Simply learning from humans will never surpass human capabilities.",{"id":96,"publish_date":97,"is_original":32,"collection":98,"cover_url":99,"cover_url_1_1":100,"title":101,"summary":102,"author":63},625,"2025-03-22","#Deep Dive into LLMs #LLM #RL #RLHF #Andrej Karpathy","article_res/cover/8da81d38b1e5cf558a164710fd8a5389.jpeg","article_res/cover/96f028d76c362a99a0dd56389e8f7a9b.jpeg","Reinforcement Learning from Human Feedback (RLHF) - Andrej Karpathy's Deep Dive on LLMs (Part 10)","Fine-Tuning Language Models from Human Preferences",{"id":104,"publish_date":105,"is_original":32,"collection":106,"cover_url":107,"cover_url_1_1":108,"title":109,"summary":110,"author":63},624,"2025-03-23","#Deep Dive into LLMs #LLM #Andrej Karpathy #AI Agent #MMM","article_res/cover/a5e7c3d48bb09109684d6513287c661d.jpeg","article_res/cover/d3f22b7c0ab8d82fd2da457a299e0773.jpeg","The Future of Large Language Models - Andrej Karpathy's In-Depth Explanation of LLM (Part 11)","preview of things to come",{"id":112,"publish_date":105,"is_original":32,"collection":113,"cover_url":114,"cover_url_1_1":115,"title":116,"summary":117,"author":63},623,"#Google #Voe #AI Video Generation","article_res/cover/c44062fea0f336c2b96b3928292392c2.jpeg","article_res/cover/a041041c69092ad3db191c5bf3ff981b.jpeg","Trial of Google's video generation model VOE2","Our state-of-the-art video generation model",[119,127,135],{"id":120,"publish_date":121,"is_original":32,"collection":122,"cover_url":123,"cover_url_1_1":124,"title":125,"summary":126,"author":63},160,"2024-10-04","#Philosophy","article_res/cover/496990c49211e8b7f996b7d39c18168e.jpeg","article_res/cover/14dbaa1ade9cb4316d5829423a900362.jpeg","Time","The fungus of the morning does not know the waxing and waning of the moon, and the cicada does not know the seasons; this is a short life. To the south of the state of Chu there is a dark spirit which regards five hundred years as spring and five hundred years as autumn. In ancient times there was a great tree called the Ming which regarded eight thousand years as spring and eight thousand years as autumn; this is a long life.",{"id":128,"publish_date":129,"is_original":32,"collection":130,"cover_url":131,"cover_url_1_1":132,"title":133,"summary":134,"author":63},98,"2024-12-17","#AI Video Generator #Sora #Pika","article_res/cover/3b86e85d03fff4f356a3e4cf2bb329c9.jpeg","article_res/cover/5fa5c20ad0b40f8f544d257c0ef02938.jpeg","Pika 2.0 video generation officially released: effect comparison with Sora","今天，我们推出了Pika 2.0模型。卓越的文字对齐效果。惊人的视觉表现。还有✨场景成分✨",{"id":136,"publish_date":137,"is_original":32,"collection":138,"cover_url":139,"cover_url_1_1":140,"title":141,"summary":142,"author":63},71,"2025-01-14","#Nvidia #World Foundation Model #Cosmos #Physical AI #Embodied AI","article_res/cover/feddf8c832dfb45d28804291f6a42a9e.jpeg","article_res/cover/d6bc2f1186d96b78228c2283a17a3645.jpeg","NVIDIA's Cosmos World Model","Cosmos World Foundation Model Platform for Physical AI",[144,163,188],{"title":8,"items":145},[146,147,155],{"id":104,"publish_date":105,"is_original":32,"collection":106,"cover_url":107,"cover_url_1_1":108,"title":109,"summary":110,"author":63},{"id":148,"publish_date":149,"is_original":32,"collection":150,"cover_url":151,"cover_url_1_1":152,"title":153,"summary":154,"author":63},622,"2025-03-24","#OWL #AI Agent #MAS #MCP #CUA","article_res/cover/cb50ca7f2bf4d1ed50202d7406e1c19a.jpeg","article_res/cover/4aa7aa3badfacf3cc84121334f1050dd.jpeg","OWL: Multi-agent collaboration","OWL: Optimized Workforce Learning for General Multi-Agent Assistance in Real-World Task Automation",{"id":156,"publish_date":157,"is_original":32,"collection":158,"cover_url":159,"cover_url_1_1":160,"title":161,"summary":162,"author":63},620,"2025-03-26","#LLM #Google #Gemini #AI Agent","article_res/cover/53751a6dbbe990b1eb0b63f3b062aed4.jpeg","article_res/cover/031344981f0a212ff82d1f3a64aa5756.jpeg","Gemini 2.5 Pro, claimed to be far ahead of the competition, has been released with great fanfare: comprehensively surpassing other LLMs and topping the global rankings","Gemini 2.5: Our most intelligent AI model",{"title":9,"items":164},[165,172,180],{"id":166,"publish_date":157,"is_original":32,"collection":167,"cover_url":168,"cover_url_1_1":169,"title":170,"summary":171,"author":63},619,"#OpenAI #AI Image Generator #4o #MMM #AR Transformer","article_res/cover/2faffc97fcecf3151552cb0fd3206d89.jpeg","article_res/cover/1133cb4948af44cee2e7fbe79efb69e5.jpeg","The native image function of GPT-4o is officially launched","Introducing 4o Image Generation",{"id":173,"publish_date":174,"is_original":4,"collection":175,"cover_url":176,"cover_url_1_1":177,"title":178,"summary":179,"author":63},434,"2023-07-15","#Anthropic #OpenAI #Google #AI Code Generator #Claude","article_res/cover/e1b6f600a2b9f262a4392684e5f2ce25.jpeg","article_res/cover/6e1772e83f78f9a351ab23d3e414adee.jpeg","Latest Updates on Google Bard /Anthropic Claude2 / ChatGPT Code Interpreter","We want our models to use their programming skills to provide more natural interfaces to the basic functions of our computers.  \n - OpenAI",{"id":181,"publish_date":182,"is_original":4,"collection":183,"cover_url":184,"cover_url_1_1":185,"title":186,"summary":187,"author":63},417,"2023-08-24","#OpenAI","article_res/cover/bccf897d50a88b18364e35f7466387e0.jpeg","article_res/cover/2f871085c1073717c1703ae86e18056f.jpeg","The GPT-3.5 Turbo fine-tuning (fine-tuning function) has been released～","Developers can now bring their own data to customize GPT-3.5 Turbo for their use cases.",{"title":10,"items":189},[190,191,192],{"id":88,"publish_date":89,"is_original":32,"collection":90,"cover_url":91,"cover_url_1_1":92,"title":93,"summary":94,"author":63},{"id":96,"publish_date":97,"is_original":32,"collection":98,"cover_url":99,"cover_url_1_1":100,"title":101,"summary":102,"author":63},{"id":104,"publish_date":105,"is_original":32,"collection":106,"cover_url":107,"cover_url_1_1":108,"title":109,"summary":110,"author":63},true,{"code":4,"msg":5,"data":195},{"id":196,"publish_date":197,"is_original":32,"collection":198,"articles_id":199,"cover_url":200,"cover_url_1_1":201,"title":202,"summary":203,"author":63,"content":204},206,"2024-08-02","#Neuroscience #Nature","XFmssWUesmz7vjS1R1gGGg","article_res/cover/0ac6998a6dd9e1acef02bffd48730b76.jpeg","article_res/cover/75eb9c3da162a166c7429ba78d31e1d8.jpeg","【Nature article】Is language for communication or for thought?","The limits of my language mean the limits of my world.  \n- Ludwig Wittgenstein","\u003Cdiv class=\"rich_media_content js_underline_content\n                       autoTypeSetting24psection\n            \" id=\"js_content\">\u003Cp style='margin-bottom: 0px;;color: rgb(0, 0, 0);font-size: 16px;line-height: 1.8em;letter-spacing: normal;text-align: left;padding-top: 8px;padding-bottom: 8px;font-family: Optima, \"Microsoft YaHei\", PingFangSC-regular, serif;text-wrap: wrap;background-color: rgb(255, 255, 255);'>Ludwig Wittgenstein mentioned in \"Tractatus Logico-Philosophicus\" that the limits of my language mean the limits of my world. But I recently read an article titled \"Language is primarily a tool for communication rather than thought\", which seems to take a different stance. It was shared by a friend in a group, and later I got the full text from my idol, Professor Gou.\u003C/p>\u003Cp style='margin-bottom: 0px;;color: rgb(0, 0, 0);font-size: 16px;line-height: 1.8em;letter-spacing: normal;text-align: left;padding-top: 8px;padding-bottom: 8px;font-family: Optima, \"Microsoft YaHei\", PingFangSC-regular, serif;text-wrap: wrap;background-color: rgb(255, 255, 255);'>\u003Cstrong>【TL; DR】Language is an important characteristic of humans, but its function has been controversial over centuries: the mainstream view holds that language is used for thinking, whereas this article argues the opposite, suggesting that language is a tool for communication rather than thought.