Journal of memory and language最新文献_第9页

LLMs as models for analogical reasoning 法学硕士作为类比推理的模型

IF 3 1区心理学 Q1 LINGUISTICS

Journal of memory and language

Pub Date : 2025-07-31 DOI: 10.1016/j.jml.2025.104676

Sam Musker , Alex Duchnowski , Raphaël Millière , Ellie Pavlick

Analogical reasoning — the capacity to identify and map structural relationships between different domains — is fundamental to human cognition and learning. Recent studies have shown that large language models (LLMs) can sometimes match humans in analogical reasoning tasks, opening the possibility that analogical reasoning might emerge from domain-general processes. However, it is still debated whether these emergent capacities are largely superficial and limited to simple relations seen during training or whether they encompass the flexible representational and mapping capabilities which are the focus of leading cognitive models of analogy. In this study, we introduce novel analogical reasoning tasks that require participants to map between semantically contentful words and sequences of letters and other abstract characters. This task necessitates the ability to flexibly re-represent rich semantic information—an ability which is known to be central to human analogy but which is thus far not well-captured by existing cognitive theories and models. We assess the performance of both human participants and LLMs on tasks focusing on reasoning from semantic structure and semantic content, introducing variations that test the robustness of their analogical inferences. Advanced LLMs match human performance across several conditions, though humans and LLMs respond differently to certain task variations and semantic distractors. Our results thus provide new evidence that LLMs might offer a how-possibly explanation of human analogical reasoning in contexts that are not yet well modeled by existing theories, but that even today’s best models are unlikely to yield how-actually explanations.

类比推理——识别和绘制不同领域之间结构关系的能力——是人类认知和学习的基础。最近的研究表明，大型语言模型（llm）有时可以在类比推理任务中与人类相匹配，这开启了从领域通用过程中出现类比推理的可能性。然而，这些涌现能力是否很大程度上是肤浅的，局限于训练期间看到的简单关系，或者它们是否包含灵活的表征和映射能力，这是领先的类比认知模型的焦点，仍然存在争议。在这项研究中，我们引入了新的类比推理任务，要求参与者在语义内容的单词和字母序列以及其他抽象字符之间进行映射。这项任务需要灵活地重新表示丰富的语义信息的能力——一种已知的人类类比的核心能力，但迄今为止还没有被现有的认知理论和模型很好地捕获。我们评估了人类参与者和法学硕士在从语义结构和语义内容进行推理的任务上的表现，并引入了测试其类比推理鲁棒性的变化。尽管人类和llm对某些任务变化和语义干扰的反应不同，但高级llm在几种情况下的表现与人类相当。因此，我们的研究结果提供了新的证据，证明法学硕士可能会在现有理论尚未很好地建模的情况下为人类类比推理提供一种可能的解释，但即使是当今最好的模型也不太可能产生实际的解释。

{"title":"LLMs as models for analogical reasoning","authors":"Sam Musker , Alex Duchnowski , Raphaël Millière , Ellie Pavlick","doi":"10.1016/j.jml.2025.104676","DOIUrl":"10.1016/j.jml.2025.104676","url":null,"abstract":"<div><div>Analogical reasoning — the capacity to identify and map structural relationships between different domains — is fundamental to human cognition and learning. Recent studies have shown that large language models (LLMs) can sometimes match humans in analogical reasoning tasks, opening the possibility that analogical reasoning might emerge from domain-general processes. However, it is still debated whether these emergent capacities are largely superficial and limited to simple relations seen during training or whether they encompass the flexible representational and mapping capabilities which are the focus of leading cognitive models of analogy. In this study, we introduce novel analogical reasoning tasks that require participants to map between semantically contentful words and sequences of letters and other abstract characters. This task necessitates the ability to flexibly <em>re-represent</em> rich semantic information—an ability which is known to be central to human analogy but which is thus far not well-captured by existing cognitive theories and models. We assess the performance of both human participants and LLMs on tasks focusing on reasoning from semantic structure and semantic content, introducing variations that test the robustness of their analogical inferences. Advanced LLMs match human performance across several conditions, though humans and LLMs respond differently to certain task variations and semantic distractors. Our results thus provide new evidence that LLMs might offer a <em>how-possibly</em> explanation of human analogical reasoning in contexts that are not yet well modeled by existing theories, but that even today’s best models are unlikely to yield <em>how-actually</em> explanations.</div></div>","PeriodicalId":16493,"journal":{"name":"Journal of memory and language","volume":"145 ","pages":"Article 104676"},"PeriodicalIF":3.0,"publicationDate":"2025-07-31","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144749191","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"心理学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Memory for prediction: A Transformer-based theory of sentence processing 预测记忆：基于变换的句子处理理论

IF 2.9 1区心理学 Q1 LINGUISTICS

Journal of memory and language

Pub Date : 2025-07-25 DOI: 10.1016/j.jml.2025.104670

Soo Hyun Ryu , Richard L. Lewis

We demonstrate that Transformer-based neural network language models provide a new foundation for mechanistic theories of sentence processing that seamlessly integrate expectation-based and memory-based accounts. First, we show that the attention mechanism in GPT2-small operates as a kind of cue-based retrieval architecture that is subject to similarity-based interference. Second, we show that it provides accounts of classic memory effects in parsing, including contrasts involving relative clauses and center-embedding. Third, we show that a simple word-by-word entropy metric computed over the internal attention patterns provides an index of memory interference that explains variance in eye-tracking and self-paced reading time measures (independent of surprisal and other predictors) in two natural story reading time corpora. Because the cues and representations are learned, there is no need for the theorist to postulate representational features and cues. Transformers provide practical modeling tools for exploring the effects of memory and experience, given the increasing availability of both pre-trained models and software for training new models, and the ease with which surprisal and attention entropy metrics may be computed.

