什么是“认知架构”（如 ACT-R、SOAR）？
What are cognitive architectures like ACT-R or SOAR?
认知架构是用一套固定结构与规则模拟人类感知、记忆、决策的计算框架，目标是统一解释多种认知任务。它们把心理学理论变成可运行模型，用来预测反应时与错误模式。
Cognitive architectures are computational frameworks with fixed structures and rules simulating human perception, memory and decision-making, aiming to unify explanations across tasks. They turn psychological theory into runnable models predicting reaction times and errors.
难度 专业 · 适合 高中(15-18)
CC BY 4.0 · 已核验 · Verified 2026-11-04
来源 · Source
Liu,《QN-ACES: Integrating Queueing Network and ACT-R, CAPS, EPIC, and Soar Architectures for Multitask Cognitive Modeling》,International Journal of Human-Computer Interaction,2009 (DOI: 10.1080/10447310902973182)
Liu, 'QN-ACES: Integrating Queueing Network and ACT-R, CAPS, EPIC, and Soar Architectures for Multitask Cognitive Modeling', International Journal of Human-Computer Interaction, 2009, DOI: 10.1080/10447310902973182
独立核验依据 · Independent verification: 核验依据 1 CROSSREF 标识 10.1080/10447310902973182：QN-ACES: Integrati
#认知架构 #计算模型 #ACT-R
大语言模型表现出“心智理论”了吗？
Do large language models show 'theory of mind'?
近期模型在标准化错误信念任务上通过率接近人类，但它们是从海量文本统计出行为模式，而非拥有真实信念表征。这引发争论：表面通过不等于具备内省式的他人心智理解，更像高级模式匹配。
Recent models pass standard false-belief tasks near human level, but they derive patterns from massive text, not genuine belief representations. Debate continues: passing is not possessing introspective understanding of others' minds—it may be advanced pattern matching.
难度 专业 · 适合 高中(15-18)
CC BY 4.0 · 已核验 · Verified 2026-11-04
来源 · Source
Lombardi, Lenci,《What Do You Mean? Exploring the Alleged Theory of Mind Abilities of Large Language Models》, Transactions of the Association for Computational Linguistic, 2026 (DOI: 10.1162/tacl.a.807)
Lombardi, Lenci, 'What Do You Mean? Exploring the Alleged Theory of Mind Abilities of Large Language Models', Transactions of the Association for Computational Linguistic, 2026, DOI: 10.1162/tacl.a.807
独立核验依据 · Independent verification: 核验依据 1 Crossref DOI 10.1162/tacl.a.807
#大语言模型 #心智理论 #AI
深度学习的人脸识别与人类视觉有何异同？
How does deep-learning face recognition compare with human vision?
卷积网络在海量数据上能超越人类辨脸，却会被对抗样本（细微噪点）轻易骗过，而人类对此稳健。差异提示人脑用了结构化的三维与语义先验，AI 更多是统计相关性，二者互补而非等同。
Convolutional networks can surpass humans at face ID given data, yet are fooled by adversarial perturbations (tiny noise) that humans shrug off. The gap suggests human vision uses structured 3D and semantic priors, while AI leans on correlations—complementary, not equivalent.
难度 专业 · 适合 高中(15-18)
CC BY 4.0 · 已核验 · Verified 2026-10-04
来源 · Source
Goodfellow, I., Bengio, Y. 与 Courville, A. (2016). 《深度学习》. MIT Press.
Goodfellow, I., Bengio, Y., & Courville, A. (2016). Deep Learning. MIT Press.
独立核验依据 · Independent verification: 核验依据 1 ISBN 978-0-262-03561-3 (MIT Press)
#深度学习 #对抗样本 #视觉
什么是“预测编码”理论？大脑是预测机器吗？
What is predictive coding theory—is the brain a prediction machine?
预测编码认为大脑不断生成对感官输入的预测，只向上传递“预测误差”而非全部信息，再用误差更新模型。这一框架统一了解释感知、注意与行动，也启发了类脑 AI 的设计。
Predictive coding says the brain constantly predicts sensory input and passes only 'prediction errors' upward, updating its model from those errors. The framework unifies perception, attention and action and inspires brain-like AI design.
难度 专业 · 适合 高中(15-18)
CC BY 4.0 · 已核验 · Verified 2026-11-04 · 2 条独立证据
来源 · Source
Rauss, Pourtois,《What is Bottom-Up and What is Top-Down in Predictive Coding?》,Frontiers in Psychology,2013 (DOI: 10.3389/fpsyg.2013.00276)
Rauss, Pourtois, 'What is Bottom-Up and What is Top-Down in Predictive Coding?', Frontiers in Psychology, 2013, DOI: 10.3389/fpsyg.2013.00276
独立核验依据 · Independent verification: 核验依据 1 Crossref DOI 10.3389/fpsyg.2013.00276 · 核验依据 2 ARCHIVE 标识 pubmed-PMC3656342：What is Bottom-Up and What
#预测编码 #贝叶斯 #大脑
机器会有“意识”吗？意识的“困难问题”是什么？
Could machines be conscious? What is the 'hard problem' of consciousness?
困难问题指：即使说清大脑如何加工信息，仍解释不了“为何有主观体验”（为何红色对你是那种感受）。当前 AI 能模拟功能却无证据拥有内在感受，因此机器意识仍是开放且高度争议的哲学-科学难题。
The hard problem: even if we explain how the brain processes information, we still can't explain why there is subjective experience (why red feels that way to you). Today's AI can simulate function but shows no evidence of inner feel, so machine consciousness remains an open, contested question.
难度 专业 · 适合 高中(15-18)
CC BY 4.0 · 已核验 · Verified 2026-08-23 · 2 条独立证据
来源 · Source
Richard Gerrig Philip Zimbardo,《心理学与生活》第19版中译,人民邮电出版社,2016
R. J. Gerrig and P. G. Zimbardo, Psychology and Life, 19th ed., Chinese ed., Posts & Telecom Press, 2016
独立核验依据 · Independent verification: 核验依据 2 OPENLIBRARY 标识 /works/OL23493W：Psychology and life（按标题+作
#意识 #困难问题 #AI
“具身认知”对人工智能有什么启示？
What does embodied cognition imply for AI?
具身观点认为心智依赖身体与环境的实时交互，而非孤立符号运算。它推动机器人研究从纯文本走向传感器-动作闭环，认为真正灵活智能需要身体在真实世界中试错学习。
Embodied cognition holds that mind depends on real-time body-environment interaction, not isolated symbol manipulation. It pushes robotics from text-only toward sensor-actuator loops, arguing flexible intelligence needs a body learning by trial in the real world.
难度 进阶 · 适合 大众
CC BY 4.0 · 已核验 · Verified 2026-08-23 · 2 条独立证据
来源 · Source
Richard Gerrig Philip Zimbardo,《心理学与生活》第19版中译,人民邮电出版社,2016
R. J. Gerrig and P. G. Zimbardo, Psychology and Life, 19th ed., Chinese ed., Posts & Telecom Press, 2016
独立核验依据 · Independent verification: 核验依据 2 OPENLIBRARY 标识 /works/OL23493W：Psychology and life（按标题+作
#具身 #机器人 #智能