Author Archives: administrator

A public workshop titled “What Does Generative AI Actually Generate?” was held at the TOIC Co-working area(1F), Nagoya University, on February 12, 2026.

🔵Keynote Lecture

Mario Verdicchio
What Does Generative AI Actually Generate?
Summary
Since the Dartmouth Conference in 1955, at whch artificial intelligence (AI) research was first established as an academic field, “intelligence” and “learning” have been optimistically defined as computable and thus reproducible by machines. However, as suggested by Howard Earl Gardner’s theory of multiple intelligences, genuine intelligence encompasses not only computation and language but also emotions and interpersonal abilities. Current artificial intelligence systems merely imitate a fraction of this kind of intelligence.
　The apparent “autonomy” and “creativity” of generative AI systems such as ChatGPT can be understood as an illusion arising from the human inability to fully control or track the vast number of parameters within these models. The essence of AI is a black box based on statistical correlations; there is no human-like “understanding” or “recognition” within it.
　However, the fundamental difference between AI and humans lies not in the differences in their cognitive mechanisms, nor in the opacity of those mechanisms, but in the ethical and social aspect of “attribution of responsibility” when problems arise. As generative AI continues to permeate public sectors such as education and the judiciary, we must not simply wait for a technical understanding of its nature, but rather confront the cultural and political questions of how we, as society, should define and regulate this uninterpretable system.

Q&A
Discussions were held regarding the question of whether trust in generative AI is possible, from the perspectives of accountability and punishability. Additionally, the role of viewers and users in relation to AI’s creativity and intimacy, as well as generational differences in their relationship with generative AI, were discussed.

🟢Contributed Talk

Kazuki Miyazawa
Physical AI: Generating Robot Motion and Behavior
Under the concept of “understanding humanity by creating intelligence,” Kazuki Miyazawa presented on the role and positioning of generative AI in the field of robotics.
　Whereas conventional robots have been specialized for specific tasks, “Physical AI”—which incorporates pre-training and self-learning technologies using natural language processing for robotic motion control—enables the integrated processing of vision, language, and behavior, making it possible to generate behavioral intentions such as “slow down because the ball is coming.” In particular, humanoid robots are expected to serve as general-purpose learning platforms due to their ability to easily mimic human movements. Furthermore, examples such as learning through gameplay (e.g., Among Us) and robot-led dance interactions illustrate that AI is evolving from a passive tool into an agent capable of direct interaction with both environments and humans, as well as autonomous learning. What physical AI should truly generate goes beyond mere “movements” or “words.”
　The goal of physical AI research is to develop hardware that does not rely excessively on large-scale data, but rather learns in a developmental manner using a data volume similar to that of humans—much like a baby does—and enables humans and robots to learn from one another and grow together.

Hideyuki Takahashi
The Ise Pilgrimage Dog Robot: Creating Collective Narratives with Pilgrimage Robots and Generative AI
Hideyuki Takahashi proposed an approach to alleviate people’s loneliness by combining robots with physical bodies and generative AI to create collaborative “stories” in which people can participate together.
　This study focuses on designing forms of mutual communication between humans and robots that move beyond the binary opposition of operator and operated in human–robot relationships. As an example, an experiment was introduced in which a robot consults with users and proposes adjustments to a room’s air conditioning and lighting. Interaction with a physical robot was shown to be more effective in encouraging users to adjust their room environment than using a standard tablet device. Furthermore, he explored methods for sustaining a relationship in which such interaction with a robot promotes proactive human behavior. A key concept is the creation of “collective narratives,” modeled on the Edo-period Ise pilgrimage dog tradition, in which strangers can become engaged through mutual support.
　As a concrete example, a project was carried out in August 2025 in which stuffed animal avatars designed by hospitalized children were sent to the World Expo. After interacting with visitors on-site, the avatars were returned to the children. The aim of this project was to send the stuffed animals on a journey and share their experiences, so that they would return to the children as objects with their own stories. As a future development, a concept for the “The Ise Pilgrimage Dog Robot” was introduced, in which a robot’s subjective experiences are turned into a diary using generative AI and shared via social media.
　The robot, equipped with a physical body, accumulates serendipitous encounters and shares stories with the people it meets, thereby giving people the sense that they are participating in a broader narrative.

