Situational systematicity: A role for schema in understanding the differences between abstract and concrete concepts

Abstract concepts and expertise: the case of institutional concepts

Recent views recognize that abstract concepts encompass a variety of exemplars, each relying on different dimensions, including not only sensorimotor but also inner, linguistic, and social experiences. How these dimensions characterize types of abstract concepts, and whether their weight varies across contexts and individuals remains an open question. We investigated the role of linguistic and social situations in the processing of institutional concepts, such as justice, by individuals with different levels of expertise. In a priming study, legal experts and non-experts were asked to respond to target words (go-trials) consisting of different kinds of abstract (institutional, theoretical) and concrete concepts (food, tools) and to ignore filler words (no-go trials). The verbal stimuli were primed by pictures depicting social-action, linguistic-social, linguistic-textual situations, and a control condition. As predicted, critical priming modulated performance on abstract concepts, likely due to their highly context-dependent meaning. Interestingly, we found that the processing of institutional concepts was selectively facilitated by social action prime, suggesting that this situational content may be integrated to support their representation. Crucially, the dialogic context, the linguistic social prime, affected more non-experts than law-experts, who tended to frame institutional concepts as shared idea for regulating social practices. Our results show that linguistic and social inputs become differently salient for institutional concepts representation depending on individual competence.

Symbol ungrounding: what the successes (and failures) of large language models reveal about human cognition

Large language models can handle sophisticated natural language processing tasks. This raises the question of how their understanding of semantic meaning compares to that of human beings. Supporters of embodied cognition often point out that because these models are trained solely on text, their representations of semantic content are not grounded in sensorimotor experience. This paper contends that human cognition exhibits capabilities that fit with both the embodied and artificial intelligence approaches. Evidence suggests that semantic memory is partially grounded in sensorimotor systems and dependent on language-specific learning. From this perspective, large language models demonstrate the richness of language as a source of semantic information. They show how our experience with language might scaffold and extend our capacity to make sense of the world. In the context of an embodied mind, language provides access to a valuable form of ungrounded cognition.This article is part of the theme issue 'Minds in movement: embodied cognition in the age of artificial intelligence'.

Language with vision: A study on grounded word and sentence embeddings

Grounding language in vision is an active field of research seeking to construct cognitively plausible word and sentence representations by incorporating perceptual knowledge from vision into text-based representations. Despite many attempts at language grounding, achieving an optimal equilibrium between textual representations of the language and our embodied experiences remains an open field. Some common concerns are the following. Is visual grounding advantageous for abstract words, or is its effectiveness restricted to concrete words? What is the optimal way of bridging the gap between text and vision? To what extent is perceptual knowledge from images advantageous for acquiring high-quality embeddings? Leveraging the current advances in machine learning and natural language processing, the present study addresses these questions by proposing a simple yet very effective computational grounding model for pre-trained word embeddings. Our model effectively balances the interplay between language and vision by aligning textual embeddings with visual information while simultaneously preserving the distributional statistics that characterize word usage in text corpora. By applying a learned alignment, we are able to indirectly ground unseen words including abstract words. A series of evaluations on a range of behavioral datasets shows that visual grounding is beneficial not only for concrete words but also for abstract words, lending support to the indirect theory of abstract concepts. Moreover, our approach offers advantages for contextualized embeddings, such as those generated by BERT (Devlin et al, 2018), but only when trained on corpora of modest, cognitively plausible sizes. Code and grounded embeddings for English are available at ( https://github.com/Hazel1994/Visually_Grounded_Word_Embeddings_2 ).

How does the "default mode" network contribute to semantic cognition?

This review examines whether and how the "default mode" network (DMN) contributes to semantic processing. We review evidence implicating the DMN in the processing of individual word meanings and in sentence- and discourse-level semantics. Next, we argue that the areas comprising the DMN contribute to semantic processing by coordinating and integrating the simultaneous activity of local neuronal ensembles across multiple unimodal and multimodal cortical regions, creating a transient, global neuronal ensemble. The resulting ensemble implements an integrated simulation of phenomenological experience - that is, an embodied situation model - constructed from various modalities of experiential memory traces. These situation models, we argue, are necessary not only for semantic processing but also for aspects of cognition that are not traditionally considered semantic. Although many aspects of this proposal remain provisional, we believe it provides new insights into the relationships between semantic and non-semantic cognition and into the functions of the DMN.

Grounded Cognition, Linguistic Relativity, and Abstract Concepts

Kemmerer's paper convincingly claims that the grounded cognition model (GCM) entails linguistic relativity. Here, we underline that tackling linguistic relativity and cultural differences is vital for GCM. First, it allows GCM to focus more on flexible rather than stable aspects of cognition. Second, it highlights the centrality of linguistic experience for human cognition. While GCM-inspired research underscored the similarity between linguistic and nonlinguistic concepts, it is now paramount to understand when and how language(s) influence knowledge. To this aim, we argue that linguistic variation might be particularly relevant for more abstract concepts-which are more debatable and open to revisions.