我们证明了基于transformer的神经网络语言模型为句子处理的机制理论提供了新的基础，该理论无缝地集成了基于期望和基于记忆的帐户。首先，我们发现GPT2-small中的注意机制是一种基于线索的检索架构，它会受到基于相似性的干扰。其次，我们表明它提供了解析中经典记忆效应的解释，包括涉及相对分句和中心嵌入的对比。第三，我们展示了在内部注意模式上计算的一个简单的逐字熵度量提供了一个记忆干扰指数，解释了在两个自然故事阅读时间语料库中眼动追踪和自定节奏阅读时间测量（独立于惊奇和其他预测因素）的差异。因为线索和表征是习得的，所以理论家不需要假设表征特征和线索。变形金刚为探索记忆和经验的影响提供了实用的建模工具，考虑到预训练模型和训练新模型的软件的可用性不断增加，并且可以轻松地计算惊喜和注意力熵度量。

{"title":"Memory for prediction: A Transformer-based theory of sentence processing","authors":"Soo Hyun Ryu , Richard L. Lewis","doi":"10.1016/j.jml.2025.104670","DOIUrl":"10.1016/j.jml.2025.104670","url":null,"abstract":"<div><div>We demonstrate that Transformer-based neural network language models provide a new foundation for mechanistic theories of sentence processing that seamlessly integrate expectation-based and memory-based accounts. First, we show that the attention mechanism in GPT2-small operates as a kind of cue-based retrieval architecture that is subject to similarity-based interference. Second, we show that it provides accounts of classic memory effects in parsing, including contrasts involving relative clauses and center-embedding. Third, we show that a simple word-by-word entropy metric computed over the internal attention patterns provides an index of memory interference that explains variance in eye-tracking and self-paced reading time measures (independent of surprisal and other predictors) in two natural story reading time corpora. Because the cues and representations are learned, there is no need for the theorist to postulate representational features and cues. Transformers provide practical modeling tools for exploring the effects of memory and experience, given the increasing availability of both pre-trained models and software for training new models, and the ease with which surprisal and attention entropy metrics may be computed.</div></div>","PeriodicalId":16493,"journal":{"name":"Journal of memory and language","volume":"145 ","pages":"Article 104670"},"PeriodicalIF":2.9,"publicationDate":"2025-07-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144703412","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"心理学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Reframing linguistic bootstrapping as joint inference using visually-grounded grammar induction models 用视觉基础语法归纳模型重构语言自举作为联合推理

IF 2.9 1区心理学 Q1 LINGUISTICS

Journal of memory and language

Pub Date : 2025-07-24 DOI: 10.1016/j.jml.2025.104672

Eva Portelance , Siva Reddy , Timothy J. O’Donnell

Semantic and syntactic bootstrapping posit that children use their prior knowledge of one linguistic domain, say syntactic relations, to help later acquire another, such as the meanings of new words. Empirical results supporting both theories may tempt us to believe that these are different independent learning strategies. Here, we argue for a unified approach, where instead they are both contingent on a more general learning strategy for language acquisition: joint learning. Using a series of neural visually-grounded grammar induction models, we demonstrate that both syntactic and semantic bootstrapping effects are strongest when syntax and semantics are learnt simultaneously via joint learning. This more general learning strategy results in better grammar induction, realistic lexical category learning, and better interpretations of novel sentence and verb meanings. Joint learning makes language acquisition easier for learners by mutually constraining the hypotheses spaces for both syntax and semantics. Studying the dynamics of joint inference over many input sources and modalities represents an important new direction for language modeling and learning research in both cognitive sciences and AI, as it may help us explain how language can be acquired in more constrained learning settings.

语义和句法引导假定儿童使用他们先前对一个语言领域的知识，比如句法关系，来帮助以后获得另一个语言领域的知识，比如新单词的意思。支持这两种理论的实证结果可能会诱使我们相信这是不同的独立学习策略。在这里，我们主张一种统一的方法，相反，它们都取决于一种更普遍的语言习得学习策略：联合学习。使用一系列基于神经视觉的语法归纳模型，我们证明了当语法和语义通过联合学习同时学习时，句法和语义的自举效应都是最强的。这种更普遍的学习策略有助于更好的语法归纳、现实的词汇类别学习以及对新句子和动词含义的更好解释。联合学习通过对句法和语义假设空间的相互约束，使学习者更容易习得语言。研究多种输入源和模态的联合推理动态是认知科学和人工智能中语言建模和学习研究的一个重要新方向，因为它可以帮助我们解释语言是如何在更有限的学习环境中习得的。