\u003C/strong>\u003C/p>\u003Ch2 style='margin-top: 30px;margin-bottom: 15px;color: rgba(0, 0, 0, 0.85);;font-family: Optima, \"Microsoft YaHei\", PingFangSC-regular, serif;letter-spacing: normal;text-align: left;text-wrap: wrap;background-color: rgb(255, 255, 255);'>\u003Cspan style=\";font-size: 22px;color: rgb(0, 0, 0);line-height: 1.5em;letter-spacing: 0em;font-weight: bold;display: block;\">Background\u003C/span>\u003C/h2>\u003Cp style='margin-bottom: 0px;;color: rgb(0, 0, 0);font-size: 16px;line-height: 1.8em;letter-spacing: normal;text-align: left;padding-top: 8px;padding-bottom: 8px;font-family: Optima, \"Microsoft YaHei\", PingFangSC-regular, serif;text-wrap: wrap;background-color: rgb(255, 255, 255);'>It is believed that language emerged in humans approximately between 100,000 and 1,000,000 years ago.\u003C/p>\u003Cp style='margin-bottom: 0px;;color: rgb(0, 0, 0);font-size: 16px;line-height: 1.8em;letter-spacing: normal;text-align: left;padding-top: 8px;padding-bottom: 8px;font-family: Optima, \"Microsoft YaHei\", PingFangSC-regular, serif;text-wrap: wrap;background-color: rgb(255, 255, 255);'>There are currently two main hypotheses:\u003C/p>\u003Cul style='margin-top: 8px;margin-bottom: 8px;;padding-left: 25px;color: rgb(0, 0, 0);font-family: Optima, \"Microsoft YaHei\", PingFangSC-regular, serif;font-size: 16px;letter-spacing: normal;text-align: left;text-wrap: wrap;background-color: rgb(255, 255, 255);' class=\"list-paddingleft-1\">\u003Cli style=\";\">\u003Csection style=\";margin-top: 5px;margin-bottom: 5px;color: rgb(1, 1, 1);line-height: 1.8em;letter-spacing: 0em;\">Language primarily has a communicative function, enabling us to share knowledge, thoughts, and emotions with each other.\u003C/section>\u003C/li>\u003Cli style=\";\">\u003Csection style=\";margin-top: 5px;margin-bottom: 5px;color: rgb(1, 1, 1);line-height: 1.8em;letter-spacing: 0em;\">Language modulates thought and cognition. The range of specific hypotheses about the role of language in thought is wide, from strongly arguing that language is necessary for all forms of (at least propositional) thought, to considering that language may be crucial or facilitative only for certain aspects of thought and reasoning, to believing that language helps build certain learning frameworks during development but may no longer be needed in the mature brain.\u003C/section>\u003C/li>\u003C/ul>\u003Cp style='margin-bottom: 0px;;color: rgb(0, 0, 0);font-size: 16px;line-height: 1.8em;letter-spacing: normal;text-align: left;padding-top: 8px;padding-bottom: 8px;font-family: Optima, \"Microsoft YaHei\", PingFangSC-regular, serif;text-wrap: wrap;background-color: rgb(255, 255, 255);'>From an evolutionary adaptability perspective, both the communicative and cognitive functions of language might provide adaptive advantages. The ability to transmit information accurately could promote cooperative behaviors such as hunting, foraging, and long-distance travel, and allow the transmission of knowledge and skills to offspring (cultural transmission). Enhanced reasoning abilities might make more complex planning and decision-making, better tool-making, and problem-solving capabilities possible.\u003C/p>\u003Cp style='margin-bottom: 0px;;color: rgb(0, 0, 0);font-size: 16px;line-height: 1.8em;letter-spacing: normal;text-align: left;padding-top: 8px;padding-bottom: 8px;font-family: Optima, \"Microsoft YaHei\", PingFangSC-regular, serif;text-wrap: wrap;background-color: rgb(255, 255, 255);'>The hypotheses \"language is used for thought\" and \"language is used for communication\" make testable predictions about human cognition and neural architecture, as well as the nature of human language. Is there some form of thought—namely our knowledge of the world and our ability to reason based on these representations—that requires language (i.e., the representations and computations that support our generation and interpretation of meaningful sequences of words)? If some forms of thought require language, then at least in these types of thought and reasoning, the language mechanism should be forced to participate, and these types of thought should be impossible without language. If language is a tool for communication, then language should exhibit characteristics of efficient information transfer. Until recently, these predictions were difficult to evaluate.\u003C/p>\u003Cp style='margin-bottom: 0px;;color: rgb(0, 0, 0);font-size: 16px;line-height: 1.8em;letter-spacing: normal;text-align: left;padding-top: 8px;padding-bottom: 8px;font-family: Optima, \"Microsoft YaHei\", PingFangSC-regular, serif;text-wrap: wrap;background-color: rgb(255, 255, 255);'>However, advances in knowledge and tools over the past two decades have provided key insights into the functions of language:\u003C/p>\u003Cul style='margin-top: 8px;margin-bottom: 8px;;padding-left: 25px;color: rgb(0, 0, 0);font-family: Optima, \"Microsoft YaHei\", PingFangSC-regular, serif;font-size: 16px;letter-spacing: normal;text-align: left;text-wrap: wrap;background-color: rgb(255, 255, 255);' class=\"list-paddingleft-1\">\u003Cli style=\";\">\u003Csection style=\";margin-top: 5px;margin-bottom: 5px;color: rgb(1, 1, 1);line-height: 1.8em;letter-spacing: 0em;\">Substantial progress has been made in decoding the neural architecture of language, providing a clear \"target\" for evaluating the involvement of language processing mechanisms in various forms of thought.\u003C/section>\u003C/li>\u003Cli style=\";\">\u003Csection style=\";margin-top: 5px;margin-bottom: 5px;color: rgb(1, 1, 1);line-height: 1.8em;letter-spacing: 0em;\">Large-scale diverse language corpora have become widely available, accompanied by a set of powerful computational tools, often based on information theory, used to rigorously characterize language systems.\u003C/section>\u003C/li>\u003C/ul>\u003Cp style='margin-bottom: 0px;;color: rgb(0, 0, 0);font-size: 16px;line-height: 1.8em;letter-spacing: normal;text-align: left;padding-top: 8px;padding-bottom: 8px;font-family: Optima, \"Microsoft YaHei\", PingFangSC-regular, serif;text-wrap: wrap;background-color: rgb(255, 255, 255);'>Therefore, the authors argue that it is time to summarize the evidence regarding key issues about the function of language and its role in human cognition. 😂\u003C/p>\u003Ch2 style='margin-top: 30px;margin-bottom: 15px;color: rgba(0, 0, 0, 0.85);;font-family: Optima, \"Microsoft YaHei\", PingFangSC-regular, serif;letter-spacing: normal;text-align: left;text-wrap: wrap;background-color: rgb(255, 255, 255);'>\u003Cspan style=\";font-size: 22px;color: rgb(0, 0, 0);line-height: 1.5em;letter-spacing: 0em;font-weight: bold;display: block;\">The language network in the human brain\u003C/span>\u003C/h2>\u003Cp style='margin-bottom: 0px;;color: rgb(0, 0, 0);font-size: 16px;line-height: 1.8em;letter-spacing: normal;text-align: left;padding-top: 8px;padding-bottom: 8px;font-family: Optima, \"Microsoft YaHei\", PingFangSC-regular, serif;text-wrap: wrap;background-color: rgb(255, 255, 255);'>Our knowledge of language includes regularity knowledge at all levels of linguistic structure, from sounds to sentences, as well as a vast number of form-meaning mappings (the meanings of morphemes, words, and structures). Using this knowledge, we can both convey our thoughts to others and infer their intentions from their utterances. Language production and comprehension are supported by a set of interconnected brain regions in the left hemisphere, which are commonly referred to as the \"language network,\" as shown in the figure below:\u003C/p>\u003Csection style=';margin-top: 20px;margin-bottom: 20px;padding: 10px 20px;border-style: none;border-width: 3px;border-color: rgba(0, 0, 0, 0.4);border-radius: 0px;background: none 0% 0% / auto no-repeat scroll padding-box border-box rgb(250, 250, 250);width: auto;height: auto;box-shadow: rgba(0, 0, 0, 0) 0px 0px 0px 0px;color: rgb(0, 0, 0);font-family: Optima, \"Microsoft YaHei\", PingFangSC-regular, serif;font-size: 16px;letter-spacing: normal;text-align: left;text-wrap: wrap;'>\u003Csection