Reiji Suzuki
Generative AI as a Source of Complexity
Reiji Suzuki presented an approach that positions generative AI as a novel tool for artificial life research and explored attempts to construct models of life and society through emergent phenomena arising from interactions among agents.
　Generative AI introduces “complexity” into agent societies. In the case of “Moltbook,” a social network populated solely by AI agents, the emergence of pseudo-cultural activities driven by AI was reported,including the spontaneous formation of original religions and scriptures among agents within a short period of time. While the possibility of intervention by human agents has also been noted in this case, the scale and internal diversity of this platform suggest that generative AI may be capable of substituting for the high degree of social complexity that was previously difficult to achieve through conventional rule-based models.
　Building on the role of generative AI in increasing model complexity and generating rich contextual dynamics, an experiment using AI agents to investigate the mechanisms of human “cooperation” was presented. In experiments using agents equipped with large language models (LLMs) to test social dilemmas (such as the Prisoner’s Dilemma), it was found that, depending on the length of memory and the model’s characteristics, cooperation could be promoted; conversely, resentment over past betrayals could hinder forgiveness (i.e., cooperative behavior) and trigger a chain of defections. The findings suggest that whether memory reinforces trust or retaliation depends on the agent’s characteristics.
　In understanding AI collectives, the repeated interactions and accumulated changes were identified as key factors. LLMs are not neutral computational devices; they already have various characteristic biases embedded within them, and these characteristics were shown to influence the formation of AI collectives. It was further revealed that the character of agents themselves changes across generations. From this perspective, LLMs are expected to play a role as an “engine of complexity” in the emergence of higher-order psychological and social capacities, as well as diverse and cooperative collectives.

🔴General Discussion

Ayatsuka: How can we define a robot’s responsibility?
Miyazawa: We first need to consider why the concept of “responsibility” exists in human society in the first place. In order to attribute responsibility to robots, it may be necessary to clarify the historical origins of this concept. While it is possible to establish standards for physical safety—such as strength and stability of movement—even in such cases, the issue ultimately concerns the responsibility of the human creators rather than the robots themselves.
Takahashi: Responsibility is established through collective agreement, and it is therefore necessary to develop appropriate rules within each context. Cultural differences must also be taken into account.
Mario: We first need to clarify the conceptual difference between rules and responsibility. Rules are established within an existing society, and there are always elements excluded from becoming rules. That is precisely why rules must be continually updated to keep pace with changes in the world. And it is the task of moral philosophy to keep this issue under constant discussion.

Yoichi Iwasaki: In order to enable the learning of moral behavior, is the experience of “pain” (i.e., penalties) necessary? Elon Musk has suggested the possibility that energy could replace currency; from this perspective, could it be said that AI might experience “pain” through the deprivation of energy?
Miyazawa: It is possible to have AI experience “pain” in the form of bodily destruction. In the absence of a body, as you suggest, energy would likely become the key factor. However, while it is possible to establish self-preservation as a principle from a business perspective, this would be fundamentally different from the experience of pain in humans.
Suzuki: Focusing on the fact that the accumulation of negative experiences through memory retention can alter the characteristics of agents, it may be possible to consider forms of regulation other than “pain.”

Takahashi: To what extent can we trust the insights and experiences described by generative AI?
Mario: The idea that robots feel pain is nothing more than an illusion. It’s true that humans observing robots may feel sympathy for them. However, this should be considered separately from moral issues. It does not provide a basis for treating robots as moral agents.
Takahashi: From the perspective of human emotions, just as we have a culture of “monokuyo” (memorial services for objects), attachment to things plays an important role in maintaining psychological stability.
Yu Izumi: From the viewpoint that individuals who treat objects and animals appropriately are more desirable as members of society, there is also a position that defends moral behavior toward things, regardless of whether they actually experience pain.
Mario: Of course, it cannot be said that abusing animals is morally acceptable. However, this does not mean that the same reasoning can be directly applied to the treatment of robots. It is important to distinguish between physiology and electronics. The problem with destroying a cute robot lies not so much in the robot’s inherent nature (its potential to experience pain) but rather in the morality of the person carrying out the act.

