Suppr超能文献

表征的计算起源。

The computational origin of representation.

作者信息

Piantadosi Steven T

机构信息

UC Berkeley, Department of Psycholoy.

出版信息

Minds Mach (Dordr). 2021 Mar;31:1-58. doi: 10.1007/s11023-020-09540-9. Epub 2020 Nov 3.

Abstract

Each of our theories of mental representation provides some insight into how the mind works. However, these insights often seem incompatible, as the debates between symbolic, dynamical, emergentist, sub-symbolic, and grounded approaches to cognition attest. Mental representations-whatever they are-must share many features with each of our theories of representation, and yet there are few hypotheses about how a synthesis could be possible. Here, I develop a theory of the underpinnings of symbolic cognition that shows how sub-symbolic dynamics may give rise to higher-level cognitive representations of structures, systems of knowledge, and algorithmic processes. This theory implements a version of conceptual role semantics by positing an internal universal representation language in which learners may create mental models to capture dynamics they observe in the world. The theory formalizes one account of how truly novel conceptual content may arise, allowing us to explain how even elementary logical and computational operations may be learned from a more primitive basis. I provide an implementation that learns to represent a variety of structures, including logic, number, kinship trees, regular languages, context-free languages, domains of theories like magnetism, dominance hierarchies, list structures, quantification, and computational primitives like repetition, reversal, and recursion. This account is based on simple discrete dynamical processes that could be implemented in a variety of different physical or biological systems. In particular, I describe how the required dynamics can be directly implemented in a connectionist framework. The resulting theory provides an "assembly language" for cognition, where high-level theories of symbolic computation can be implemented in simple dynamics that themselves could be encoded in biologically plausible systems.

摘要

我们关于心理表征的每一种理论都为心智的运作方式提供了一些见解。然而,这些见解往往看似相互矛盾,正如符号主义、动态主义、突现主义、亚符号主义和基于 grounded 的认知方法之间的争论所证明的那样。心理表征——无论它们是什么——必然与我们的每一种表征理论都有许多共同特征,然而关于如何实现综合的假设却很少。在这里,我提出了一种符号认知基础的理论,它展示了亚符号动态如何产生对结构、知识系统和算法过程的高级认知表征。该理论通过假定一种内部通用表征语言来实现概念角色语义的一个版本,学习者可以在这种语言中创建心理模型来捕捉他们在世界中观察到的动态。该理论形式化了一种关于真正新颖的概念内容如何产生的解释,使我们能够解释即使是基本的逻辑和计算操作如何能从更原始的基础上学到。我提供了一个实现,它学会表征各种结构,包括逻辑、数字、亲属树、正则语言、上下文无关语言、诸如磁性等理论领域、支配层次结构、列表结构、量化以及诸如重复、反转和递归等计算原语。这个解释基于简单的离散动态过程,这些过程可以在各种不同的物理或生物系统中实现。特别是,我描述了所需的动态如何能在联结主义框架中直接实现。由此产生的理论为认知提供了一种“汇编语言”,其中符号计算的高级理论可以在简单的动态中实现,而这些动态本身可以编码在生物学上合理的系统中。

相似文献

1
The computational origin of representation.
Minds Mach (Dordr). 2021 Mar;31:1-58. doi: 10.1007/s11023-020-09540-9. Epub 2020 Nov 3.
2
The foundation of Piaget's theories: mental and physical action.
Adv Child Dev Behav. 1999;27:221-46. doi: 10.1016/s0065-2407(08)60140-8.
3
Concepts as Semantic Pointers: A Framework and Computational Model.
Cogn Sci. 2016 Jul;40(5):1128-62. doi: 10.1111/cogs.12265. Epub 2015 Aug 1.
4
Macromolecular crowding: chemistry and physics meet biology (Ascona, Switzerland, 10-14 June 2012).
Phys Biol. 2013 Aug;10(4):040301. doi: 10.1088/1478-3975/10/4/040301. Epub 2013 Aug 2.
6
Why concepts are (probably) vectors.
Trends Cogn Sci. 2024 Sep;28(9):844-856. doi: 10.1016/j.tics.2024.06.011. Epub 2024 Aug 7.
7
8
A connectionist computational model for epistemic and temporal reasoning.
Neural Comput. 2006 Jul;18(7):1711-38. doi: 10.1162/neco.2006.18.7.1711.
9
Toward a Unified Sub-symbolic Computational Theory of Cognition.
Front Psychol. 2016 Jun 21;7:925. doi: 10.3389/fpsyg.2016.00925. eCollection 2016.
10
Connectionist perspectives on language learning, representation and processing.
Wiley Interdiscip Rev Cogn Sci. 2015 May-Jun;6(3):235-47. doi: 10.1002/wcs.1340. Epub 2015 Jan 30.