{"title":"Reframing linguistic bootstrapping as joint inference using visually-grounded grammar induction models","authors":"Eva Portelance , Siva Reddy , Timothy J. O’Donnell","doi":"10.1016/j.jml.2025.104672","DOIUrl":"10.1016/j.jml.2025.104672","url":null,"abstract":"<div><div>Semantic and syntactic bootstrapping posit that children use their prior knowledge of one linguistic domain, say syntactic relations, to help later acquire another, such as the meanings of new words. Empirical results supporting both theories may tempt us to believe that these are different independent learning strategies. Here, we argue for a unified approach, where instead they are both contingent on a more general learning strategy for language acquisition: joint learning. Using a series of neural visually-grounded grammar induction models, we demonstrate that both syntactic and semantic bootstrapping effects are strongest when syntax and semantics are learnt simultaneously via joint learning. This more general learning strategy results in better grammar induction, realistic lexical category learning, and better interpretations of novel sentence and verb meanings. Joint learning makes language acquisition <em>easier</em> for learners by mutually constraining the hypotheses spaces for both syntax and semantics. Studying the dynamics of joint inference over many input sources and modalities represents an important new direction for language modeling and learning research in both cognitive sciences and AI, as it may help us explain how language can be acquired in more constrained learning settings.</div></div>","PeriodicalId":16493,"journal":{"name":"Journal of memory and language","volume":"145 ","pages":"Article 104672"},"PeriodicalIF":2.9,"publicationDate":"2025-07-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144694551","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"心理学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Type and token frequency jointly drive learning of morphology 词型和词频共同驱动词法学习

IF 3 1区心理学 Q1 LINGUISTICS

Journal of memory and language

Pub Date : 2025-07-22 DOI: 10.1016/j.jml.2025.104666

Gaja Jarosz , Cerys Hughes , Andrew Lamont , Brandon Prickett , Maggie Baird , Seoyoung Kim , Max Nelson

We examine the joint roles of type frequency and token frequency in three artificial language learning experiments involving lexicalized plural allomorphy. The primary role of type frequency in productivity is well-established, but debates about the precise relationship between type frequency and productivity continue. The effect of token frequency on productivity is even more controversial: some lines of research suggest token frequency and productivity are inversely related, other results indicate they are positively related, and yet others argue token frequency plays no role in productivity. We address both of these questions. Our learning framework makes it possible to examine the effects of these variables on generalization to novel forms and to examine how sensitivity to these factors affects the time-course of learning. The first two experiments differentiate predictions for generalization of three distinct hypotheses about the role of type frequency, while the third experiment investigates the independent role of token frequency. We find that both type and token frequency independently and positively contribute to learning rates and generalization across the three experiments. We also apply two computational learning theories – implementing two prominent theoretical linguistic frameworks – to the learning of the lexically-conditioned allomorphy patterns in our experiments. Despite their differences, we show that the incremental learning dynamics of both models correctly predict the general trends in generalization rates, learning curves, and the influence of token frequency observed across the experimental conditions.

在三个涉及词汇化复数异型的人工语言学习实验中，研究了类型频率和标记频率的联合作用。类型频率在生产力中的主要作用是公认的，但关于类型频率和生产力之间确切关系的争论仍在继续。代币频率对生产力的影响更具争议性：一些研究表明代币频率和生产力呈负相关，其他结果表明它们呈正相关，还有一些人认为代币频率对生产力没有作用。我们解决了这两个问题。我们的学习框架可以检验这些变量对新形式泛化的影响，并检验对这些因素的敏感性如何影响学习的时间过程。前两个实验区分了关于类型频率作用的三个不同假设的泛化预测，而第三个实验研究了标记频率的独立作用。我们发现类型和标记频率对三个实验的学习率和泛化都有独立和积极的贡献。我们还应用了两种计算学习理论-实现了两种突出的理论语言学框架-在我们的实验中学习词汇条件异形模式。尽管它们存在差异，但我们表明，两种模型的增量学习动态都正确地预测了在实验条件下观察到的泛化率、学习曲线和标记频率影响的总体趋势。

{"title":"Type and token frequency jointly drive learning of morphology","authors":"Gaja Jarosz , Cerys Hughes , Andrew Lamont , Brandon Prickett , Maggie Baird , Seoyoung Kim , Max Nelson","doi":"10.1016/j.jml.2025.104666","DOIUrl":"10.1016/j.jml.2025.104666","url":null,"abstract":"<div><div>We examine the joint roles of type frequency and token frequency in three artificial language learning experiments involving lexicalized plural allomorphy. The primary role of type frequency in productivity is well-established, but debates about the precise relationship between type frequency and productivity continue. The effect of token frequency on productivity is even more controversial: some lines of research suggest token frequency and productivity are inversely related, other results indicate they are positively related, and yet others argue token frequency plays no role in productivity. We address both of these questions. Our learning framework makes it possible to examine the effects of these variables on generalization to novel forms and to examine how sensitivity to these factors affects the time-course of learning. The first two experiments differentiate predictions for generalization of three distinct hypotheses about the role of type frequency, while the third experiment investigates the independent role of token frequency. We find that both type and token frequency independently and positively contribute to learning rates and generalization across the three experiments. We also apply two computational learning theories – implementing two prominent theoretical linguistic frameworks – to the learning of the lexically-conditioned allomorphy patterns in our experiments. Despite their differences, we show that the incremental learning dynamics of both models correctly predict the general trends in generalization rates, learning curves, and the influence of token frequency observed across the experimental conditions.</div></div>","PeriodicalId":16493,"journal":{"name":"Journal of memory and language","volume":"144 ","pages":"Article 104666"},"PeriodicalIF":3.0,"publicationDate":"2025-07-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144723277","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"心理学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Learning filler-gap dependencies with neural language models: Testing island sensitivity in Norwegian and English 用神经语言模型学习填空依赖性：测试挪威语和英语的岛屿敏感性

