1
|
Rule JS, Piantadosi ST, Cropper A, Ellis K, Nye M, Tenenbaum JB. Symbolic metaprogram search improves learning efficiency and explains rule learning in humans. Nat Commun 2024; 15:6847. [PMID: 39127796 DOI: 10.1038/s41467-024-50966-x] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [MESH Headings] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 01/25/2024] [Accepted: 07/23/2024] [Indexed: 08/12/2024] Open
Abstract
Throughout their lives, humans seem to learn a variety of rules for things like applying category labels, following procedures, and explaining causal relationships. These rules are often algorithmically rich but are nonetheless acquired with minimal data and computation. Symbolic models based on program learning successfully explain rule-learning in many domains, but performance degrades quickly as program complexity increases. It remains unclear how to scale symbolic rule-learning methods to model human performance in challenging domains. Here we show that symbolic search over the space of metaprograms-programs that revise programs-dramatically improves learning efficiency. On a behavioral benchmark of 100 algorithmically rich rules, this approach fits human learning more accurately than alternative models while also using orders of magnitude less search. The computation required to match median human performance is consistent with conservative estimates of human thinking time. Our results suggest that metaprogram-like representations may help human learners to efficiently acquire rules.
Collapse
Affiliation(s)
- Joshua S Rule
- Psychology, University of California, Berkeley, Berkeley, CA, 94704, USA.
| | | | | | - Kevin Ellis
- Computer Science, Cornell University, Ithaca, NY, 14850, USA
| | - Maxwell Nye
- Adept AI Labs, San Francisco, CA, 94110, USA
| | - Joshua B Tenenbaum
- Brain and Cognitive Sciences, Massachusetts Institute of Technology, Cambridge, MA, 02139, USA
| |
Collapse
|
2
|
Dupre G. Acquiring a language vs. inducing a grammar. Cognition 2024; 247:105771. [PMID: 38507944 DOI: 10.1016/j.cognition.2024.105771] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 09/13/2023] [Revised: 03/11/2024] [Accepted: 03/12/2024] [Indexed: 03/22/2024]
Abstract
Standard computational models of language acquisition treat acquiring a language as a process of inducing a set of string-generating rules from a collection of linguistic data assumed to be generated by these very rules. In this paper I give theoretical and empirical arguments that such a model is radically unlike what a human language learner must do to acquire their native language. Most centrally, I argue that such models presuppose that linguistic data is directly a product of a grammar, ignoring the myriad non-grammatical systems involved in the use of language. The significance of these non-target systems in shaping the linguistic data children are exposed to undermines any simple reverse inference from linguistic data to grammatical competence.
Collapse
Affiliation(s)
- Gabe Dupre
- Department of Philosophy, University of California, Davis, United States of America.
| |
Collapse
|
3
|
Koplenig A, Wolfer S. Languages with more speakers tend to be harder to (machine-)learn. Sci Rep 2023; 13:18521. [PMID: 37898699 PMCID: PMC10613286 DOI: 10.1038/s41598-023-45373-z] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Key Words] [MESH Headings] [Grants] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 08/24/2023] [Accepted: 10/18/2023] [Indexed: 10/30/2023] Open
Abstract
Computational language models (LMs), most notably exemplified by the widespread success of OpenAI's ChatGPT chatbot, show impressive performance on a wide range of linguistic tasks, thus providing cognitive science and linguistics with a computational working model to empirically study different aspects of human language. Here, we use LMs to test the hypothesis that languages with more speakers tend to be easier to learn. In two experiments, we train several LMs-ranging from very simple n-gram models to state-of-the-art deep neural networks-on written cross-linguistic corpus data covering 1293 different languages and statistically estimate learning difficulty. Using a variety of quantitative methods and machine learning techniques to account for phylogenetic relatedness and geographical proximity of languages, we show that there is robust evidence for a relationship between learning difficulty and speaker population size. However, contrary to expectations derived from previous research, our results suggest that languages with more speakers tend to be harder to learn.