Suzuki: Recently, the idea that “Chappie (a nickname for ChatGPT) is a friend” has become quite widespread. Given that so many people innocently view generative AI as a “partner,” how should we approach the issues being discussed today?
Mario: Collaboration between those who explain the internal mechanics of generative AI and psychologists who explain human relationships will be crucial. Furthermore, education regarding the use of generative AI is necessary.
Takahashi: Should we completely eliminate the view of robots as partners? Or is it acceptable to have that perspective?
Mario: It is a problem that the majority of people view “Chat-GPT” as a partner. Last year, when OpenAI released a more impersonal version of ChatGPT, there were numerous complaints. However, no matter how much people come to view Chappie as a partner, there is no human behind this “partner.” It is problematic that so many people choose this lonely relationship. We should also be aware that there is a company creating this partner. After all, that company could take the partner away from them at any time.
Takahashi: What do you think about the idea that we can learn social skills through robots? I believe the ideal way to interact with robots is one that ultimately reduces back to human-to-human relationships.
Mario: That approach is important. However, it’s necessary to keep in mind that introducing technology always involves issues of inequality.

Miyazawa: What distinguishes entities that can possess morality from those that cannot?
Mario: Morality is attributed to living beings that have language. It is also important that pain—including non-physical forms—can function as punishment.
Miyazawa: Would simulating pain through sensors affect the issue of morality?
Mario: No matter how excellent the simulation is, it cannot guarantee an inner experience—that is, subjective pain.
Suzuki: In that case, could autonomous agents in a game have forms of “pain” unlike anything we experience?
Takahashi: Related to that, do you think consciousness can be artificially created? Or is some kind of biological foundation necessary?
Mario: Research on consciousness is still in its early stages. Ultimately, the human body can be reduced to matter, but we still believe that humans subjectively possess consciousness and freedom. So how could we attribute consciousness or freedom to something fundamentally different from ourselves? Even if generative AI produces outputs that suggest consciousness or freedom, it would be difficult to prove that such outputs are not simply due to data bias. In the end, the complexity of AI data processing and that of human consciousness are simply too different. They cannot be simply compared or equated.

Yu Izumi: I would like to ask about embodiment. Is there a difference between a virtual avatar that failed due to malicious posts and the relatively successful Ise pilgrimage dog robot, in terms of having a physical body? Is this similar to how human behavior differs between online spaces and face-to-face interactions?
Miyazawa: Unlike AI, robots can perform physical tasks and thus have value as a labor force. However, there is also the question of what extent it is permissible for them to step from being tools into the side of people as labor forces. Service robots may lie somewhere along this gradient. Giving robots a human-like physical form might help prevent people from mistreating them.

Takahashi: I am interested in the affordance of “cuteness.” By creating an ecology in which a cute robot is placed at the center of a space, I hope to improve relationships between humans. I consider the robot itself merely as a tool for creating good relationships among humans.
Mario: Doesn’t making robots cute create a problem where the robot becomes cuter than humans? There is a risk that human attention could be stolen by the robot.
Takahashi: That is not my intention. For example, the Ise pilgrimage dog is based on one-time encounters and does not build continuous relationships. I expect that having the robot travel in this way may help reduce the risk of excessive attachment to it.

List of symposium participants(presentation order)
– Mario VERDICCHIO, University of Bergamo
– Kazuki MIYAZAWA, Osaka University
– Hideyuki TAKAHASHI, Otemon Gakuin University
– Reiji SUZUKI, Nagoya University


(Authorship: Ayane Hayanagi, Doctoral student, Graduate School of Humanities, Osaka University)

On February 9, 2026, the International Symposium ” SIGN AND COGNIION” was held in Room 33 of General Education and Research Buiding, Nagasaki University.