IF 2.9 1区心理学 Q1 LINGUISTICS

Journal of memory and language

Pub Date : 2025-07-20 DOI: 10.1016/j.jml.2025.104663

Anastasia Kobzeva , Suhas Arehalli , Tal Linzen , Dave Kush

Human linguistic input is often claimed to be impoverished with respect to linguistic evidence for complex structural generalizations that children induce. The field of language acquisition is currently debating the ability of various learning algorithms to accurately derive target generalizations from the input. A growing body of research explores whether Neural Language Models (NLMs) can induce human-like generalizations about filler-gap dependencies (FGDs) in English, including island constraints on their distribution. Based on positive results for select test cases, some authors have argued that the relevant generalizations can be learned without domain-specific learning biases (Wilcox et al., 2023), though other researchers dispute this conclusion ((Lan et al., 2024b; Howitt et al.,2024). Previous work focuses solely on English, but broader claims about filler-gap dependency learnability can only be made based on multiple languages and dependency types. To address this gap, we compare the ability of NLMs to learn restrictions on FGDs in English and Norwegian. Our results are mixed: they show that although these models acquire some sophisticated generalizations about filler-gap dependencies in the two languages, their generalizations still diverge from those of humans. When tested on structurally complex environments, the models sometimes adopt narrower generalizations than humans do or overgeneralize beyond their input in non-human-like ways. We conclude that current evidence does not support the claim that FGDs and island constraints on them can be learned without domain-specific biases.

人类的语言输入通常被认为是贫乏的，相对于儿童诱导的复杂结构概括的语言证据。语言习得领域目前正在讨论各种学习算法能否准确地从输入中获得目标泛化。越来越多的研究探索神经语言模型（NLMs）是否可以诱导出英语中填充-间隙依赖关系（fgd）的类人概括，包括它们分布的岛约束。基于选择测试用例的积极结果，一些作者认为可以在没有特定领域学习偏差的情况下学习相关的概括（Wilcox et al., 2023），尽管其他研究人员对这一结论提出异议(Lan et al., 2024b；Howitt et al.,2024)。以前的工作只关注于英语，但是关于填空依赖的可学习性的更广泛的主张只能基于多种语言和依赖类型。为了解决这一差距，我们比较了nlm在英语和挪威语中学习FGDs限制的能力。我们的结果是混合的：它们表明，尽管这些模型获得了关于两种语言中填充-间隙依赖关系的一些复杂的概括，但它们的概括仍然与人类的不同。当在结构复杂的环境中进行测试时，模型有时会采用比人类更狭窄的泛化，或者以非人类的方式过度泛化。我们的结论是，目前的证据并不支持在没有特定领域偏见的情况下可以学习fgd和岛屿约束的说法。

{"title":"Learning filler-gap dependencies with neural language models: Testing island sensitivity in Norwegian and English","authors":"Anastasia Kobzeva , Suhas Arehalli , Tal Linzen , Dave Kush","doi":"10.1016/j.jml.2025.104663","DOIUrl":"10.1016/j.jml.2025.104663","url":null,"abstract":"<div><div>Human linguistic input is often claimed to be impoverished with respect to linguistic evidence for complex structural generalizations that children induce. The field of language acquisition is currently debating the ability of various learning algorithms to accurately derive target generalizations from the input. A growing body of research explores whether Neural Language Models (NLMs) can induce human-like generalizations about filler-gap dependencies (FGDs) in English, including island constraints on their distribution. Based on positive results for select test cases, some authors have argued that the relevant generalizations can be learned without domain-specific learning biases (Wilcox et al., 2023), though other researchers dispute this conclusion ((Lan et al., 2024b; Howitt et al.,2024). Previous work focuses solely on English, but broader claims about filler-gap dependency learnability can only be made based on multiple languages and dependency types. To address this gap, we compare the ability of NLMs to learn restrictions on FGDs in English and Norwegian. Our results are mixed: they show that although these models acquire some sophisticated generalizations about filler-gap dependencies in the two languages, their generalizations still diverge from those of humans. When tested on structurally complex environments, the models sometimes adopt narrower generalizations than humans do or overgeneralize beyond their input in non-human-like ways. We conclude that current evidence does not support the claim that FGDs and island constraints on them can be learned without domain-specific biases.</div></div>","PeriodicalId":16493,"journal":{"name":"Journal of memory and language","volume":"144 ","pages":"Article 104663"},"PeriodicalIF":2.9,"publicationDate":"2025-07-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144665627","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"心理学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Exploring the animacy effect in focal prospective memory tasks: When animates don’t stand out 探究焦点前瞻记忆任务中的动画效果：当动画不突出时

IF 2.9 1区心理学 Q1 LINGUISTICS

Journal of memory and language

Pub Date : 2025-07-18 DOI: 10.1016/j.jml.2025.104673

Sara B. Félix , Marie Poirier , Josefa N.S. Pandeirada

The animacy effect refers to a memory advantage for animates/living beings as compared to inanimates/nonliving things. So far, the animacy effect has been investigated mostly in retrospective memory. Given that memory serves a future-oriented function, and considering the adaptive significance of animacy, it has been proposed that it should also confer an advantage in prospective memory (i.e., memory for intentions/actions to-be-performed in the future). Recent research reported an animacy effect in nonfocal event-based prospective memory tasks. The present work explored this effect in focal prospective memory. In a series of five studies, conducted in different countries and languages, we employed various ongoing tasks. Across all studies, no differences in prospective memory performance between animates and inanimates were found. This result held in a sign-test including all participants (N = 408 young adults) for a more powered analysis. Also, no differences between animates and inanimates were obtained in the baseline and filler trials. These results are discussed considering the mechanisms that have been proposed to explain the effect in retrospective memory tasks, namely attention-prioritization and richness of encoding. Overall, our results are partially explained by the attention-prioritization account of the animacy effect and also provide support for the Multiprocess Framework.