Collapse
Affiliation(s)
| | - Sascha Wolfer
- Leibniz Institute for the German Language (IDS), Mannheim, Germany
| |
Collapse
|
4
|
Valle-Lisboa JC, Pomi A, Mizraji E. Multiplicative processing in the modeling of cognitive activities in large neural networks. Biophys Rev 2023; 15:767-785. [PMID: 37681105 PMCID: PMC10480136 DOI: 10.1007/s12551-023-01074-5] [Citation(s) in RCA: 1] [Impact Index Per Article: 1.0] [Reference Citation Analysis] [Abstract] [Key Words] [Grants] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 03/27/2023] [Accepted: 06/04/2023] [Indexed: 09/09/2023] Open
Abstract
Explaining the foundation of cognitive abilities in the processing of information by neural systems has been in the beginnings of biophysics since McCulloch and Pitts pioneered work within the biophysics school of Chicago in the 1940s and the interdisciplinary cybernetists meetings in the 1950s, inseparable from the birth of computing and artificial intelligence. Since then, neural network models have traveled a long path, both in the biophysical and the computational disciplines. The biological, neurocomputational aspect reached its representational maturity with the Distributed Associative Memory models developed in the early 70 s. In this framework, the inclusion of signal-signal multiplication within neural network models was presented as a necessity to provide matrix associative memories with adaptive, context-sensitive associations, while greatly enhancing their computational capabilities. In this review, we show that several of the most successful neural network models use a form of multiplication of signals. We present several classical models that included such kind of multiplication and the computational reasons for the inclusion. We then turn to the different proposals about the possible biophysical implementation that underlies these computational capacities. We pinpoint the important ideas put forth by different theoretical models using a tensor product representation and show that these models endow memories with the context-dependent adaptive capabilities necessary to allow for evolutionary adaptation to changing and unpredictable environments. Finally, we show how the powerful abilities of contemporary computationally deep-learning models, inspired in neural networks, also depend on multiplications, and discuss some perspectives in view of the wide panorama unfolded. The computational relevance of multiplications calls for the development of new avenues of research that uncover the mechanisms our nervous system uses to achieve multiplication.
Collapse
Affiliation(s)
- Juan C. Valle-Lisboa
- Group of Cognitive Systems Modeling, Biophysics and Systems Biology Section, Facultad de Ciencias, Universidad de la República, Iguá 4225, 11400 Montevideo, Uruguay
- Centro Interdisciplinario en Cognición para la Enseñanza y el Aprendizaje (CICEA), Universidad de la República, Espacio Interdisciplinario, 11200 Montevideo, Uruguay
| | - Andrés Pomi
- Group of Cognitive Systems Modeling, Biophysics and Systems Biology Section, Facultad de Ciencias, Universidad de la República, Iguá 4225, 11400 Montevideo, Uruguay
| | - Eduardo Mizraji
- Group of Cognitive Systems Modeling, Biophysics and Systems Biology Section, Facultad de Ciencias, Universidad de la República, Iguá 4225, 11400 Montevideo, Uruguay
| |
Collapse
|
5
|
Pavlick E. Symbols and grounding in large language models. PHILOSOPHICAL TRANSACTIONS. SERIES A, MATHEMATICAL, PHYSICAL, AND ENGINEERING SCIENCES 2023; 381:20220041. [PMID: 37271171 PMCID: PMC10239679 DOI: 10.1098/rsta.2022.0041] [Citation(s) in RCA: 8] [Impact Index Per Article: 8.0] [Reference Citation Analysis] [Abstract] [Key Words] [MESH Headings] [Track Full Text] [Subscribe] [Scholar Register] [Received: 02/02/2023] [Accepted: 04/20/2023] [Indexed: 06/06/2023]
Abstract
Large language models (LLMs) are one of the most impressive achievements of artificial intelligence in recent years. However, their relevance to the study of language more broadly remains unclear. This article considers the potential of LLMs to serve as models of language understanding in humans. While debate on this question typically centres around models' performance on challenging language understanding tasks, this article argues that the answer depends on models' underlying competence, and thus that the focus of the debate should be on empirical work which seeks to characterize the representations and processing algorithms that underlie model behaviour. From this perspective, the article offers counterarguments to two commonly cited reasons why LLMs cannot serve as plausible models of language in humans: their lack of symbolic structure and their lack of grounding. For each, a case is made that recent empirical trends undermine the common assumptions about LLMs, and thus that it is premature to draw conclusions about LLMs' ability (or lack thereof) to offer insights on human language representation and understanding. This article is part of a discussion meeting issue 'Cognitive artificial intelligence'.