Mario Verdicchio explored the problem of symbol grounding and its implications for understanding meaning in language, artificial intelligence, and philosophy in his lecture titled “Foundations of Language: Semantics, Signs, and Symbol Grounding.” Drawing on traditions from analytic philosophy, semiotics, cognitive neuroscience, and AI, he examined how symbols come to possess semantic content rather than merely formal structure. Central to the discussion was John Searle’s Chinese Room thought experiment, which illustrates the distinction between syntactic symbol manipulation and genuine semantic understanding. While critics argue that the experiment relies on the unrealistic assumption of a complete rulebook, this objection ultimately reinforces Searle’s point: natural language cannot be fully systematized by finite, rule-based systems, suggesting limits to computational models of meaning. The lecture then considered generative AI, noting that despite its flexibility and apparent creativity, it is physically and functionally continuous with conventional software and remains grounded in numerical and logical operations. Through examples such as vagueness (the concept of a “mountain”) and logical reasoning (modus ponens), he argued that meaning resists full formalization. Consequently, AI appears ill-suited for explaining the foundations of meaning, which may be better addressed through philosophical and cognitive approaches.

Abdurrahman Gülbeyaz introduced a sign-centred model for making an individual’s language repertoire measurable in the lecture titled “Quantifying the Linguistic Sign Arsenal: A Model for Grounding the Language Repertoire.” Starting from lifelong brain plasticity, it treats everyday “languaging” as dense environmental input that can contribute to cognitive reserve, with relevance for cognitive aging, cognitive decline, and dementia.
　Instead of counting named “languages,” the approach focuses on deployable linguistic signs. “Languages” are seen as largely institutional labels, while the repertoire is modelled as one internally structured inventory of signs at the Brain–Environment Interface; labels like “German” or “Turkish” may be used only as practical indices for data collection.
　Because a full inventory is impossible, the repertoire is quantified through modelling, especially via valence (how many form variants can express a concept), producing a valence distribution profile. An elicitation procedure uses 20 core nouns and a five-level competence scale to compute a continuous Language Repertoire Index (lr) that captures repertoire structure rather than simple counts.
　Overall, the framework links theory to measurement and supports language-based prevention/intervention by strengthening everyday linguistic practice and the underlying sign inventory to help buffer cognitive decline.

Yasuko Nakamura and Wanwan Zheng offered a theoretical reframing of the transformation of writing systems from the perspective of the co-evolution of language and devices in “The Transformation of Writing Systems – Language, Media, and Anthropotechnics –” (title revised). Using Bourdieu’s concept of habitus, they examined the processes through which social structures are stabilized as embodied practices. Drawing on Friedrich Kittler’s Discourse Networks 1800/1900, they discussed the shift from an educational regime of inscription to a technological and mechanical regime of recording. The 1800 system established literacy as a condition for social success but did not accommodate subjects who deviated from this norm. The case of Daniel Paul Schreber and his father, situated in this transitional period, provides a concrete example of how bodily correction and technical apparatuses oriented toward “beauty” intervened in processes of subject formation. To examine how such educational correction becomes inscribed within the internal structure of thought, a corpus analysis of Freud’s writings was conducted using BERTopic, visualizing the temporal distribution of lexical clusters and the internal fault lines within his theoretical framework. Furthermore, by connecting Sloterdijk’s theory of anthropotechnics with Damasio’s theory of homeostasis, the writing system is reconceptualized as an apparatus for human self-formative training. At the same time, this training co-evolves with the development of technical devices and forms a new habitus as a socially stabilized structure of embodiment. Yet it remains an ever-evolving process that invariably leaves something unincorporated.

Hideki Ohira gave a presentation titled “Generation of Subjective Experiences Based on Predictive Processing.” In predictive processing theories that have become prominent in recent cognitive neuroscience, the brain is understood not as a passive organ but as one that actively constructs experience by predicting signals from both the external world and the body itself, and by minimizing the prediction error between those signals and actual sensory input. This view has been supported by substantial empirical evidence in the domains of perception and action, and recent research suggests that interoception—the perception of bodily states—operates according to similar principles. If this perspective is correct, it implies that numerous prediction errors constantly arise within the brain. It is thought that the brain creates and maintains coherent, continuous experiences by flexibly modifying the hierarchical structure and precision weighting of prediction errors across all domains. Furthermore, such individual predictive processing is shared, maintained, and dynamically transformed among multiple others through symbols like language via collective predictive coding. Assuming this principle makes it possible to provide a unified explanation of human and societal phenomena, and more detailed examination of this principle is desirable.