style=\";\">\u003Cp style=\";line-height: 1.8em;letter-spacing: 0em;text-indent: 0em;padding-top: 8px;padding-bottom: 8px;\">\u003Cimg class=\"rich_pages wxw-img\" data-galleryid=\"\" data-imgfileid=\"100006152\" data-ratio=\"0.7619047619047619\" data-s=\"300,640\" data-type=\"png\" data-w=\"357\" style=\"text-align: center;letter-spacing: 0em;text-indent: 0em;\" src=\"https://res.cooltool.vip/article_res/assets/17423802979910.2348368080309955.png\">\u003C/p>\u003Csection style=\"text-align: center;margin-bottom: 24px;\">\u003Cimg class=\"rich_pages wxw-img\" data-galleryid=\"\" data-imgfileid=\"100006153\" data-ratio=\"0.6785009861932939\" data-s=\"300,640\" data-type=\"png\" data-w=\"507\" style=\"\" src=\"https://res.cooltool.vip/article_res/assets/17423802979930.4497286820141795.png\">\u003C/section>\u003Cp style=\"text-align: left;\">\u003Cstrong style=\"letter-spacing: 0em;text-indent: 0em;background: none 0% 0% / auto no-repeat scroll padding-box border-box rgba(0, 0, 0, 0);text-align: left;;width: auto;height: auto;border-style: none;border-width: 3px;border-color: rgba(0, 0, 0, 0.4);border-radius: 0px;\">Description\u003C/strong>\u003Cspan style=\"letter-spacing: 0em;text-indent: 0em;text-align: left;\">：\u003C/span>\u003Cspan style=\"letter-spacing: 0em;text-indent: 0em;text-align: left;\">The language network in the human brain.\u003C/span>\u003Cspan style=\"letter-spacing: 0em;text-indent: 0em;text-align: left;\">The language network supports computations for lexical access and syntactic structure building in both comprehension and production, across multiple modalities, suggesting that the representations operated on by the network are abstract.\u003C/span>\u003Cspan style=\"letter-spacing: 0em;text-indent: 0em;text-align: left;\">Given that acquiring words and syntax is thought to be critical for certain aspects of thought, this brain network is a clear target for evaluating the hypothesis that language is used for thought.\u003C/span>\u003Cspan style=\"letter-spacing: 0em;text-indent: 0em;text-align: left;\">The top left shows the language network in five sample individuals.\u003C/span>\u003Cspan style=\"letter-spacing: 0em;text-indent: 0em;text-align: left;\">These activation maps were obtained using fMRI with a language localizer paradigm, which contrasts language processing with perceptually similar control conditions.\u003C/span>\u003Cspan style=\"letter-spacing: 0em;text-indent: 0em;text-align: left;\">The brain regions in these maps show greater neural activity under the critical language-processing condition compared to the control condition.\u003C/span>\u003Cspan style=\"letter-spacing: 0em;text-indent: 0em;text-align: left;\">In the lower left is a schematic of the response profile of the language network (e.g., as measured by fMRI) to sentences for comprehension or production, lists of unrelated words, and lists of non-words.\u003C/span>\u003Cspan style=\"letter-spacing: 0em;text-indent: 0em;text-align: left;\">A stronger response to sentences is generally taken to indicate involvement of a brain region in combinatorial (syntactic and semantic) computations that are required for processing sentences but not word lists;\u003C/span>\u003Cspan style=\"letter-spacing: 0em;text-indent: 0em;text-align: left;\">a stronger response to word lists indicates involvement in lexical-semantic access, which is required for processing real words but not non-words.\u003C/span>\u003Cspan style=\"letter-spacing: 0em;text-indent: 0em;text-align: left;\">On the right are sample stimuli used in brain-imaging experiments to study responses to sentences, word lists, and nonword lists during comprehension (top row) and production (bottom row).\u003C/span>\u003C/p>\u003C/section>\u003C/section>\u003Csection style=';margin-top: 20px;margin-bottom: 20px;padding: 10px 20px;border-style: none;border-width: 3px;border-color: rgba(0, 0, 0, 0.4);border-radius: 0px;background: none 0% 0% / auto no-repeat scroll padding-box border-box rgb(250, 250, 250);width: auto;height: auto;box-shadow: rgba(0, 0, 0, 0) 0px 0px 0px 0px;color: rgb(0, 0, 0);font-family: Optima, \"Microsoft YaHei\", PingFangSC-regular, serif;font-size: 16px;letter-spacing: normal;text-align: left;text-wrap: wrap;'>\u003Csection style=\";\">\u003Cp style=\";line-height: 1.8em;letter-spacing: 0em;text-indent: 0em;padding-top: 8px;padding-bottom: 8px;\">\u003Cimg class=\"rich_pages wxw-img\" data-galleryid=\"\" data-imgfileid=\"100006154\" data-ratio=\"0.4143222506393862\" data-s=\"300,640\" data-type=\"png\" data-w=\"782\" style=\"text-align: center;letter-spacing: 0em;text-indent: 0em;\" src=\"https://res.cooltool.vip/article_res/assets/17423802979930.46131752769468104.png\">\u003C/p>\u003Cp style=\";line-height: 1.8em;letter-spacing: 0em;text-indent: 0em;padding-top: 8px;padding-bottom: 8px;\">: Classical and current models:\u003C/p>\u003Col class=\"list-paddingleft-1\" style=\"list-style-type: lower-alpha;\">\u003Cli>\u003Cp style=\";line-height: 1.8em;letter-spacing: 0em;text-indent: 0em;padding-top: 8px;padding-bottom: 8px;\">The classical model of language neurobiology\u003C/p>\u003C/li>\u003Cli>\u003Cp style=\";line-height: 1.8em;letter-spacing: 0em;text-indent: 0em;padding-top: 8px;padding-bottom: 8px;\">A model based on current knowledge of language neurobiology\u003C/p>\u003C/li>\u003C/ol>\u003Cp style=\";line-height: 1.8em;letter-spacing: 0em;text-indent: 0em;padding-top: 8px;padding-bottom: 8px;\">This updated model still includes Broca's area (articulatory planning) and Wernicke's area (language perception), but also encompasses a set of frontal and temporal regions that jointly support advanced language comprehension and production. To provide context, we also show the primary auditory cortex, which may supply input to Wernicke's area (language perception), as well as the sensorimotor cortex, which may receive input from Broca's area (articulatory planning).\u003C/p>\u003C/section>\u003C/section>\u003Cp style='margin-bottom: 0px;;color: rgb(0, 0, 0);font-size: 16px;line-height: 1.8em;letter-spacing: normal;text-align: left;padding-top: 8px;padding-bottom: 8px;font-family: Optima, \"Microsoft YaHei\", PingFangSC-regular, serif;text-wrap: wrap;background-color: rgb(255, 255, 255);'>Two characteristics of the language network are particularly important for understanding the functions of language.\u003C/p>\u003Cul style='margin-top: 8px;margin-bottom: 8px;;padding-left: 25px;color: rgb(0, 0, 0);font-family: Optima, \"Microsoft YaHei\", PingFangSC-regular, serif;font-size: 16px;letter-spacing: normal;text-align: left;text-wrap: wrap;background-color: rgb(255, 255, 255);' class=\"list-paddingleft-1\">\u003Cli style=\";\">\u003Csection style=\";margin-top: 5px;margin-bottom: 5px;color: rgb(1, 1, 1);line-height: 1.8em;letter-spacing: 0em;\">Language areas exhibit independence in their input and output patterns—a key marker of abstraction. During comprehension, these brain regions respond to various modalities of linguistic input (spoken, written, or signed languages). Similarly, during language production, these areas are active whether we convey information through speaking or writing. These findings suggest that these areas support both language comprehension and production, implying that they might store our linguistic knowledge, which is needed to encode and decode linguistic information. The abstract nature of the representations in the language network aligns with the \"narrow language faculty\" or \"abstract computational system\" described in an influential article on language evolution by Hauser, Chomsky, and Fitch, distinct from low-level mechanisms of phonetic perception and articulation.\u003C/section>\u003C/li>\u003Cli style=\";\">\u003Csection style=\";margin-top: 5px;margin-bottom: 5px;color: rgb(1, 1, 1);line-height: 1.8em;letter-spacing: 0em;\">Language areas represent and process word meanings and syntactic structures—key components of the hypothesis that language is used for thought. Specifically, evidence from multiple experiments and natural paradigms using functional MRI (fMRI), magnetoencephalography, and intracranial recordings shows that all regions of the language network are sensitive to word meanings as well as syntactic and semantic dependencies between words.