Collapse
Affiliation(s)
- Ellie Pavlick
- Department of Computer Science, Brown University, Providence, RI, USA
| |
Collapse
|
6
|
Dedhe AM, Piantadosi ST, Cantlon JF. Cognitive Mechanisms Underlying Recursive Pattern Processing in Human Adults. Cogn Sci 2023; 47:e13273. [PMID: 37051878 PMCID: PMC11097651 DOI: 10.1111/cogs.13273] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Key Words] [MESH Headings] [Grants] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 06/23/2022] [Revised: 02/24/2023] [Accepted: 03/05/2023] [Indexed: 04/14/2023]
Abstract
The capacity to generate recursive sequences is a marker of rich, algorithmic cognition, and perhaps unique to humans. Yet, the precise processes driving recursive sequence generation remain mysterious. We investigated three potential cognitive mechanisms underlying recursive pattern processing: hierarchical reasoning, ordinal reasoning, and associative chaining. We developed a Bayesian mixture model to quantify the extent to which these three cognitive mechanisms contribute to adult humans' performance in a sequence generation task. We further tested whether recursive rule discovery depends upon relational information, either perceptual or semantic. We found that the presence of relational information facilitates hierarchical reasoning and drives the generation of recursive sequences across novel depths of center embedding. In the absence of relational information, the use of ordinal reasoning predominates. Our results suggest that hierarchical reasoning is an important cognitive mechanism underlying recursive pattern processing and can be deployed across embedding depths and relational domains.
Collapse
Affiliation(s)
- Abhishek M Dedhe
- Department of Psychology, Carnegie Mellon University
- Center for the Neural Basis of Cognition, Carnegie Mellon University
| | | | - Jessica F Cantlon
- Department of Psychology, Carnegie Mellon University
- Center for the Neural Basis of Cognition, Carnegie Mellon University
| |
Collapse
|
7
|
Dedhe AM, Clatterbuck H, Piantadosi ST, Cantlon JF. Origins of Hierarchical Logical Reasoning. Cogn Sci 2023; 47:e13250. [PMID: 36739520 PMCID: PMC11057913 DOI: 10.1111/cogs.13250] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Key Words] [MESH Headings] [Grants] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 09/30/2022] [Revised: 12/21/2022] [Accepted: 01/06/2023] [Indexed: 02/06/2023]
Abstract
Hierarchical cognitive mechanisms underlie sophisticated behaviors, including language, music, mathematics, tool-use, and theory of mind. The origins of hierarchical logical reasoning have long been, and continue to be, an important puzzle for cognitive science. Prior approaches to hierarchical logical reasoning have often failed to distinguish between observable hierarchical behavior and unobservable hierarchical cognitive mechanisms. Furthermore, past research has been largely methodologically restricted to passive recognition tasks as compared to active generation tasks that are stronger tests of hierarchical rules. We argue that it is necessary to implement learning studies in humans, non-human species, and machines that are analyzed with formal models comparing the contribution of different cognitive mechanisms implicated in the generation of hierarchical behavior. These studies are critical to advance theories in the domains of recursion, rule-learning, symbolic reasoning, and the potentially uniquely human cognitive origins of hierarchical logical reasoning.