Is time an embodied property of concepts?

A haircut usually lasts under an hour. But how long does it take to recognize that something is an instance of a haircut? And is this "time-to-perceive" a part of the representation of concepts like haircut? Across three experiments testing lexical decision, word recognition, and semantic decision, we show that the amount of time people say it takes to perceive a concept in the world (e.g., haircut, dandelion, or merit) predicts how long it takes for them to respond to a word referring to that thing, over and above the effects of other lexical-semantic variables (e.g., word frequency, concreteness) and other variables related to conceptual complexity (e.g., how confusable a concept is with other, similar concepts, or the diversity of the contexts in which a concept appears). These results suggest that our experience of how long it takes to recognize an instance of a concept can become a part of its representation, and that we simulate this information when reading words. Consequently, we suggest that time may be an embodied property of concepts.

Semantic Bimodal Presentation Differentially Slows Working Memory Retrieval

Although evidence has shown that working memory (WM) can be differentially affected by the multisensory congruency of different visual and auditory stimuli, it remains unclear whether different multisensory congruency about concrete and abstract words could impact further WM retrieval. By manipulating the attention focus toward different matching conditions of visual and auditory word characteristics in a 2-back paradigm, the present study revealed that for the characteristically incongruent condition under the auditory retrieval condition, the response to abstract words was faster than that to concrete words, indicating that auditory abstract words are not affected by visual representation, while auditory concrete words are. Alternatively, for concrete words under the visual retrieval condition, WM retrieval was faster in the characteristically incongruent condition than in the characteristically congruent condition, indicating that visual representation formed by auditory concrete words may interfere with WM retrieval of visual concrete words. The present findings demonstrated that concrete words in multisensory conditions may be too aggressively encoded with other visual representations, which would inadvertently slow WM retrieval. However, abstract words seem to suppress interference better, showing better WM performance than concrete words in the multisensory condition.

Leveraging Prior Knowledge to Support Short-term Memory: Exploring the Role of the Ventromedial Prefrontal Cortex

It is well established that the ventromedial prefrontal cortex (vmPFC) plays a critical role in memory consolidation and the retrieval of remote long-term memories. Recent evidence suggests that the vmPFC also supports rapid neocortical learning and consolidation over shorter timescales, particularly when novel events align with stored knowledge. One mechanism by which the vmPFC has been proposed to support this learning is by integrating congruent information into existing neocortical knowledge during memory encoding. An important outstanding question is whether the vmPFC also plays a critical role in linking congruent information with existing knowledge before storage in long-term memory. The current study investigated this question by testing whether lesions to the vmPFC disrupt the ability to leverage stored knowledge in support of short-term memory. Specifically, we investigated the visuospatial bootstrapping effect, the phenomenon whereby immediate verbal recall of visually presented stimuli is better when stimuli appear in a familiar visuospatial array that is congruent with prior knowledge compared with an unfamiliar visuospatial array. We found that the overall magnitude of the bootstrapping effect did not differ between patients with vmPFC lesions and controls. However, a reliable bootstrapping effect was not present in the patient group alone. Post hoc analysis of individual patient performance revealed that the bootstrapping effect did not differ from controls in nine patients but was reduced in two patients. Although mixed, these results suggest that vmPFC lesions do not uniformly disrupt the ability to leverage stored knowledge in support of short-term memory.

A context-sensitive and non-linguistic approach to abstract concepts

Despite the recent upsurge in research on abstract concepts, there remain puzzles at the foundation of their empirical study. These are most evident when we consider what is required to assess a person's abstract conceptual abilities without using language as a prompt or requiring it as a response-as in classic non-verbal categorization tasks, which are standardly considered tests of conceptual understanding. After distinguishing two divergent strands in the most common conception of what it is for a concept to be abstract, we argue that neither reliably captures the kind of abstraction required to successfully categorize in non-verbal tasks. We then present a new conception of concept abstractness-termed 'trial concreteness'-that is keyed to individual categorization trials. It has advantages in capturing the context-relativity of the degree of abstraction required for the application of a concept and fittingly correlates with participant success in recent experiments. This article is part of the theme issue 'Concepts in interaction: social engagement and inner experiences'.

