Cognitive Science最新文献

This study investigated cross-cultural differences in visual attention patterns during comic reading, focusing on participants with Japanese and American cultural backgrounds. Using an eye-tracking paradigm, we examined attention processes as participants viewed pages from American comics and Japanese manga featuring objective or subjective viewpoints. The results showed that for objective pages, American readers exhibited relatively longer fixations on focal objects, while Japanese readers allocated relatively more attention to backgrounds, aligning with analytic versus holistic cognitive styles. By contrast, for subjective materials, Japanese readers demonstrated greater attention to focal objects than American readers did, suggesting that the subjective perspective embedded in manga shifts Japanese readers toward a focal-object-oriented attentional style. Individual differences in self-reported analytic-holistic cognitive styles and manga reading experience, in addition to cultural background, were associated with attentional patterns for manga. The results underscore the influence of artistic design in shaping visual attention in ways that both mirror and transcend culturally ingrained attentional biases. This study deepens our understanding of cross-cultural variations in visual processing and comic reading behaviors, providing fresh insights into the complex interplay among culture, cognition, and visual narrative comprehension.

The majority of research in computational psycholinguistics on sentence processing has focused on word-by-word incremental processing within sentences, rather than holistic sentence-level representations. This study introduces two novel computational approaches for quantifying sentence-level processing: sentence surprisal and sentence relevance. Using multilingual large language models (LLMs), we compute sentence surprisal through three methods, chain rule, next sentence prediction, and negative log-likelihood, and apply a “memory-aware” approach to calculate sentence-level semantic relevance based on convolution operations. The sentence-level metrics developed are tested and compared to validate whether they can predict the reading speed of sentences, and, further, we explore how sentence-level metrics take effects on human processing and comprehending sentences as a whole across languages. The results show that sentence-level metrics are highly capable of predicting sentence reading speed. Our results also indicate that these computational sentence-level metrics are exceptionally effective at predicting and explaining the processing difficulties encountered by readers in processing sentences as a whole across a variety of languages. The proposed sentence-level metrics offer significant interpretability and achieve high accuracy in predicting human sentence reading speed, as they capture unique aspects of comprehension difficulty beyond word-level measures. These metrics serve as valuable computational tools for investigating human sentence processing and advancing our understanding of naturalistic reading. Their strong performance and generalization capabilities highlight their potential to drive progress at the intersection of LLMs and cognitive science.

Five-to-six-year-olds’ abilities to detect and solve ambiguities in spoken language have been found to be a predictor of their later reading abilities in first-to-third grade. However, the origins of this relationship remain unclear. Success in ambiguity detection may be reflective of overall language attainment, which varies with socioeconomic status (SES) and is known to predict reading. Yet, it is also possible that children's ability to detect ambiguity is explained by domain-general cognitive control skills, which can also vary with SES and predict literacy attainment. In this cross-sectional study, we examined within the same children the contributions of overall language knowledge, SES, and cognitive control skills to their ability to detect ambiguities in speech. Five-to-six-year-old French-learning preschoolers (n = 38) performed three different tasks: ambiguity detection, a cognitive control (Flanker/No-Go) task, and standard assessments of vocabulary and oral language comprehension in French (BSEDS). Years of maternal education after the end of high school were used as a proxy of family SES. Individual differences in the ability to detect ambiguity were strongly related to children's cognitive control abilities, as indexed by congruency effects in the Flanker task. No relations with SES or language assessment were observed. These results lend support to the idea that children's reading development may hinge upon their ability to deal effectively with temporary lexical, syntactic, and semantic ambiguities that pervade real-time sentence interpretation and that their ability to deal with representational conflict in speech is reflective of their domain-general cognitive control skills.

It has been repeatedly found that idioms are processed faster than syntactically matched literal phrases, in both comprehension and production. This has led to debate about whether idioms are accessed as chunks or built compositionally, with different studies attempting to measure the effect of compositionality on processing, with differing conclusions. This paper looks at idiom processing through the lens of information update, in particular surprisal theory, which is a standard theory of sentence processing. Compositionality is just one aspect of a word's predictability; we argue that surprisal, as an expectation-based theory, provides a more general unifying framework for understanding the idiom processing advantage. In this paper, comprehension and production experiments on verb-object idioms reveal that the idiom processing advantage can be largely explained by the fact that idioms have lower surprisal than matched literal phrases. The results indicate that the idiom advantage manifests primarily on the noun in verb-object idioms.

Sentence production is a stage-like process of mapping a conceptual representation to the linear speech signal via grammatical rules. While the typological diversity of languages is vast and thus must necessarily influence sentence production, psycholinguistic studies of diverse languages are comparatively rare. Here, we present data from a sentence planning and production study in Pitjantjatjara, an Australian Indigenous language that has highly flexible word order. Forty-nine (N = 49) native speakers described pictures of two-participant scenes while their eye-movements were recorded. Participants produced all possible orders of agent, patient, and verb. There was a general preference to produce agent-initial orders, but word order was influenced by the semantic properties of agent and patient referents (± human). Analyses of participants’ eye-movements revealed early relational encoding of the entire event, whereby speakers distributed their attention between agent and patient referents in a manner that is different than typically observed in languages that have more restricted word order options. Relational encoding was influenced by the word order that participants eventually produced. The results provide evidence to suggest that sentence planning in Pitjantjatjara is a hierarchical process, in which early relational encoding creates a wholistic conceptualization of an event, possibly driven by pressure to decide upon one of many possible word orders.