有生命效应是指有生命/生物相对于无生命/非生物的记忆优势。到目前为止，动画效应的研究主要集中在回顾性记忆中。鉴于记忆服务于面向未来的功能，并考虑到动画的适应性意义，有人提出它也应该赋予前瞻性记忆（即对未来将要执行的意图/行动的记忆）的优势。最近的研究报道了在非焦点事件为基础的前瞻性记忆任务中的动画效应。本研究在焦点前瞻记忆中探讨了这种效应。在用不同国家和语言进行的五项系列研究中，我们采用了各种正在进行的任务。在所有的研究中，没有发现动物和非动物在前瞻记忆表现上的差异。这一结果在包括所有参与者（N = 408名年轻人）的符号测试中得到了更有力的分析。此外，在基线和填充试验中，动物和非动物之间没有差异。这些结果讨论了已提出的机制，以解释在回顾记忆任务的影响，即注意优先和编码的丰富性。总的来说，我们的结果部分地解释了动画效应的注意优先级，也为多进程框架提供了支持。

{"title":"Exploring the animacy effect in focal prospective memory tasks: When animates don’t stand out","authors":"Sara B. Félix , Marie Poirier , Josefa N.S. Pandeirada","doi":"10.1016/j.jml.2025.104673","DOIUrl":"10.1016/j.jml.2025.104673","url":null,"abstract":"<div><div>The animacy effect refers to a memory advantage for animates/living beings as compared to inanimates/nonliving things. So far, the animacy effect has been investigated mostly in retrospective memory. Given that memory serves a future-oriented function, and considering the adaptive significance of animacy, it has been proposed that it should also confer an advantage in prospective memory (i.e., memory for intentions/actions to-be-performed in the future). Recent research reported an animacy effect in nonfocal event-based prospective memory tasks. The present work explored this effect in focal prospective memory. In a series of five studies, conducted in different countries and languages, we employed various ongoing tasks. Across all studies, no differences in prospective memory performance between animates and inanimates were found. This result held in a sign-test including all participants (<em>N</em> = 408 young adults) for a more powered analysis. Also, no differences between animates and inanimates were obtained in the baseline and filler trials. These results are discussed considering the mechanisms that have been proposed to explain the effect in retrospective memory tasks, namely attention-prioritization and richness of encoding. Overall, our results are partially explained by the attention-prioritization account of the animacy effect and also provide support for the Multiprocess Framework.</div></div>","PeriodicalId":16493,"journal":{"name":"Journal of memory and language","volume":"144 ","pages":"Article 104673"},"PeriodicalIF":2.9,"publicationDate":"2025-07-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144656880","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"心理学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

The development of the adaptive use of different forms of rehearsal in verbal serial recall tasks. A multi-method study 不同形式的排练在言语序列回忆任务中的适应性运用。多方法研究

IF 2.9 1区心理学 Q1 LINGUISTICS

Journal of memory and language

Pub Date : 2025-07-17 DOI: 10.1016/j.jml.2025.104674

Sebastian Poloczek , Christopher Jarrold

Verbal rehearsal is a key feature of certain working memory models that have previously assumed that children develop adult-like rehearsal around the age of 7. However, a broader literature indicates that younger children are capable of rehearsal. The present study, consisting of two experiments with 191 primary school children in total, combined methods that are rarely used to study rehearsal in serial recall. Self-paced presentation times were obtained as a behavioural indicator of strategy use. On half of trials, children additionally reported their strategies via think-aloud (Expt. 1) or immediate trial-by trial self-reports (Expt. 1 & 2). Results from the three methods employed in Experiment 1 with 10- to 11-year-olds converged on the conclusion that multiple strategies were used across trials. Listening, single rehearsal, and cumulative rehearsal were common strategies that were validly reported with no or only small effects of reactivity of strategy reporting. Experiment 2 revealed that between the ages of 6 to 11 years children employed a range of strategies across trials. Listening without rehearsal was common and cumulative rehearsal rare among the younger children, but cumulative rehearsal and strategy adaptivity to list length gradually increased with age. Importantly, self-reports were corroborated by self-presentation times even in younger children. We conclude that rehearsal development does not follow a stage-like progression. Rather, the data support an overlapping waves model as several strategies coexist, the likelihood of using a strategy changes gradually, and adaptivity of strategy choices still improves among older children.