Acquiring religious words: dialogical and individual construction of a word's meaning

By the age of eight, there is a significant increase in abstract words in the child's lexicon. A crucial contribution can be seen in the linguistic input, i.e. the way how abstract words are presented by caregivers by means of linguistic perspectivation and emotionalization. Following an interactionist way, we were interested in how the semantics of abstract words is constructed by child and caregiver in duet. We focused on a subset of abstract words and studied the acquisition of meaning of the religious concept mercy. We expected religious words to be emotionally anchored and presented with perspectivation, both contributing to learning. Exploring the dialogic constructions, we investigated eight 7- to 8-year olds and their parents during dialogic reading and studied their strategies focusing on the linguistic means of emotionalization and perspectivation in contextualizing the word. In a subsequent test, we analysed these means used by the children and assessed their individual understanding of mercy. Our analyses indicate that during reading, the enrichment of semantics by emotionalization was related between child and caregiver, whereas cross-situationally, a simultaneous enrichment of emotionalization and perspectivation was present. Moreover, the children demonstrated a conceptual understanding of mercy in religious contexts, but not in secular contexts. This article is part of the theme issue 'Concepts in interaction: social engagement and inner experiences'.

Abstract and concrete concepts in conversation

Concepts allow us to make sense of the world. Most evidence on their acquisition and representation comes from studies of single decontextualized words and focuses on the opposition between concrete and abstract concepts (e.g., "bottle" vs. "truth"). A significant step forward in research on concepts consists in investigating them in online interaction during their use. Our study examines linguistic exchanges analyzing the differences between sub-kinds of concepts. Participants were submitted to an online task in which they had to simulate a conversational exchange by responding to sentences involving sub-kinds of concrete (tools, animals, food) and abstract concepts (PS, philosophical-spiritual; EMSS, emotional-social, PSTQ, physical-spatio-temporal-quantitative). We found differences in content: foods evoked interoception; tools and animals elicited materials, spatial, auditive features, confirming their sensorimotor grounding. PS and EMSS yielded inner experiences (e.g., emotions, cognitive states, introspections) and opposed PSTQ, tied to visual properties and concrete agency. More crucially, the various concepts elicited different interactional dynamics: more abstract concepts generated higher uncertainty and more interactive exchanges than concrete ones. Investigating concepts in situated interactions opens new possibilities for studying conceptual knowledge and its pragmatic and social aspects.

Effects of social experience on abstract concepts in semantic priming

Humans can understand thousands of abstract words, even when they do not have clearly perceivable referents. Recent views highlight an important role of social experience in grounding of abstract concepts and sub-kinds of abstract concepts, but empirical work in this area is still in its early stages. In the present study, a picture-word semantic priming paradigm was employed to investigate the contribution effect of social experience that is provided by real-life pictures to social abstract (SA, e.g., friendship, betrayal) concepts and emotional abstract (EA, e.g., happiness, anger) concepts. Using a lexical decision task, we examined responses to picture-SA word pairs (Experiment 1) and picture-EA word pairs (Experiment 2) in social/emotional semantically related and unrelated conditions. All pairs shared either positive or negative valence. The results showed quicker responses to positive SA and EA words that were preceded by related vs. unrelated prime pictures. Specifically, positive SA words were facilitated by the corresponding social scene pictures, whereas positive EA words were facilitated by pictures depict the corresponding facial expressions and gestures. However, such facilitatory effect was not observed in negative picture-SA/EA word conditions. This pattern of results suggests that a facilitatory effect of social experience on abstract concepts varies with different sub-kinds of abstract concepts, that seems to be limited to positive SA concepts. Overall, our findings confirm the crucial role of social experience for abstract concepts and further suggest that not all abstract concepts can benefit from social experience, at least in the semantic priming.

Modelling concrete and abstract concepts using brain-constrained deep neural networks

A neurobiologically constrained deep neural network mimicking cortical areas relevant for sensorimotor, linguistic and conceptual processing was used to investigate the putative biological mechanisms underlying conceptual category formation and semantic feature extraction. Networks were trained to learn neural patterns representing specific objects and actions relevant to semantically 'ground' concrete and abstract concepts. Grounding sets consisted of three grounding patterns with neurons representing specific perceptual or action-related features; neurons were either unique to one pattern or shared between patterns of the same set. Concrete categories were modelled as pattern triplets overlapping in their 'shared neurons', thus implementing semantic feature sharing of all instances of a category. In contrast, abstract concepts had partially shared feature neurons common to only pairs of category instances, thus, exhibiting family resemblance, but lacking full feature overlap. Stimulation with concrete and abstract conceptual patterns and biologically realistic unsupervised learning caused formation of strongly connected cell assemblies (CAs) specific to individual grounding patterns, whose neurons were spread out across all areas of the deep network. After learning, the shared neurons of the instances of concrete concepts were more prominent in central areas when compared with peripheral sensorimotor ones, whereas for abstract concepts the converse pattern of results was observed, with central areas exhibiting relatively fewer neurons shared between pairs of category members. We interpret these results in light of the current knowledge about the relative difficulty children show when learning abstract words. Implications for future neurocomputational modelling experiments as well as neurobiological theories of semantic representation are discussed.