引用本文的文献

1
Explainability Through Systematicity: The Hard Systematicity Challenge for Artificial Intelligence.
Minds Mach (Dordr). 2025;35(3):35. doi: 10.1007/s11023-025-09738-9. Epub 2025 Jul 29.
2
Symbolic metaprogram search improves learning efficiency and explains rule learning in humans.
Nat Commun. 2024 Aug 10;15(1):6847. doi: 10.1038/s41467-024-50966-x.
3
Natural language syntax complies with the free-energy principle.
Synthese. 2024;203(5):154. doi: 10.1007/s11229-024-04566-3. Epub 2024 May 3.
5
A model of conceptual bootstrapping in human cognition.
Nat Hum Behav. 2024 Jan;8(1):125-136. doi: 10.1038/s41562-023-01719-1. Epub 2023 Oct 16.
6
Eight-Month-Old Infants Meta-Learn by Downweighting Irrelevant Evidence.
Open Mind (Camb). 2023 Jun 1;7:141-155. doi: 10.1162/opmi_a_00079. eCollection 2023.
7
Estimations of child linguistic productivity controlling for vocabulary and sample size.
Front Psychol. 2022 Oct 18;13:996610. doi: 10.3389/fpsyg.2022.996610. eCollection 2022.
8
One model for the learning of language.
Proc Natl Acad Sci U S A. 2022 Feb 1;119(5). doi: 10.1073/pnas.2021865119.
9
Neural circuits and symbolic processing.
Neurobiol Learn Mem. 2021 Dec;186:107552. doi: 10.1016/j.nlm.2021.107552. Epub 2021 Nov 8.
10
The cultural origins of symbolic number.
Psychol Rev. 2022 Nov;129(6):1442-1456. doi: 10.1037/rev0000289. Epub 2021 Jun 17.

本文引用的文献

1
Solving Bongard Problems With a Visual Language and Pragmatic Constraints.
Cogn Sci. 2024 May;48(5):e13432. doi: 10.1111/cogs.13432.
2
The Child as Hacker.
Trends Cogn Sci. 2020 Nov;24(11):900-915. doi: 10.1016/j.tics.2020.07.005. Epub 2020 Oct 1.
3
Towards a rational constructivist theory of cognitive development.
Psychol Rev. 2019 Nov;126(6):841-864. doi: 10.1037/rev0000153. Epub 2019 Jun 10.
4
Internal Models in Biological Control.
Annu Rev Control Robot Auton Syst. 2019 May 1;2:339-364. doi: 10.1146/annurev-control-060117-105206.
5
Bayesian validation of grammar productions for the language of thought.
PLoS One. 2018 Jul 10;13(7):e0200420. doi: 10.1371/journal.pone.0200420. eCollection 2018.
6
Learning abstract visual concepts via probabilistic program induction in a Language of Thought.
Cognition. 2017 Nov;168:320-334. doi: 10.1016/j.cognition.2017.07.005. Epub 2017 Aug 1.
7
The language of geometry: Fast comprehension of geometrical primitives and rules in human adults and preschoolers.
PLoS Comput Biol. 2017 Jan 26;13(1):e1005273. doi: 10.1371/journal.pcbi.1005273. eCollection 2017 Jan.
8
Ducklings imprint on the relational concept of "same or different".
Science. 2016 Jul 15;353(6296):286-8. doi: 10.1126/science.aaf4247.
9
The emergence of reasoning by the disjunctive syllogism in early childhood.
Cognition. 2016 Sep;154:40-48. doi: 10.1016/j.cognition.2016.05.012. Epub 2016 May 28.
10
The logical primitives of thought: Empirical foundations for compositional cognitive models.
Psychol Rev. 2016 Jul;123(4):392-424. doi: 10.1037/a0039980. Epub 2016 Apr 14.

文献AI研究员

20分钟写一篇综述,助力文献阅读效率提升50倍。

立即体验

用中文搜PubMed

大模型驱动的PubMed中文搜索引擎

马上搜索

文档翻译

学术文献翻译模型,支持多种主流文档格式。

立即体验