\u003C/section>\u003C/li>\u003C/ul>\u003Cp style='margin-bottom: 0px;;color: rgb(0, 0, 0);font-size: 16px;line-height: 1.8em;letter-spacing: normal;text-align: left;padding-top: 8px;padding-bottom: 8px;font-family: Optima, \"Microsoft YaHei\", PingFangSC-regular, serif;text-wrap: wrap;background-color: rgb(255, 255, 255);'>In summary, the abstract nature of language representation in the language network, along with its sensitivity to meaning and structure, makes it a clear target for evaluating hypotheses about the role of language in thought and cognition.\u003C/p>\u003Ch2 style='margin-top: 30px;margin-bottom: 15px;color: rgba(0, 0, 0, 0.85);;font-family: Optima, \"Microsoft YaHei\", PingFangSC-regular, serif;letter-spacing: normal;text-align: left;text-wrap: wrap;background-color: rgb(255, 255, 255);'>\u003Cspan style=\";font-size: 22px;color: rgb(0, 0, 0);line-height: 1.5em;letter-spacing: 0em;font-weight: bold;display: block;\">Language is neither necessary nor sufficient for thought.\u003C/span>\u003C/h2>\u003Cp style='margin-bottom: 0px;;color: rgb(0, 0, 0);font-size: 16px;line-height: 1.8em;letter-spacing: normal;text-align: left;padding-top: 8px;padding-bottom: 8px;font-family: Optima, \"Microsoft YaHei\", PingFangSC-regular, serif;text-wrap: wrap;background-color: rgb(255, 255, 255);'>The ontology of human cognition and the nature of psychological representations that mediate thought remain active areas of research. Broadly speaking, thought includes our knowledge of the world (including domain-specific knowledge such as the physical properties of objects or knowledge about social agents) and reasoning over these representations, which involves making inferences and predictions. In addition to reasoning within specific domains of knowledge, reasoning can involve integrating information across domains (a key component of analogical reasoning) or be domain-general, abstract reasoning that is not tied to any particular domain. Domain-general reasoning is often associated with the concept of fluid intelligence. Moreover, empirically, all aspects of thought tested so far exhibit similar behavior in terms of their use of linguistic resources.\u003C/p>\u003Cp style='margin-bottom: 0px;;color: rgb(0, 0, 0);font-size: 16px;line-height: 1.8em;letter-spacing: normal;text-align: left;padding-top: 8px;padding-bottom: 8px;font-family: Optima, \"Microsoft YaHei\", PingFangSC-regular, serif;text-wrap: wrap;background-color: rgb(255, 255, 255);'>If language modulates certain forms of thought, then those forms of thought should not be possible without language because they should critically depend on linguistic representations (i.e., the \"necessity\" argument for language in thought). Furthermore, the presence of language (or full linguistic competence) should correlate with the ability to engage in these forms of thought (i.e., the \"sufficiency\" argument for language in thought). Below, we will address these claims in turn:\u003C/p>\u003Ch3 style='margin-top: 30px;margin-bottom: 15px;color: rgba(0, 0, 0, 0.85);;font-family: Optima, \"Microsoft YaHei\", PingFangSC-regular, serif;letter-spacing: normal;text-align: left;text-wrap: wrap;background-color: rgb(255, 255, 255);'>\u003Cspan style=\";font-size: 20px;color: rgb(0, 0, 0);line-height: 1.5em;letter-spacing: 0em;font-weight: bold;display: block;\">Language is not necessary for any form of thought tested.\u003C/span>\u003C/h3>\u003Cp style='margin-bottom: 0px;;color: rgb(0, 0, 0);font-size: 16px;line-height: 1.8em;letter-spacing: normal;text-align: left;padding-top: 8px;padding-bottom: 8px;font-family: Optima, \"Microsoft YaHei\", PingFangSC-regular, serif;text-wrap: wrap;background-color: rgb(255, 255, 255);'>A classic approach to inferring brain-behavior associations and dissociations is through the study of individuals with brain damage or disease. If linguistic abilities modulate our capacity to perform certain forms of thought, then language impairments should co-occur with difficulties in these forms of thought and reasoning. However, the evidence is clear — there are many individuals with severe language impairments, affecting both their lexicon and syntax, who nonetheless demonstrate intact abilities across a variety of types of thought. They can solve mathematical problems, make executive plans and follow non-verbal instructions, engage in various forms of reasoning, including formal logical reasoning, causal reasoning about the world, and scientific reasoning, understand others' beliefs or thoughts and engage in pragmatic reasoning, navigate the world, and make semantic judgments about objects and events. See the figure below:\u003C/p>\u003Csection style=';margin-top: 20px;margin-bottom: 20px;padding: 10px 20px;border-style: none;border-width: 3px;border-color: rgba(0, 0, 0, 0.4);border-radius: 0px;background: none 0% 0% / auto no-repeat scroll padding-box border-box rgb(250, 250, 250);width: auto;height: auto;box-shadow: rgba(0, 0, 0, 0) 0px 0px 0px 0px;color: rgb(0, 0, 0);font-family: Optima, \"Microsoft YaHei\", PingFangSC-regular, serif;font-size: 16px;letter-spacing: normal;text-align: left;text-wrap: wrap;'>\u003Csection style=\";\">\u003Cp style=\";line-height: 1.8em;letter-spacing: 0em;text-indent: 0em;padding-top: 8px;padding-bottom: 8px;\">\u003Cimg class=\"rich_pages wxw-img\" data-galleryid=\"\" data-imgfileid=\"100006155\" data-ratio=\"0.45925925925925926\" data-s=\"300,640\" data-type=\"png\" data-w=\"810\" style=\"text-align: center;\" src=\"https://res.cooltool.vip/article_res/assets/17423802979920.8186973069667163.png\">\u003Cbr>\u003C/p>\u003Cp style=\";line-height: 1.8em;letter-spacing: 0em;text-indent: 0em;padding-top: 8px;padding-bottom: 8px;\">：The separation of language and thought in the brain. On the left is a schematic representation of the response characteristics of the language network (e.g., measured by fMRI). This network reacts strongly to language comprehension and production but does not respond to non-linguistic tasks that require thought and reasoning. The core regions of the language network are represented in red on the brain template.\u003C/p>\u003Cp style=\";line-height: 1.8em;letter-spacing: 0em;text-indent: 0em;padding-top: 8px;padding-bottom: 8px;\">On the right is a schematic representation of the response characteristics of two networks that support thought and reasoning (e.g., measured by fMRI). The multiple-demand network (represented in blue on the brain template) responds to various cognitively demanding tasks, including executive function tasks, novel problem-solving tasks, and mathematical and logical reasoning, but does not react to language or social reasoning. The theory-of-mind network (represented in green on the brain template) responds during social reasoning but does not react to language or demanding executive function tasks.\u003C/p>\u003C/section>\u003C/section>\u003Ch3 style='margin-top: 30px;margin-bottom: 15px;color: rgba(0, 0, 0, 0.85);;font-family: Optima, \"Microsoft YaHei\", PingFangSC-regular, serif;letter-spacing: normal;text-align: left;text-wrap: wrap;background-color: rgb(255, 255, 255);'>\u003Cspan style=\";font-size: 20px;color: rgb(0, 0, 0);line-height: 1.5em;letter-spacing: 0em;font-weight: bold;display: block;\">Full language ability does not imply full thinking ability.