语言排练是某些工作记忆模型的一个关键特征，这些模型先前假设儿童在7岁左右发展出像成人一样的排练能力。然而，更广泛的文献表明，年幼的孩子有能力排练。本研究以191名小学生为研究对象，采用两种常用的方法对连续记忆中的排练进行研究。自定的陈述时间被作为策略使用的行为指标。在一半的试验中，孩子们通过大声思考（见图1）或直接一次又一次的自我报告（见图1）来报告他们的策略。2）. 实验1中对10- 11岁儿童采用的三种方法的结果一致表明，在不同的试验中使用了多种策略。聆听、单一排练和累积排练是有效报告的常见策略，对策略报告的反应性没有或只有很小的影响。实验2显示，6至11岁的儿童在不同的试验中采用了一系列的策略。在低龄儿童中，不排练的听是常见的，累积排练是罕见的，但累积排练和对列表长度的策略适应性随着年龄的增长而逐渐增加。重要的是，即使在年幼的孩子中，自我报告也得到了自我呈现次数的证实。我们的结论是，排练的发展并不遵循一个阶段式的进展。相反，数据支持重叠波模型，因为几种策略共存，使用策略的可能性逐渐变化，并且策略选择的适应性在较大的儿童中仍然提高。

{"title":"The development of the adaptive use of different forms of rehearsal in verbal serial recall tasks. A multi-method study","authors":"Sebastian Poloczek , Christopher Jarrold","doi":"10.1016/j.jml.2025.104674","DOIUrl":"10.1016/j.jml.2025.104674","url":null,"abstract":"<div><div>Verbal rehearsal is a key feature of certain working memory models that have previously assumed that children develop adult-like rehearsal around the age of 7. However, a broader literature indicates that younger children are capable of rehearsal. The present study, consisting of two experiments with 191 primary school children in total, combined methods that are rarely used to study rehearsal in serial recall. Self-paced presentation times were obtained as a behavioural indicator of strategy use. On half of trials, children additionally reported their strategies via think-aloud (Expt. 1) or immediate trial-by trial self-reports (Expt. 1 & 2). Results from the three methods employed in Experiment 1 with 10- to 11-year-olds converged on the conclusion that multiple strategies were used across trials. Listening, single rehearsal, and cumulative rehearsal were common strategies that were validly reported with no or only small effects of reactivity of strategy reporting. Experiment 2 revealed that between the ages of 6 to 11 years children employed a range of strategies across trials. Listening without rehearsal was common and cumulative rehearsal rare among the younger children, but cumulative rehearsal and strategy adaptivity to list length gradually increased with age. Importantly, self-reports were corroborated by self-presentation times even in younger children. We conclude that rehearsal development does not follow a stage-like progression. Rather, the data support an overlapping waves model as several strategies coexist, the likelihood of using a strategy changes gradually, and adaptivity of strategy choices still improves among older children.</div></div>","PeriodicalId":16493,"journal":{"name":"Journal of memory and language","volume":"144 ","pages":"Article 104674"},"PeriodicalIF":2.9,"publicationDate":"2025-07-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144656879","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"心理学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

IF 2.9 1区心理学 Q1 LINGUISTICS

Journal of memory and language

Pub Date : 2025-07-16 DOI: 10.1016/j.jml.2025.104669

Hailin Hao , Zuzanna Fuchs , Shravan Vasishth

During the processing of linguistic dependencies, the presence of a non-dependent word—referred to as a distractor—can sometimes complicate the identification of the correct subject. This phenomenon, known as similarity-based interference, provides a valuable testing ground for competing theories of sentence processing and has garnered significant interest in the field of psycholinguistics. One prominent theory, cue-based retrieval, suggests that the parser initiates a search for the relevant linguistic dependent at the retrieval site (e.g., the verb) based on a set of retrieval cues. In this work, we explore the use of lexicon-specific cues set by classifiers in the retrieval of noun dependents in Mandarin Chinese to provide evidence for the cue-based retrieval mechanism. A further open question is whether the distractor must intervene between the co-dependents (so-called retroactive interference) or whether the distractor can appear to the left of the dependent elements (so-called proactive interference). Previous work has suggested that proactive interference is weaker than retroactive interference, i.e., that the distractor has to intervene between the co-dependents to influence the dependency completion process. Using self-paced reading and A-Maze tasks, and Bayes Factors for hypothesis testing, we found robust evidence for a predicted interference effect in retroactive configurations, but no interference in proactive configurations. We discuss the theoretical implications of the current work for theories of retrieval and sentence processing in general.

在语言依赖的处理过程中，非依赖词（即干扰物）的存在有时会使正确主语的识别复杂化。这种现象被称为基于相似性的干扰，为句子处理的竞争理论提供了一个有价值的试验场，并在心理语言学领域引起了极大的兴趣。一个突出的理论是基于线索的检索，它建议解析器根据一组检索线索在检索站点（例如，动词）启动对相关语言依赖的搜索。在本研究中，我们探讨了分类器在汉语名词依存词检索中的词典特定线索的使用，为基于线索的检索机制提供证据。一个进一步开放的问题是，干扰物是否必须在相互依赖的元素之间进行干预（所谓的追溯干扰），或者干扰物是否可以出现在依赖元素的左侧（所谓的主动干扰）。先前的研究表明，主动干扰比追溯干扰弱，即干扰物必须在共同依赖者之间进行干预，以影响依赖完成过程。使用自定节奏阅读和迷宫任务，并使用贝叶斯因子进行假设检验，我们发现在回溯配置中预测干扰效应的有力证据，但在主动配置中没有干扰。我们讨论了当前工作对检索和句子处理理论的理论意义。