People use many kinds of cues that help them navigate social interactions. We examined how perceived foreignness affected people's ability to map speaker-specific naming preferences, align with their interlocutors concerning these preferences, and make social inferences based on them. In a pseudo-interactive experiment, participants engaged with two simulated speakers: one with a common native name who consistently used favored words, and one who consistently used the disfavored alternatives, and had either a native name, a foreign name associated with positive stereotypes (American), or a foreign name associated with negative stereotypes (Former Soviet Union; FSU). We assessed participants’ tendencies to align with each speaker's lexical choices, their ability to generalize disfavored lexical use to other sorts of language use, and the social inferences they drew about each speaker. Results showed that perceived foreignness modulated both linguistic alignment and social judgments. The alignment effect was larger for FSU and native speakers compared to the American speakers. Interestingly, this stemmed from the increased tendency to use the disfavored words with the common native speaker when the uncommon speaker was American, suggesting that speakers’ nationality modulated words’ perceived disfavoredness. Further, generalizations about social traits (e.g., cooperativeness) varied by nationality, with American speakers rated more positively despite similar linguistic behaviors. These findings reveal that foreignness-associated stereotypes can modulate the social consequences of language use, suggesting a bidirectional dynamics where social identity both shapes language processing and is shaped by it. This extends theories of social meaning by demonstrating how social expectations conditionally interact with linguistic behaviors.

Reading comprehension has been mostly studied using traditional texts and very little is known about reading comprehension in comics. We wanted to find out whether comics could enhance comprehension processes, compared to traditional text and what cognitive processes might be involved in this effect. Furthermore, we explored the functional role of pictures in understanding comics. Forty skilled readers read the comic and text versions of two already published stories and answered comprehension questions. Eye movements were recorded during reading. We found no differences in reading comprehension performance. However, comics were explored faster than traditional texts. Importantly, reading speed of words in balloons was faster than in traditional texts. An analysis of eye movements suggests that the presence of pictures facilitates the extraction of information, with shorter total saccadic amplitude on the pages of comics than in text. When reading comics, participants spent less time on the pictures than the balloons, and this behavior was associated with shorter and fewer fixations. Pictures were also used as an entry point for reading a panel, as the first fixation in the panel fell more frequently on the pictures and the readers returned to them more often than to the balloons. Because pictures are processed faster than words, they may be used to construct a first representation of the content of the story, which can be used to facilitate the processing of the whole story and, more specifically, of its verbal component. This strategy is not available in traditional texts.

Can metaphorical reasoning involving embodied experience—such as color perception—be learned from the statistics of language alone? Recent work finds that colorblind individuals robustly understand and reason abstractly about color, implying that color associations in everyday language might contribute to the metaphorical understanding of color. However, it is unclear how much colorblind individuals’ understanding of color is driven by language versus their limited (but no less embodied) visual experience. A more direct test of whether language supports the acquisition of humans’ understanding of color is whether large language models (LLMs)—those trained purely on text with no visual experience—can nevertheless learn to generate consistent and coherent metaphorical responses about color. Here, we conduct preregistered surveys that compare colorseeing adults, colorblind adults, and LLMs in how they (1) associate colors to words that lack established color associations and (2) interpret conventional and novel color metaphors. Colorblind and colorseeing adults exhibited highly similar and replicable color associations with novel words and abstract concepts. Yet, while GPT (a popular LLM) also generated replicable color associations with impressive consistency, its associations departed considerably from colorseeing and colorblind participants. Moreover, GPT frequently failed to generate coherent responses about its own metaphorical color associations when asked to invert its color associations or explain novel color metaphors in context. Consistent with this view, painters who regularly work with color pigments were more likely than all other groups to understand novel color metaphors using embodied reasoning. Thus, embodied experience may play an important role in metaphorical reasoning about color and the generation of conceptual connections between embodied associations.

Human thinking has long been posited to involve two different cognitive processes, also known as intuition and deliberation. While deliberation is effortful and cognitively costly, intuition is effortless. A central issue for reasoning theories is to account for the trigger of deliberation. Compelling theories explain the trigger of deliberative processes by the existence of a metacognitive experience. A feeling of rightness, of error, or of uncertainty would accompany our intuitions and, depending on their strength, triggers the need to use deliberation. Despite the emotional component that can be assumed in these metacognitive phenomena, and a whole literature linking emotion to cognition, these models do not fully embrace the emotional nature of these experiences, both empirically and theoretically. We believe that the psychology of reasoning, and particularly dual-process theories, would benefit from fully accepting this emotional dimension of reasoning.

Social interaction depends on turn-taking and adapting to one's conversational partner, yet little is known about the typical and atypical development of these abilities. We investigated this in a longitudinal corpus of spontaneous speech in 64 parent–child dyads: 32 typically developing children (20.27 months at start, six girls, 24 White) and 32 with autism (linguistically matched, 32.76 months, four girls, 31 White). Contrary to prior studies, children with autism responded 189 ms faster on average than typically developing children due to more overlapping speech. Latency decreased in both groups (47–78 ms every 4 months) and depended on individual differences in socio-cognitive, linguistic, and motor skills, which for autism explained all variance by age. Both groups equally adapted their tempo to their interlocutors. With robust conceptualization and modeling techniques, we highlight the importance of overlapping speech, show that latencies in autism might be faster than in typical development and situate turn-taking into fine-grained developmental and interpersonal contexts.