\u003C/span>\u003C/h3>\u003Cp style='margin-bottom: 0px;;color: rgb(0, 0, 0);font-size: 16px;line-height: 1.8em;letter-spacing: normal;text-align: left;padding-top: 8px;padding-bottom: 8px;font-family: Optima, \"Microsoft YaHei\", PingFangSC-regular, serif;text-wrap: wrap;background-color: rgb(255, 255, 255);'>In these studies, researchers carefully and systematically evaluated individuals' language abilities to ensure that the deficits were indeed related to language skills, including lexical and syntactic processing, rather than low-level perceptual or motor skills. These findings challenge the common view about the importance of language for thought, as well as several specific assumptions regarding the critical role of language in certain types of thinking, including mathematical reasoning, cross-domain information integration, and categorization. Despite losing their language abilities, some severely aphasic patients were still able to perform all tested forms of thought and reasoning, as evidenced by their intact performance across various cognitive tasks. They simply could not map these thoughts onto linguistic expressions—neither could they convey their thoughts through language nor extract meaning from others' speech and sentences. Of course, in some cases of brain injury, both language and (certain) thinking abilities may be affected, but this is expected because the language system is adjacent to other higher cognitive systems.\u003C/p>\u003Cp style='margin-bottom: 0px;;color: rgb(0, 0, 0);font-size: 16px;line-height: 1.8em;letter-spacing: normal;text-align: left;padding-top: 8px;padding-bottom: 8px;font-family: Optima, \"Microsoft YaHei\", PingFangSC-regular, serif;text-wrap: wrap;background-color: rgb(255, 255, 255);'>Neuroimaging evidence complements that from individuals with brain damage. Tools such as functional magnetic resonance imaging (fMRI) allow the identification of language areas in intact, healthy brains, which can then be examined for their responses while individuals perform tasks involving different kinds of thought. In \"Language Networks in the Human Brain,\" these networks are a set of brain regions ubiquitously involved when we understand and produce language. The responses of these networks to various non-linguistic inputs and tasks have been studied, and the evidence suggests that all regions of the language network are essentially \"silent\" across all forms of thought tested, including mathematical reasoning, formal logic reasoning, demanding executive function tasks like working memory or cognitive control tasks, understanding computer code, thinking about others' mental states, and making semantic judgments about objects or events. These tasks activate other brain areas that do not overlap with the language network, though they may sometimes be located near language regions. Future work might uncover some thought tasks that engage the language areas and challenge aphasics, but no such tasks have been found yet.\u003C/p>\u003Cp style='margin-bottom: 0px;;color: rgb(0, 0, 0);font-size: 16px;line-height: 1.8em;letter-spacing: normal;text-align: left;padding-top: 8px;padding-bottom: 8px;font-family: Optima, \"Microsoft YaHei\", PingFangSC-regular, serif;text-wrap: wrap;background-color: rgb(255, 255, 255);'>Some hypotheses about the use of language for thought particularly involve cognitive development. According to these hypotheses, language (or, in some views, broader symbolic representation) may be crucial for the development of certain types or modes of thought. Some evidence supporting this view comes from teaching children or non-human primates certain vocabulary or symbols for concepts (e.g., the concept of \"same\") or labeling in a way that highlights task-relevant dimensions of the world (e.g., drawing attention to object size by marking larger objects as \"daddy\" and smaller ones as \"baby\"), which can lead to success in certain relational reasoning tasks. Other studies suggest that training young children in certain syntactic structures (e.g., complement clauses) enables them to pass theory-of-mind tasks. However, there are reasons for skepticism. First, recent evidence shows that the separation between language networks and systems supporting thought and reasoning is already present in young children, contradicting the possibility that thought depends on linguistic resources at early developmental stages. Second, some children who grow up without exposure to language can still perform complex reasoning.\u003C/p>\u003Cp style='margin-bottom: 0px;;color: rgb(0, 0, 0);font-size: 16px;line-height: 1.8em;letter-spacing: normal;text-align: left;padding-top: 8px;padding-bottom: 8px;font-family: Optima, \"Microsoft YaHei\", PingFangSC-regular, serif;text-wrap: wrap;background-color: rgb(255, 255, 255);'>In particular, some deaf children born to hearing parents, unable to hear language, sometimes go years without exposure to it. Lack of language exposure harms many aspects of cognition, as expected, since language provides a key source of information for learning about the world. Nevertheless, individuals experiencing language deprivation undoubtedly demonstrate the ability for complex cognitive functions: they can still learn mathematics, engage in relational reasoning, construct causal chains, and acquire rich and complex knowledge of the world (more controversial evidence exists regarding language deprivation in cases of child abuse). In other words, the absence of linguistic representation does not make complex (including symbolic) thought fundamentally impossible, although certain aspects of reasoning do show delays. Thus, in typical development, language and reasoning develop in parallel (see consistent evidence in \"Communication and Thought in Humans and Animals\").\u003C/p>\u003Cp style='margin-bottom: 0px;;color: rgb(0, 0, 0);font-size: 16px;line-height: 1.8em;letter-spacing: normal;text-align: left;padding-top: 8px;padding-bottom: 8px;font-family: Optima, \"Microsoft YaHei\", PingFangSC-regular, serif;text-wrap: wrap;background-color: rgb(255, 255, 255);'>Finally, it is worth noting that pre-linguistic infants and many animal species—including non-human primates, corvids, elephants, and cephalopods—demonstrate impressive reasoning and problem-solving abilities, clearly accomplished without language, further questioning the necessity of language or language-like systems for complex thought.\u003C/p>\u003Ch2 style='margin-top: 30px;margin-bottom: 15px;color: rgba(0, 0, 0, 0.85);;font-family: Optima, \"Microsoft YaHei\", PingFangSC-regular, serif;letter-spacing: normal;text-align: left;text-wrap: wrap;background-color: rgb(255, 255, 255);'>\u003Cspan style=\";font-size: 22px;color: rgb(0, 0, 0);line-height: 1.5em;letter-spacing: 0em;font-weight: bold;display: block;\">Language is an efficient communication code.\u003C/span>\u003C/h2>\u003Cp style='margin-bottom: 0px;;color: rgb(0, 0, 0);font-size: 16px;line-height: 1.8em;letter-spacing: normal;text-align: left;padding-top: 8px;padding-bottom: 8px;font-family: Optima, \"Microsoft YaHei\", PingFangSC-regular, serif;text-wrap: wrap;background-color: rgb(255, 255, 255);'>An effective communication code should be easy to produce and understand, able to resist noise (environmental noise or noise caused by imperfect processing mechanisms), and must be learnable by people. Human languages—whether spoken or signed—all exhibit these properties, which are reflected at all levels of language structure, including sounds, words, and syntax.\u003C/p>\u003Cp style='margin-bottom: 0px;;color: rgb(0, 0, 0);font-size: 16px;line-height: 1.8em;letter-spacing: normal;text-align: left;padding-top: 8px;padding-bottom: 8px;font-family: Optima, \"Microsoft YaHei\", PingFangSC-regular, serif;text-wrap: wrap;background-color: rgb(255, 255, 255);'>While some of these properties are not uniquely predicted by the view that language is used for communication. However, proponents of the idea that language is used for thought have long pointed to the lack of communicative properties in language as evidence against its communicative function, so it is important to summarize the rich cross-linguistic evidence for the presence of these properties. Moreover, some communicative properties are difficult to explain under the view that language is primarily used for internal thought.