{"title":"Similarity-based interference in the processing of classifier-noun dependencies in Mandarin Chinese","authors":"Hailin Hao , Zuzanna Fuchs , Shravan Vasishth","doi":"10.1016/j.jml.2025.104669","DOIUrl":"10.1016/j.jml.2025.104669","url":null,"abstract":"<div><div>During the processing of linguistic dependencies, the presence of a non-dependent word—referred to as a distractor—can sometimes complicate the identification of the correct subject. This phenomenon, known as similarity-based interference, provides a valuable testing ground for competing theories of sentence processing and has garnered significant interest in the field of psycholinguistics. One prominent theory, cue-based retrieval, suggests that the parser initiates a search for the relevant linguistic dependent at the retrieval site (e.g., the verb) based on a set of retrieval cues. In this work, we explore the use of lexicon-specific cues set by classifiers in the retrieval of noun dependents in Mandarin Chinese to provide evidence for the cue-based retrieval mechanism. A further open question is whether the distractor must intervene between the co-dependents (so-called retroactive interference) or whether the distractor can appear to the left of the dependent elements (so-called proactive interference). Previous work has suggested that proactive interference is weaker than retroactive interference, i.e., that the distractor has to intervene between the co-dependents to influence the dependency completion process. Using self-paced reading and A-Maze tasks, and Bayes Factors for hypothesis testing, we found robust evidence for a predicted interference effect in retroactive configurations, but no interference in proactive configurations. We discuss the theoretical implications of the current work for theories of retrieval and sentence processing in general.</div></div>","PeriodicalId":16493,"journal":{"name":"Journal of memory and language","volume":"144 ","pages":"Article 104669"},"PeriodicalIF":2.9,"publicationDate":"2025-07-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144633035","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"心理学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Initially encoding attended but outdated information into working memory: behavioral and neural evidence 最初是将参与但过时的信息编码到工作记忆中：行为和神经证据

IF 2.9 1区心理学 Q1 LINGUISTICS

Journal of memory and language

Pub Date : 2025-06-23 DOI: 10.1016/j.jml.2025.104668

Chunqian Xiao , Yihong Luo , Yuan Gao , Jiejie Liao , Mengxia Yu , Lei Mo

Attention has traditionally been regarded as a gateway to working memory, largely determining whether information enters it. Recent work suggests that the brain actively inhibits attended but outdated information to prevent it from entering working memory. However, it remains unknown whether this information is blocked directly by attention before entering working memory, or after being encoded into working memory, given that such information has already been attended to and processed. This study explored this question by manipulating stimulus onset asynchronies (SOAs) in three experiments, including behavioral and electroencephalography (EEG) measures, and examining memory traces of attended but outdated information at different time points. Behavioral evidence demonstrated the stability of the memory trace of the attended but outdated information only when SOA was short. This finding was observed across different features and paradigms. Time-frequency analysis indicated that the brain inhibited attention to information matching the attended but outdated information in the early stage, with behavioral performance predicted by alpha modulation of the right hemisphere. These results suggest that attended but outdated information is initially encoded into working memory, even though it does not need to be remembered. These findings enhance our understanding of the impact of attention on working memory.

传统上，注意力被认为是进入工作记忆的门户，在很大程度上决定了信息是否进入工作记忆。最近的研究表明，大脑会主动抑制已参与但过时的信息，以防止其进入工作记忆。然而，目前尚不清楚这些信息是在进入工作记忆之前被注意力直接阻止，还是在被编码到工作记忆之后，因为这些信息已经被注意和处理了。本研究通过三个实验来探讨这一问题，包括行为和脑电图（EEG）测量，以及在不同时间点检查参与但过时信息的记忆痕迹。行为证据表明，只有在SOA很短的情况下，参与但过时的信息的内存跟踪才具有稳定性。这一发现是在不同的特征和范式中观察到的。时间-频率分析表明，大脑在早期阶段抑制了对与已注意但已过时的信息相匹配的信息的注意，其行为表现由右半球的α调制预测。这些结果表明，被关注但过时的信息最初被编码到工作记忆中，即使它不需要被记住。这些发现增强了我们对注意力对工作记忆影响的理解。

{"title":"Initially encoding attended but outdated information into working memory: behavioral and neural evidence","authors":"Chunqian Xiao , Yihong Luo , Yuan Gao , Jiejie Liao , Mengxia Yu , Lei Mo","doi":"10.1016/j.jml.2025.104668","DOIUrl":"10.1016/j.jml.2025.104668","url":null,"abstract":"<div><div>Attention has traditionally been regarded as a gateway to working memory, largely determining whether information enters it. Recent work suggests that the brain actively inhibits attended but outdated information to prevent it from entering working memory. However, it remains unknown whether this information is blocked directly by attention before entering working memory, or after being encoded into working memory, given that such information has already been attended to and processed. This study explored this question by manipulating stimulus onset asynchronies (SOAs) in three experiments, including behavioral and electroencephalography (EEG) measures, and examining memory traces of attended but outdated information at different time points. Behavioral evidence demonstrated the stability of the memory trace of the attended but outdated information only when SOA was short. This finding was observed across different features and paradigms. Time-frequency analysis indicated that the brain inhibited attention to information matching the attended but outdated information in the early stage, with behavioral performance predicted by alpha modulation of the right hemisphere. These results suggest that attended but outdated information is initially encoded into working memory, even though it does not need to be remembered. These findings enhance our understanding of the impact of attention on working memory.</div></div>","PeriodicalId":16493,"journal":{"name":"Journal of memory and language","volume":"144 ","pages":"Article 104668"},"PeriodicalIF":2.9,"publicationDate":"2025-06-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144366673","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"心理学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Judgments of learning enhance elaborative rather than relational processing: Implications from phonologically related and phonological-semantic mediated pairs 学习判断促进了精细加工而不是关系加工：来自语音相关和语音语义介导对的启示