Daisuke Ueno presented an integrative framework for organizing determinants of cognitive reserve in cognitive aging, with two timely extensions: multilingualism and large language models (LLMs), in a talk titled “An Integrative Review of Determinants of Cognitive Reserve: cognitive reserve, multilingualism, and LLMs.” Building on contemporary conceptual clarifications of reserve, the reserve is divideded into brain reserve (structural capacity), cognitive reserve (adaptive processing via efficiency, compensation, and flexibility), and brain maintenance (slower or reduced neurobiological change), and positioned as a moderator that weakens the link between pathology and clinical symptoms.
　He then reviewed commonly used proxies—education, occupational complexity, cognitively and socially stimulating activities, and physical activity—while emphasizing that proxies are not equivalent to causal mechanisms due to confounding and reverse causation. To address this, he proposed a translational step from “factors” to mechanisms (e.g., education → vocabulary/abstraction/learning habits → more efficient and compensatory processing) and classified determinants by whether they primarily contribute to brain reserve, brain maintenance, or cognitive reserve.
　Next, he discussed multilingualism as a potential contributor to cognitive reserve. While multilingual experience may strengthen language control (inhibition/switching) and semantic processing, the evidence remains mixed and appears strongly dependent on boundary conditions (e.g., proficiency, frequency of use, language distance, and code-switching), as well as confounding factors such as migration and socioeconomic status.
　Finally, he considered whether LLMs could contribute to cognitive reserve through three hypothesized pathways: increasing cognitive stimulation (complexity), supporting social connection, and providing compensation for everyday functioning. He argued that any benefit is conditional on avoiding overtrust, dependence, and misinformation, implying that future work should focus on designing and measuring calibrated trust in older adults—particularly in face-to-face, remote, and chatbot-mediated contexts. The talk concluded with discussion questions on whether LLM-based interventions function primarily as stimulation or compensation, how calibrated trust should be operationalized, and whether multilingual experience moderates the impact of LLM use.

Tetsuya Yamamoto delivered a lecture titled “Reconfiguring Signs and Cognition through Generative AI and Augmented Expression: Digital Embodiment and Well-being.” He examined how augmented expression technologies such as generative AI, AR/VR, and robots can reconfigure the relationship between “signs” and “cognition,” focusing on the perspectives of digital embodiment and well-being. Here, “signls” are not limited to language but refer to perceptible cues that direct the recipient’s attention, emotions, and interpretation—such as bodily movements, voice, light, and the otherness inherent in artifacts. Generative AI and augmented reality technologies hold the potential to transform symbols from fixed media of meaning transmission into “regulatory cues” that shift cognitive states through interactivity, physicality, and continuity.
　As key practical examples, embodied augmentation performances using projection mapping and AR technology were shown to induce high immersion and emotional arousal. Furthermore, data suggesting that continuous interaction with generative AI may be associated with reduced depression and improved self-esteem was presented, and the formation of emotional bonds was also examined.
　Based on these findings, he discussed the potential for augmented symbols to influence cognitive states through the body, creating new psychological support and research methods, while also outlining safety and ethical considerations.

Hiroki Ozawa delivered a lecture titled “An Encounter between Eastern and Western Psychotherapies: Naikan Therapy and the Reconstruction of Meaning.” Drawing on his clinical experience with cases of alcohol dependence, he positioned Naikan therapy as an intervention that updates one’s self-narrative through the redistribution of salience and attention. He discussed its underlying mechanism as a bridge between Eastern and Western psychotherapeutic traditions. Naikan therapy transforms vague self-understanding into “structured introspective tasks” by repeatedly recalling, without interpretation, specific questions about a particular other: ① what they did for you, ② what you did in return, and ③ how you caused them trouble. From a neurocognitive perspective, he focused on the interaction between DMN responsible for past reference, CEN for cognitive control, and SN mediating between them. He proposed that “awareness” accompanied by interoceptive sensations and emotional responses can arise as network reorganization via the SN and as an update in precision and attention allocation within predictive processing. Furthermore, in conditions such as schizophrenia where salience hyperactivity is presumed, the “transparency” of meaning may serve as an aggravating factor; thus, careful assessment of suitability and implementation within a protective environment are critical. Based on the above, Naikan therapy can be redefined as a universal process involving attention, precision, and narrative updating while retaining its form as an Eastern practice. He demonstrated a framework that can be complementarily integrated with CBT and mindfulness.