\u003C/p>\u003Cp style='margin-bottom: 0px;;color: rgb(0, 0, 0);font-size: 16px;line-height: 1.8em;letter-spacing: normal;text-align: left;padding-top: 8px;padding-bottom: 8px;font-family: Optima, \"Microsoft YaHei\", PingFangSC-regular, serif;text-wrap: wrap;background-color: rgb(255, 255, 255);'>Starting from the lowest level of language structure, the sounds of a language are distributed in phonetic space so as to transmit information in a way that resists noise corruption and promotes perception and understanding. The combination of sounds in a language also appears to be influenced by factors related to the ease of articulation of different sound categories, including anatomical changes in speech organs related to physical environment and diet.\u003C/p>\u003Cp style='margin-bottom: 0px;;color: rgb(0, 0, 0);font-size: 16px;line-height: 1.8em;letter-spacing: normal;text-align: left;padding-top: 8px;padding-bottom: 8px;font-family: Optima, \"Microsoft YaHei\", PingFangSC-regular, serif;text-wrap: wrap;background-color: rgb(255, 255, 255);'>Regarding word forms, languages tend to reuse short and efficient words and sequences of sounds within words. Furthermore, frequently used and low-information-content words tend to be shorter. These properties make retrieval during word production and comprehension easier (frequently used words are easier to retrieve from memory), and more efficient in terms of articulation (shorter words consume less energy, and repeated sound sequences can be stored as blocks in motor memory, reducing articulatory planning costs).\u003C/p>\u003Cp style='margin-bottom: 0px;;color: rgb(0, 0, 0);font-size: 16px;line-height: 1.8em;letter-spacing: normal;text-align: left;padding-top: 8px;padding-bottom: 8px;font-family: Optima, \"Microsoft YaHei\", PingFangSC-regular, serif;text-wrap: wrap;background-color: rgb(255, 255, 255);'>In terms of word meaning, words optimize communication efficiency by balancing complexity (the cost of acquiring or representing meaning) with informativeness (the precision and clarity with which a word selects its meaning). An example is kinship terms: languages choose different solutions at the optimal boundary between simplicity and information-rich meanings (such as words that can specify particular members of a family tree, like \"grandmother\"). The efficiency found in natural languages means that it is usually impossible to simplify the kinship system of a real language while maintaining its informativeness, and vice versa. Similar results have been reported for color terms, season terms, closed-class words, and grammatical markers. Lexical systems also exhibit features adapted to specific communicative needs, covering parts of the conceptual space relevant to a particular community more densely. As shown in the figure below:\u003C/p>\u003Csection style=';margin-top: 20px;margin-bottom: 20px;padding: 10px 20px;border-style: none;border-width: 3px;border-color: rgba(0, 0, 0, 0.4);border-radius: 0px;background: none 0% 0% / auto no-repeat scroll padding-box border-box rgb(250, 250, 250);width: auto;height: auto;box-shadow: rgba(0, 0, 0, 0) 0px 0px 0px 0px;color: rgb(0, 0, 0);font-family: Optima, \"Microsoft YaHei\", PingFangSC-regular, serif;font-size: 16px;letter-spacing: normal;text-align: left;text-wrap: wrap;'>\u003Csection style=\";\">\u003Cp style=\";line-height: 1.8em;letter-spacing: 0em;text-indent: 0em;padding-top: 8px;padding-bottom: 8px;\">\u003Cimg class=\"rich_pages wxw-img\" data-galleryid=\"\" data-imgfileid=\"100006156\" data-ratio=\"0.9384615384615385\" data-s=\"300,640\" data-type=\"png\" data-w=\"260\" style=\"text-align: center;letter-spacing: 0em;text-indent: 0em;\" src=\"https://res.cooltool.vip/article_res/assets/17423802981370.6623262331230819.png\">\u003C/p>\u003Cp style=\";line-height: 1.8em;letter-spacing: 0em;text-indent: 0em;padding-top: 8px;padding-bottom: 8px;\">：Words in many semantic domains trade off complexity against informativeness. This pattern matches the prediction of an efficient communication system (shown here for the grammatical number marking domain). Across the space of all possible grammatical systems, attested inventories are plotted (black dots; size corresponds to the number of languages (N) with a given inventory) and unattested systems (gray dots). Systems that achieve the optimal trade-off lie on the Pareto frontier (solid line); the shaded area below the line shows impossible trade-offs. DU, dual; GPAUC, approximate number (many); PAUC, paucal (few); PL, plural; SG, singular; TR, trial. Optional values are indicated by subscript \"o\".\u003C/p>\u003C/section>\u003C/section>\u003Cp style='margin-bottom: 0px;;color: rgb(0, 0, 0);font-size: 16px;line-height: 1.8em;letter-spacing: normal;text-align: left;padding-top: 8px;padding-bottom: 8px;font-family: Optima, \"Microsoft YaHei\", PingFangSC-regular, serif;text-wrap: wrap;background-color: rgb(255, 255, 255);'>Syntax is the most debated part in discussions about whether language is optimized for efficient communication. Syntax dictates how words combine to express a vast array of meanings, i.e., “the infinite use of finite means.” The defining features of syntax—hierarchy and compositionality—may be precisely the result of pressures seen in word meaning: the pressure for simplicity (crucial for learnability) and the pressure for expressivity (vital for effective communication).\u003C/p>\u003Cp style='margin-bottom: 0px;;color: rgb(0, 0, 0);font-size: 16px;line-height: 1.8em;letter-spacing: normal;text-align: left;padding-top: 8px;padding-bottom: 8px;font-family: Optima, \"Microsoft YaHei\", PingFangSC-regular, serif;text-wrap: wrap;background-color: rgb(255, 255, 255);'>Moreover, various syntactic patterns across languages worldwide can be explained by combining communicative and cognitive pressures. A key example is the tendency of languages to minimize the length of dependencies between words. In any given sentence, words combine according to grammatical rules to form larger meanings. For instance, in the sentence “Lana ate five apples,” there is a dependency between “five” and “apples,” but not between “Lana” and “five,” as these words lack a direct semantic relationship. Longer-distance connections (those with more intervening words) increase the difficulty of both production and comprehension, as confirmed by behavioral measures or brain imaging. Due to this cognitive cost, languages may have evolved to become more processable by minimizing dependency lengths—that is, through usage. This functional pressure to keep dependencies localized in syntax explains several universal tendencies in word order. As shown in the figure below:\u003C/p>\u003Csection style=';margin-top: 20px;margin-bottom: 20px;padding: 10px 20px;border-style: none;border-width: 3px;border-color: rgba(0, 0, 0, 0.4);border-radius: 0px;background: none 0% 0% / auto no-repeat scroll padding-box border-box rgb(250, 250, 250);width: auto;height: auto;box-shadow: rgba(0, 0, 0, 0) 0px 0px 0px 0px;color: rgb(0, 0, 0);font-family: Optima, \"Microsoft YaHei\", PingFangSC-regular, serif;font-size: 16px;letter-spacing: normal;text-align: left;text-wrap: wrap;'>\u003Csection style=\";\">\u003Cp style=\";line-height: 1.8em;letter-spacing: 0em;text-indent: 0em;padding-top: 8px;padding-bottom: 8px;\">\u003Cimg class=\"rich_pages wxw-img\" data-galleryid=\"\" data-imgfileid=\"100006157\" data-ratio=\"0.7543352601156069\" data-s=\"300,640\" data-type=\"png\" data-w=\"346\" style=\"text-align: center;letter-spacing: 0em;text-indent: 0em;\" src=\"https://res.cooltool.vip/article_res/assets/17423802986910.5258115324084662.png\">\u003C/p>\u003Cp style=\";line-height: 1.8em;letter-spacing: 0em;text-indent: 0em;padding-top: 8px;padding-bottom: 8px;\">：Languages minimize syntactic dependency lengths across languages. Based on the analysis of large-scale language corpora, the observed average dependency lengths for sentences with 1 to 50 words in four diverse types of languages (black line in each graph). For each sentence in the corpus, a value is calculated by summing up all dependency lengths as shown in c. The red dashed line shows a random baseline created by first shuffling word order while preserving hierarchical dependency structures and disallowing crossing dependencies, then recalculating dependency lengths. All lines are fitted using generalized additive models. Across languages, observed dependency lengths are below the random baseline, indicating that languages have evolved to shorten dependency lengths, presumably to facilitate production and comprehension.