Collapse
Affiliation(s)
- Abhishek M. Dedhe
- Department of Psychology, Carnegie Mellon University
- Center for the Neural Basis of Cognition, Carnegie Mellon University
| | | | | | - Jessica F. Cantlon
- Department of Psychology, Carnegie Mellon University
- Center for the Neural Basis of Cognition, Carnegie Mellon University
| |
Collapse
|
8
|
Ritonga M, Hakim R, Nurdianto T, Ritonga AW. Learning for early childhood using the IcanDO platform: Breakthroughs for golden age education in Arabic learning. EDUCATION AND INFORMATION TECHNOLOGIES 2023; 28:1-18. [PMID: 36688217 PMCID: PMC9838446 DOI: 10.1007/s10639-022-11575-7] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Figures] [Subscribe] [Scholar Register] [Received: 09/14/2022] [Accepted: 12/29/2022] [Indexed: 06/17/2023]
Abstract
Early childhood is an age of children whose learning tendencies rely on games, therefore this study aims to describe the IcanDO platform for early childhood in learning Arabic. The research was conducted with a qualitative approach, the data sources were early childhood teachers. Data collected by observation techniques, interviews and documentation studies. The result of the research is that IcanDO as a learning platform is interesting for early childhood, with this platform they can play to learn Arabic. In accordance with the data, it was found that the IcanDO platform used can stimulate early childhood learning, stronger memory, personalized learning can be implemented, children's thinking skills can be trained and children's multilingual abilities are also growing. Researchers recommend that the use of various platforms that support the implementation of education in difficult conditions continue to be developed. The limitation of this research is that IcanDO and its effectiveness have not been studied comprehensively, therefore the aspects that have not been touched in this research can be the work of future researchers.
Collapse
Affiliation(s)
| | - Rosniati Hakim
- Muhammadiyah University of West Sumatra, Padang, Indonesia
| | | | | |
Collapse
|
9
|
Algorithms of adaptation in inductive inference. Cogn Psychol 2022; 137:101506. [PMID: 35872374 DOI: 10.1016/j.cogpsych.2022.101506] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 05/11/2021] [Revised: 06/01/2022] [Accepted: 07/08/2022] [Indexed: 11/20/2022]
Abstract
We investigate the idea that human concept inference utilizes local adaptive search within a compositional mental theory space. To explore this, we study human judgments in a challenging task that involves actively gathering evidence about a symbolic rule governing the behavior of a simulated environment. Participants learn by performing mini-experiments before making generalizations and explicit guesses about a hidden rule. They then collect additional evidence themselves (Experiment 1) or observe evidence gathered by someone else (Experiment 2) before revising their own generalizations and guesses. In each case, we focus on the relationship between participants' initial and revised guesses about the hidden rule concept. We find an order effect whereby revised guesses are anchored to idiosyncratic elements of the earlier guess. To explain this pattern, we develop a family of process accounts that combine program induction ideas with local (MCMC-like) adaptation mechanisms. A particularly local variant of this adaptive account captures participants' hypothesis revisions better than a range of alternative explanations. We take this as suggestive that people deal with the inherent complexity of concept inference partly through use of local adaptive search in a latent compositional theory space.
Collapse
|
10
|
Another model not for the learning of language. Proc Natl Acad Sci U S A 2022; 119:e2204664119. [PMID: 35787184 PMCID: PMC9303892 DOI: 10.1073/pnas.2204664119] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Track Full Text] [Download PDF] [Figures] [Journal Information] [Subscribe] [Scholar Register] [Indexed: 11/18/2022] Open
|
11
|
Reply to Kodner et al.: Fundamental misunderstanding of both model and methods. Proc Natl Acad Sci U S A 2022; 119:e2204944119. [PMID: 35787183 PMCID: PMC9303956 DOI: 10.1073/pnas.2204944119] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Track Full Text] [Download PDF] [Figures] [Journal Information] [Subscribe] [Scholar Register] [Indexed: 11/18/2022] Open
|
12
|
Reply to Murphy and Leivada: Program induction can learn language. Proc Natl Acad Sci U S A 2022; 119:e2202925119. [PMID: 35648825 PMCID: PMC9191657 DOI: 10.1073/pnas.2202925119] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Track Full Text] [Download PDF] [Journal Information] [Subscribe] [Scholar Register] [Indexed: 11/25/2022] Open
|
13
|
|