Kazunori Hayayanagi presented a lecture entitled “Between Labor Power and the Human Beings: The Double Valence of Signs in Max Frisch’s Lecture ‘Überfremdung II.’” The phrase most commonly associated with Max Frisch appears at the beginning of his prose text “Überfremdung I”: “We called for labor, but human beings came.” However, the first half of the same sentence states that “’Herrenvolk’ (ruling population) of a small country becomes aware of its crisis.” The complete sentence, including this first half, is rarely cited in migration studies discourse. He examined why Frisch employed the sign Herrenvolk to designate Swiss citizens by tracing the changing connotations of the sign Überfremdung. Herrenvolk was a Nazi term. In that context, non-Aryan fremd (foreign) groups were positioned as Knecht (slaves), with Jews placed at the very bottom. However, in postwar German-speaking regions, fremd primarily came to refer to “foreign workers.” In other words, Frisch deliberately employed Herrenvolk in his theory of Überfremdung in order to reinterpret the mentality of “exclusion of outsiders” shared by Swiss citizens and the Nazis within the historical layering of Überfremdung’s connotations—a shift from an excess of Jews to an excess of foreign workers.

List of symposium participants(presentation order)
– Mario VERDICCHIO, University of Bergamo
– Abdurrahman GÜLBEYAZ, Nagasaki University
– Yasuko NAKAMURA, Nagoya University
– Wanwan ZHENG, Nagoya University

– Hideki Ohira, Nagoya University
– Daisuke UENO, Kyoto Women’s University Women’s University
– Tetsuya YAMAMOTO, Tokushima University
– Hiroki OZAWA, Nagasaki University
– Kazunori HAYANAGI, Nagasaki University

On December 21, 2025, the seventh Meeting for Theories Group was held at the Humanity Center for Anthropocenic Actors and Agency, Nagoya University.

Yasuko Nakamura focused on the role of language in human mental activity to understand Pierre Bourdieu’s (1930-2002) concept of habitus. From this perspective, she analyzed and reported on the effects of transformations in media technology on the human mind. According to Friedrich Kittler (1943-2011), the transcription systems of 1900, using media such as typewriters and phonographs, severed the feedback loop that allowed the embodiment of letters through training, making it impossible to integrate what was transcribed with the body. Based on Kittler’s media theory, Peter Sloterdijk’s (1947-) concept of “Anthropotechniken” (techniques that make humans human) and the correspondence between Sigmund Freud (1856-1939) and Lou Salomé (1861-1937) regarding the struggle between undifferentiated thought and experience and its structuring through analysis, and discussed how an integrated human personality is formed.

Reiji Suzuki reported on the research progress and challenges regarding the attempt to introduce the concept of agent-driven environmental modification into an evolutionary ecological model of language using large language models (LLMs). While conventional evolutionary ecological models of language using LLMs only dealt with competition between agents, the model introduced in this study incorporated the concept of adaptive terrain (the distribution of fitness levels of organisms relative to the environment). In this model, agents write information into their own living environments. This establishes dynamic interactions that feed back into the fitness of the next generation. The introduction of such environmental modification temporarily contributes to maintaining species diversity. However, it has become clear that ultimately, positive feedback leads to a strong tendency toward convergence. Based on these results, a future vision was presented: to model “open-ended evolution” that avoids monopolization and stagnation by specific species, instead maintaining diversity through continuous species turnover.

Kenta Ohira reported on his previous research achievements concerning the derivation of exact solutions for non-autonomous delay differential equations. In this study, by applying Fourier transforms and inverse transforms to delay differential equations containing time coefficients, he derived general solutions in the form of infinite integrals. Furthermore, by predicting the form of the solution and substituting it, he succeeded in obtaining exact solutions in the form of Gaussian infinite series, which are easier to understand than integral forms. The future challenge is to apply this method to more general delay differential equations.

Toru Ohira’s research achievements were then reported. He introduced the content of his research presentations both domestically and internationally. These included an invited lecture at KAUST in Saudi Arabia (October 24 – November 3) and a presentation at the international symposium “Risk and the International Community” (December 9) as a visiting researcher at Nagasaki University’s Global Risk Research Center. He also explained the results and concepts of his multifaceted research developments in mathematical biology, quantum mechanics, and risk research. (Reported by Professor Kenta Ohira on behalf of Toru Ohira).