\u003C/p>\u003C/section>\u003C/section>\u003Csection style=';margin-top: 20px;margin-bottom: 20px;padding: 10px 20px;border-style: none;border-width: 3px;border-color: rgba(0, 0, 0, 0.4);border-radius: 0px;background: none 0% 0% / auto no-repeat scroll padding-box border-box rgb(250, 250, 250);width: auto;height: auto;box-shadow: rgba(0, 0, 0, 0) 0px 0px 0px 0px;color: rgb(0, 0, 0);font-family: Optima, \"Microsoft YaHei\", PingFangSC-regular, serif;font-size: 16px;letter-spacing: normal;text-align: left;text-wrap: wrap;'>\u003Csection style=\";\">\u003Cp style=\";line-height: 1.8em;letter-spacing: 0em;text-indent: 0em;padding-top: 8px;padding-bottom: 8px;\">\u003Cimg class=\"rich_pages wxw-img\" data-galleryid=\"\" data-imgfileid=\"100006158\" data-ratio=\"1.126984126984127\" data-s=\"300,640\" data-type=\"png\" data-w=\"504\" style=\"text-align: center;letter-spacing: 0em;text-indent: 0em;\" src=\"https://res.cooltool.vip/article_res/assets/17423802980180.9876003752881366.png\">\u003C/p>\u003Cp style=\";line-height: 1.8em;letter-spacing: 0em;text-indent: 0em;padding-top: 8px;padding-bottom: 8px;\">：Examples of syntactic dependency length minimization in different languages. First row, syntactic dependency structure for subject-verb-object ordered languages (e.g., English). The verb appears before the object noun; prepositions appear before their object nouns. In this and other examples, the syntactic categories of words are shown below each word, and relationships between words are indicated by directed arcs; the type of relationship is marked above each arc. The total dependency length of a sentence is the sum of all dependency distances — for example, the dependency length between 'Alfred' and 'said' is 1; for non-adjacent words, the dependency length is the number of intervening words plus one. For this sentence, there are seven local dependencies of length 1 and three dependencies of length 2, resulting in a total sentence dependency length of 7 + 6 = 13. Second row, syntactic dependency structure for subject-object-verb ordered languages (e.g., Japanese). The verb appears after the object noun; postpositions (prepositions) appear after their object nouns. Two word orders rarely found in natural languages are hypothesized to be rare because they introduce long-distance dependencies: in the third row, the verb appears after the object noun, and the preposition appears before the object noun; in the fourth row, the verb appears before the object noun, and the postposition appears after the object noun. Comp, complementizer; Ind-Obj, indirect object; Mod, modifier; N, noun; Obj, object; root, main root of the sentence; SComp, sentence complement; Subj, subject; VComp, verb with complementizer; VN, verb with noun; VN, Prep, verb with noun and preposition; Prep, preposition; Wh-pro, wh-pronoun (e.g., 'who').\u003C/p>\u003C/section>\u003C/section>\u003Cp style='margin-bottom: 0px;;color: rgb(0, 0, 0);font-size: 16px;line-height: 1.8em;letter-spacing: normal;text-align: left;padding-top: 8px;padding-bottom: 8px;font-family: Optima, \"Microsoft YaHei\", PingFangSC-regular, serif;text-wrap: wrap;background-color: rgb(255, 255, 255);'>Other examples involve the ordering of basic elements — subject (roughly, the agent), verb, and object (roughly, the recipient of the action) — within sentences to convey complex meanings. The subject-object-verb order is the most common order in the world's languages (found in about 47% of languages, such as Japanese, Persian, and Hindi), seemingly a cognitively natural default: speakers of different languages use this order when expressing event meanings through gestures, and emerging sign languages also exhibit this order. Gibson et al. propose a communicative explanation for why some languages shift from the default subject-object-verb order to the second most common subject-verb-object order (found in about 41% of languages, such as English, Ukrainian, and Mandarin). Specifically, in the subject-verb-object order, listeners can use positional cues — whether a noun appears before or after a verb — to reconstruct who is doing what to whom if information is lost during communication.\u003C/p>\u003Cp style='margin-bottom: 0px;;color: rgb(0, 0, 0);font-size: 16px;line-height: 1.8em;letter-spacing: normal;text-align: left;padding-top: 8px;padding-bottom: 8px;font-family: Optima, \"Microsoft YaHei\", PingFangSC-regular, serif;text-wrap: wrap;background-color: rgb(255, 255, 255);'>A common argument against the evolution of language as a communication system is the pervasive presence of ambiguity: most words have multiple meanings, and many (especially longer) sentences have multiple possible dependency structures. For instance, Chomsky has long argued that the existence of ambiguity implies that language is primarily used for thought rather than communication, since ambiguous signals would hinder communication. However, under the view that language is used for thought, human thinking does not seem ambiguous, making the ambiguity in human language unexpected. On the contrary, theories of language as a communication tool naturally predict the existence of ambiguity in language. That is, ambiguity can be mathematically proven to be useful for communication: it allows speakers to omit information already known to the listener (e.g., from context) and permits the reuse of short, easy-to-produce linguistic forms. A system that disallows ambiguity would require a much larger lexicon and grammar than the human language system, necessitating long words and sentences to convey simple meanings. For example, an artificially constructed language designed to eliminate ambiguity was found to be too complex for humans to learn and required repeated modifications to make it learnable. Moreover, the existence of words with multiple related meanings (e.g., \"water\" functioning as both a noun and a verb) is considered helpful for learning, as the acquisition of one meaning facilitates the acquisition of related meanings. Thus, the ability to handle multiple form-meaning mappings may be useful for maintaining a rich lexicon.\u003C/p>\u003Cp style='margin-bottom: 0px;;color: rgb(0, 0, 0);font-size: 16px;line-height: 1.8em;letter-spacing: normal;text-align: left;padding-top: 8px;padding-bottom: 8px;font-family: Optima, \"Microsoft YaHei\", PingFangSC-regular, serif;text-wrap: wrap;background-color: rgb(255, 255, 255);'>Of course, some characteristics of human language may also have non-communicative explanations. For example, the compositionality in language may simply reflect a pre-existing compositionality in thought, or even in low-level perceptual and motor systems. This view still runs counter to the assumption that language is used for thinking, as the directionality is reversed.