Encoding and inhibition of arbitrary episodic context with abstract concepts

Context is critical for conceptual processing, but the mechanism underpinning its encoding and reinstantiation during abstract concept processing is unclear. Context may be especially important for abstract concepts-we investigated whether episodic context is recruited differently when processing abstract compared with concrete concepts. Experiments 1 and 2 presented abstract and concrete words in arbitrary contexts at encoding (Experiment 1: red/green colored frames; Experiment 2: male/female voices). Recognition memory for these contexts was worse for abstract concepts. Again using frame color and voice as arbitrary contexts, respectively, Experiments 3 and 4 presented words from encoding in the same or different context at test to determine whether there was a greater recognition memory benefit for abstract versus concrete concepts when the context was unchanged between encoding and test. Instead, abstract concepts were less likely to be remembered when context was retained. This suggests that at least some types of episodic context-when arbitrary-are attended less, and may even be inhibited, when processing abstract concepts. In Experiment 5, we utilized a context-spatial location-which (as we show) tends to be relevant during real-world processing of abstract concepts. We presented words in different locations, preserving or changing location at test. Location retention conferred a recognition memory advantage for abstract concepts. Thus, episodic context may be encoded with abstract concepts when context is relevant to real-world processing. The systematic contexts necessary for understanding abstract concepts may lead to arbitrary context inhibition, but greater attention to contexts that tend to be more relevant during real-world processing.

Thematic and other semantic relations central to abstract (and concrete) concepts

In this article, we discuss multiple types of meaningful (semantic) relations underlying abstract (as compared to concrete) concepts. We adopt the viewpoint that words act as cues to meaning (Elman in Ment Lexicon 6(1):1-34, 2011; Lupyan and Lewis in Lang Cogn Neurosci 34(10):1319-1337, 2019), which is dependent on the dynamic contents of a comprehender's mental model of the situation. This view foregrounds the importance of both linguistic and real-world context as individuals make sense of words, flexibly access relevant knowledge, and understand described events and situations. We discuss theories of, and experimental work on, abstract concepts through the lens of the importance of thematic and other semantic relations. We then tie these findings to the sentence processing literature in which such meaningful relations within sentential contexts are often experimentally manipulated. In this literature, some specific classes/types of abstract words have been studied, although not comprehensively, and with limited connection to the literature on knowledge underlying abstract concepts reviewed herein. We conclude by arguing that the ways in which humans understand relatively more abstract concepts, in particular, can be informed by the careful study of words presented not in isolation, but rather in situational and linguistic contexts, and as a function of individual differences in knowledge, goals, and beliefs.

Building semantic memory from embodied and distributional language experience

Humans seamlessly make sense of a rapidly changing environment, using a seemingly limitless knowledgebase to recognize and adapt to most situations we encounter. This knowledgebase is called semantic memory. Embodied cognition theories suggest that we represent this knowledge through simulation: understanding the meaning of coffee entails reinstantiating the neural states involved in touching, smelling, seeing, and drinking coffee. Distributional semantic theories suggest that we are sensitive to statistical regularities in natural language, and that a cognitive mechanism picks up on these regularities and transforms them into usable semantic representations reflecting the contextual usage of language. These appear to present contrasting views on semantic memory, but do they? Recent years have seen a push toward combining these approaches under a common framework. These hybrid approaches augment our understanding of semantic memory in important ways, but current versions remain unsatisfactory in part because they treat sensory-perceptual and distributional-linguistic data as interacting but distinct types of data that must be combined. We synthesize several approaches which, taken together, suggest that linguistic and embodied experience should instead be considered as inseparably entangled: just as sensory and perceptual systems are reactivated to understand meaning, so are experience-based representations endemic to linguistic processing; further, sensory-perceptual experience is susceptible to the same distributional principles as language experience. This conclusion produces a characterization of semantic memory that accounts for the interdependencies between linguistic and embodied data that arise across multiple timescales, giving rise to concept representations that reflect our shared and unique experiences. This article is categorized under: Psychology > Language Neuroscience > Cognition Linguistics > Language in Mind and Brain.

Language as a mental travel guide

Gilead et al.'s approach to human cognition places abstraction and prediction at the heart of "mental travel" under a "representational diversity" perspective that embraces foundational concepts in cognitive science. But, it gives insufficient credit to the possibility that the process of abstraction produces a gradient, and underestimates the importance of a highly influential domain in predictive cognition: language, and related, the emergence of experientially based structure through time.