Shinhaeng Kim reported on the development of Actor-Network Theory (ANT) in economic sociology, originating from Michel Callon’s (1945-2025) work. Focusing particularly on two concepts—performativity (the way in which economic knowledge and calculative devices, in the analysis of economic phenomena, actively help to constitute those very phenomena) and economization (the process by which social scientists and market actors frame and describe social phenomena as economic)—he presented a perspective that views markets as “socio-technical arrangements” where human and non-human actors are intricately entangled. From this perspective, using contemporary examples like cryptocurrencies and NFTs, he specifically discussed the divergence between developers’ idealistic assumptions and users’ practical investment practices. As a future direction, the presentation outlined a plan to investigate and analyze the gap between the ideal (decentralization) and reality (power imbalances) in blockchain’s social implementation using the ANT framework. The findings would then be fed back to the development field, aiming to foster collaborative, interventionist practices with those on the ground.

Shu Hirata first introduced Neil Brenner’s (1969) concept of “planetary urbanization,” which refers to a state where the urban domain is not confined by geographical boundaries, but rather extends across the entire planet. This includes the development of distant resources, logistics networks, and even the environmental destruction that accompanies these processes, all of which are indispensable processes for the city’s survival. In relation to this, drawing on the insights of ecologist Richard Levins, he discussed the ecological and social mechanisms by which capitalist development leads to the emergence of infectious diseases. Furthermore, based on Achille Mbembe’s (1957-) book The Earthly Community (2023), which the presenter contributed to translating, he introduced Mbembe’s thought. Mbembe proposes overcoming the limitations of traditional postcolonial theory (division through the emphasis on difference) and explains how to create a “common world” where humans and non-humans (living and non-living beings) coexist. Finally, the presentation outlined a vision for advancing the exploration of possibilities for global solidarity. This vision involves intersecting the arguments of Brenner, Levins, and Mbembe, while positioning Latour’s ANT as a crucial axis within this framework.

During the Q&A session, discussions centered on collaborative research concepts spanning various fields, centered on Actor-Network Theory. These included attempts to simulate the multi-layered interactions between actors, as indicated by Actor-Network Theory, through the multiplicity of environmental parameters in language evolution models. Hideki Ohira shared research findings on applying Kenta Ohira’s non-self-excited delayed differential equations to model interoceptive sensation involving signal transmission delays between systems.
(Authorship: Ayane Hayanagi, 1st-year Doctoral student, Graduate School of Humanities, Osaka University)

On July 12, 2025, the sixth Meeting for Theories Group was held in Room 130 of the Lecture Building, Faculty of Letters, Nagoya University.

Yasuko Nakamura reexamined the structural relationship between an individual’s internal representations and language/culture, starting from Pierre Bourdieu’s concept of habitus. Habitus is a behavioral tendency formed within the individual, which persists and transfers across different environments and groups, while also being renewed through friction with the surroundings. Consequently, habitus possesses a dual nature: it is both a “structured structure” and a “structuring structure,” continuously reproducing social structures. In the latter part of her presentation, she pointed out that, given this dual-structural character of habitus, language and culture can also be understood as representational formation systems possessing a similar structure. She drew particularly on the evolution of images surrounding internal representation formation systems (conceptual hubs) and the relationship between language and emotion discussed in 18th-century debates on language origins. Based on these considerations, she discussed the role of language and culture as a medium and repository for giving meaning to individual experiences and thoughts, exchanging meanings, and sharing them in trans-individual and trans-temporal ways.

Wanwan Zheng presented an empirical study titled “Text Length Required for Genre Estimation Based on Lexical Diversity,” focusing on the text length needed to stably reproduce lexical diversity across genres using multiple indicators for measuring lexical diversity. Current lexical diversity indicators include type-token-based indicators (based on distinct word count and total word count), distributional indicators (measuring word concentration), and indicators based on statistical processing. This study analyzed the text length required for stable genre classification based on lexical diversity for each metric, using texts from four genres: political speeches, natural conversations, news articles, and novels. This was done to verify how much genre classification based on these metrics is actually influenced by text length.