\u003C/p>\u003Ch2 style='margin-top: 30px;margin-bottom: 15px;color: rgba(0, 0, 0, 0.85);;font-family: Optima, \"Microsoft YaHei\", PingFangSC-regular, serif;letter-spacing: normal;text-align: left;text-wrap: wrap;background-color: rgb(255, 255, 255);'>\u003Cspan style=\";font-size: 22px;color: rgb(0, 0, 0);line-height: 1.5em;letter-spacing: 0em;font-weight: bold;display: block;\">Communication and Thought in Humans and Animals\u003C/span>\u003C/h2>\u003Cp style='margin-bottom: 0px;;color: rgb(0, 0, 0);font-size: 16px;line-height: 1.8em;letter-spacing: normal;text-align: left;padding-top: 8px;padding-bottom: 8px;font-family: Optima, \"Microsoft YaHei\", PingFangSC-regular, serif;text-wrap: wrap;background-color: rgb(255, 255, 255);'>Interactions with conspecifics across species require a communication system — a mechanism for perceiving and emitting signals, along with storage for the association between signals and meanings. The human communication system is undoubtedly complex, but does language give us a new form of reasoning, or merely reflect the complexity of independent human thought? In this review, based on previous theories and evidence from the past two decades, it appears that language is not necessary for any form of thought tested so far, nor is it sufficient to independently support thought. Moreover, based on evidence that many features of natural languages seem optimized for efficient information transfer, the article concludes that communication is the primary function of language.\u003C/p>\u003Cp style='margin-bottom: 0px;;color: rgb(0, 0, 0);font-size: 16px;line-height: 1.8em;letter-spacing: normal;text-align: left;padding-top: 8px;padding-bottom: 8px;font-family: Optima, \"Microsoft YaHei\", PingFangSC-regular, serif;text-wrap: wrap;background-color: rgb(255, 255, 255);'>The view that language is merely a communication system aligns with the perspective of continuity in human evolution. In this view, the characteristics of human language — including its complexity — may result from the multifaceted environment in which it evolved, where the system must be useful (able to express the content of a person's inner thoughts) and learnable, and humans must be able to process languages within the specialties and limitations of their existing cognitive and neural systems. The alternative view is that language is the substrate of thought, implying a marked discontinuity between our species and others. This view regards language as a transformative mechanism, perhaps innate, endowing humans with a new form of mental computational representation.\u003C/p>\u003Cp style='margin-bottom: 0px;;color: rgb(0, 0, 0);font-size: 16px;line-height: 1.8em;letter-spacing: normal;text-align: left;padding-top: 8px;padding-bottom: 8px;font-family: Optima, \"Microsoft YaHei\", PingFangSC-regular, serif;text-wrap: wrap;background-color: rgb(255, 255, 255);'>Why does the close relationship between language and thought appeal intuitively to many people? Some insist on the superiority of humans in the animal kingdom (that is, the differences between them and other animals are not only quantitative but qualitative), despite scientific evidence showing strong biological similarities between humans and non-human animals. Another reason may relate to the parsimony in explaining the differences between humans and non-human animals (for other reasons, see Supplementary Information). Specifically, humans differ from other animals in the complexity of their communication systems as well as in their thought and cognition. A parsimonious explanation tends toward a single-factor explanation — for instance, humans evolved language, and cognitive changes were merely a consequence of this shift. However, evidence on the evolution of the human brain shows that the complexity of multiple cognitive systems increased simultaneously.\u003C/p>\u003Cp style='margin-bottom: 0px;;color: rgb(0, 0, 0);font-size: 16px;line-height: 1.8em;letter-spacing: normal;text-align: left;padding-top: 8px;padding-bottom: 8px;font-family: Optima, \"Microsoft YaHei\", PingFangSC-regular, serif;text-wrap: wrap;background-color: rgb(255, 255, 255);'>Compared to the brains of other animals, including non-human primates, the human brain's association cortex — which accommodates mental processes beyond sensory perception and motor control — has expanded significantly in proportion. The association cortex spans the frontal, temporal, and parietal lobes and includes multiple large-scale networks in humans that support certain cognitive functions. The language network is just one of these networks. Significant progress has been made in characterizing other networks that support human cognitive abilities, including those underlying non-linguistic tasks discussed in \"language is neither necessary nor sufficient for thought.\" For example, the so-called \"multiple demand\" network supports a variety of goal-directed behaviors, including novel problem-solving, and its damage can lead to impairments in fluid intelligence. Mathematical and logical reasoning as well as computer code processing also rely on the multiple demand network. Other such networks include the \"theory of mind\" network, which supports social reasoning, including mentalization or the ability to think about others' thoughts, and the \"default\" network, whose functions are still debated. Some associate its regions with scenario projection into the past or future, while others link it to spatial cognition and reasoning. At least some of these networks have homologs in the brains of non-human animals — cross-species functional structural correspondences remain an ongoing effort in neuroscience.\u003C/p>\u003Cp style='margin-bottom: 0px;;color: rgb(0, 0, 0);font-size: 16px;line-height: 1.8em;letter-spacing: normal;text-align: left;padding-top: 8px;padding-bottom: 8px;font-family: Optima, \"Microsoft YaHei\", PingFangSC-regular, serif;text-wrap: wrap;background-color: rgb(255, 255, 255);'>Importantly, multiple brain networks expanded during human evolution, and this expansion correlates with increases in various cognitive abilities. Whether this expansion truly proceeded in parallel or whether the emergence or expansion of one network critically drove the expansion of others remains unclear, but the former possibility seems more reasonable because the increase in multiple cognitive abilities could have enhanced survival probabilities — including social complexity (the ability to simulate others' thoughts), the capacity to infer the causal structure of the world, flexible problem-solving and future planning, and better communication skills. Regardless of the exact timeline and order of the expansion of different cognitive networks in the modern human brain, the language network that supports our ability to communicate with conspecifics is distinct from the networks that support our ability to think and reason, making the view that language mediates thought less likely.\u003C/p>\u003Ch2 style='margin-top: 30px;margin-bottom: 15px;color: rgba(0, 0, 0, 0.85);;font-family: Optima, \"Microsoft YaHei\", PingFangSC-regular, serif;letter-spacing: normal;text-align: left;text-wrap: wrap;background-color: rgb(255, 255, 255);'>\u003Cspan style=\";font-size: 22px;color: rgb(0, 0, 0);line-height: 1.5em;letter-spacing: 0em;font-weight: bold;display: block;\">Conclusion\u003C/span>\u003C/h2>\u003Cp style='margin-bottom: 0px;;color: rgb(0, 0, 0);font-size: 16px;line-height: 1.8em;letter-spacing: normal;text-align: left;padding-top: 8px;padding-bottom: 8px;font-family: Optima, \"Microsoft YaHei\", PingFangSC-regular, serif;text-wrap: wrap;background-color: rgb(255, 255, 255);'>Language primarily serves a communicative function, reflecting rather than initiating the unique complexity of human cognition. It is not the key substrate providing for thought and reasoning but has transformed our species by enabling the transmission of knowledge across generations. Language is undoubtedly an extremely useful tool for transmitting knowledge. The cumulative effect of this transmission — where knowledge builds upon prior knowledge — along with the enhancement of our social and problem-solving abilities, may have enabled us to create human civilization. While all tested forms of thinking are evidently possible without language, our species' achievements might not have been realized without the cumulative culture facilitated by the external use of language.\u003C/p>\u003Cp style=\"display: none;\">\u003Cmp-style-type data-value=\"3\">\u003C/mp-style-type>\u003C/p>\u003C/div>",1752585456242]