IF 2.9 1区心理学 Q1 LINGUISTICS

Journal of memory and language

Pub Date : 2025-06-21 DOI: 10.1016/j.jml.2025.104667

Minyu Chang , C.J. Brainerd

Judgment of learning (JOL) reactivity refers to the effect that making JOLs directly impacts subsequent memory performance. One of the most replicated findings in this line of research is that JOLs enhance memory for related word pairs. However, so far, only semantic relatedness has been studied, and many existing theories of JOL reactivity, such as the cue-strengthening hypothesis and the enhanced relational processing account, are heavily dependent on findings generated with semantically related word pairs. The current study used phonologically related pairs instead of semantically related pairs. Specifically, we used rhyme pairs (e.g., fall-tall) in Experiments 1A and 1B and homophone pairs (e.g., coarse-course) in Experiments 2A, 2B, and 2C. A consistent pattern emerged that JOLs did not produce reactivity for phonologically related pairs on associative recall tests, unlike for semantically related pairs. This supports the hypothesis that JOL reactivity reflects an enhancement in elaborative processing, which presumably can involve either item-specific or relational processing that focuses on deeper, semantic content, rather than relational processing that focuses on relational content of any nature. In Experiment 3, we found positive JOL reactivity for phonological-semantic mediated pairs (e.g., coarse-class, where coarse is phonologically related to an unpresented mediator course that is semantically related to class). This, in contrast to the null reactivity for pure phonological relatedness, again supports the necessity of semantic processing in positive JOL reactivity. We discuss how the elaborative processing account offers an opportunity to reconcile existing theoretical explanations and help build a general framework for JOL reactivity.

学习判断（JOL）反应性是指做出JOL直接影响随后的记忆表现。在这方面的研究中，最被重复的发现之一是，JOLs可以增强对相关单词对的记忆。然而，到目前为止，只有语义相关性得到了研究，许多现有的JOL反应性理论，如线索强化假说和增强的关系处理说，都严重依赖于语义相关词对产生的结果。目前的研究使用语音相关对而不是语义相关对。具体来说，我们在实验1A和1B中使用了押韵对（例如，fall-tall），在实验2A、2B和2C中使用了同音对（例如，rough -course）。一个一致的模式出现，JOLs不产生反应性的语音相关对联想回忆测试，不像语义相关对。这支持了一种假设，即JOL反应性反映了精细化处理的增强，精细化处理可能涉及特定于项目的处理，也可能涉及侧重于更深层次语义内容的关系处理，而不是侧重于任何性质的关系内容的关系处理。在实验3中，我们发现语音-语义中介对的JOL反应性为正（例如，粗类，其中粗类在语音上与未呈现的中介课程相关，而在语义上与类别相关）。这与纯粹语音相关性的零反应性相反，再次支持了积极JOL反应性中语义加工的必要性。我们讨论了精细加工解释如何提供一个调和现有理论解释的机会，并帮助建立JOL反应性的一般框架。

{"title":"Judgments of learning enhance elaborative rather than relational processing: Implications from phonologically related and phonological-semantic mediated pairs","authors":"Minyu Chang , C.J. Brainerd","doi":"10.1016/j.jml.2025.104667","DOIUrl":"10.1016/j.jml.2025.104667","url":null,"abstract":"<div><div>Judgment of learning (JOL) reactivity refers to the effect that making JOLs directly impacts subsequent memory performance. One of the most replicated findings in this line of research is that JOLs enhance memory for related word pairs. However, so far, only semantic relatedness has been studied, and many existing theories of JOL reactivity, such as the cue-strengthening hypothesis and the enhanced relational processing account, are heavily dependent on findings generated with semantically related word pairs. The current study used phonologically related pairs instead of semantically related pairs. Specifically, we used rhyme pairs (e.g., <em>fall-tall</em>) in Experiments 1A and 1B and homophone pairs (e.g., <em>coarse-course</em>) in Experiments 2A, 2B, and 2C. A consistent pattern emerged that JOLs did not produce reactivity for phonologically related pairs on associative recall tests, unlike for semantically related pairs. This supports the hypothesis that JOL reactivity reflects an enhancement in elaborative processing, which presumably can involve either item-specific or relational processing that focuses on deeper, semantic content, rather than relational processing that focuses on relational content of any nature. In Experiment 3, we found positive JOL reactivity for phonological-semantic mediated pairs (e.g., <em>coarse-class</em>, where <em>coarse</em> is phonologically related to an unpresented mediator <em>course</em> that is semantically related to <em>class</em>). This, in contrast to the null reactivity for pure phonological relatedness, again supports the necessity of semantic processing in positive JOL reactivity. We discuss how the elaborative processing account offers an opportunity to reconcile existing theoretical explanations and help build a general framework for JOL reactivity.</div></div>","PeriodicalId":16493,"journal":{"name":"Journal of memory and language","volume":"144 ","pages":"Article 104667"},"PeriodicalIF":2.9,"publicationDate":"2025-06-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144329594","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"心理学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0