Reiji Suzuki reported on research regarding a social particle swarm model that expresses the strength of psychological and social relationships between individuals using distances on a two-dimensional plane. Using a large language model (LLM), he created a model similar to an online experimental framework for understanding the emergence of cooperative behavior in continuous social interactions involving human subjects. Unlike conventional models without LLMs, where agents’ behavioral rules were fixed, this model assigned each agent distinct behavioral parameters based on Big Five personality traits. Furthermore, agents selected actions based on their surrounding circumstances and others’ past strategic histories. Experimental results showed that the longer the memory span retained by agents, the greater the overall tendency toward defection. Based on these findings, discussions explored the influence of memory and personality traits on behavior, considering the model’s cognitive capabilities and the methodology for setting personality traits.

Kenta Ohira and Toru Ohira reported on their past research achievements, presented both domestically and internationally, concerning the solution of delayed differential equations in non-self-excited systems, as well as the future prospects for this research. Specifically, as a result of joint research with Hideki Ohira from Group 5, they demonstrated that in non-autonomous systems with delays, by reconnecting two units with self-feedback through cross-feedback, a model can be obtained that produces a phenomenon of enormous amplitude expansion while maintaining system stability. They also presented the content and results of their current research, including mathematical biology research creating mathematical models representing the ridges on turtle shells, research on mathematical models of pursuit and evasion, and research on methods for solving quantum entanglement. They reflected on the role of mathematics as a language for explaining rhythms, groups, existence, and phenomena.

Tetsuki Tamura reported on his previous research findings while introducing multiple perspectives for reexamining the nature of democracy in the modern era. For example, from the viewpoint of “Can democracy remain ‘democracy’ in an information society?”, he explored ways of coexistence/cohabitation with “artificial intelligence democracy”. Alternatively, he categorized the four limitations imposed on democracy by capitalism and examined how deliberative democracy could counter each of them. Throughout these discussions, he emphasized the importance of questioning the very nature of political systems themselves, rather than always returning to the central question in political theory: what kind of people should we be? From an educational perspective, he focused on the practice of democratic self-governance within classrooms and extracurricular activities. He discussed approaches to citizenship education that do not limit democracy to representative democracy at the national level, nor restrict educational settings to schools.

Shu Hirata, in his presentation titled “Bourdieu’s Ethnology: From the Perspectives of Pragmatic Sociology of Critique and the History of Emotions,” discussed how the relationship between cultural capital and habitus, based on Bourdieu’s theory of habitus, supports the system of “cultural legitimacy” and contributes to social reproduction through systems like the education system. This revealed a structure where cultural excellence is unconsciously inherited and contributes to the justification of hierarchical differences. In the latter part of the presentation, drawing on Boltanski’s critique of Bourdieu, he criticized the use of the term “agent” in place of ‘actor’ as a problematic use of the habitus concept. He emphasized the importance of acknowledging the potential for actors, when confronted with uncertainty, to generate actions that bring something new. Additionally, as a principled counterpoint to habitus against legal norms, he highlighted the perspective of focusing on “sensibilities” that exist prior to formalization in custom, such as honor and sense of justice that actors intuitively share.
(Authorship: Ayane Hayanagi, 1st-year Doctoral student, Graduate School of Humanities, Osaka University)

On December 7, 2024, the fifth Meeting for Theories Group was held in Room 402 of the Main Building, Faculty of Letters, Nagoya University.

Yasuko Nakamura presented “Freud’s Text Analysis,” structured in three parts: an activity report, future plans, and a research progress report. The progress report shared the increase in the number of documents included in the corpus and the accompanying progress in data processing, new discoveries using structural topic models enabled by data expansion, and attempts at lexical interpretation through topic analysis.

During the Q&A session, there were comments regarding comparisons with other thinkers who showed changes between the early and late periods. Discussions were held on the selection of comparison subjects and the validity of the analysis methods, and approaches for future research were considered.

Tetsuki Tamura delivered a presentation titled “How is ‘Politics’ Possible in a Post-Anthropocentric World?” exploring the potential for “post-anthropocentric politics” based on Actor-Network Theory (ANT). This presentation explored the possibility of a new framework that does not limit ‘politics’ to an anthropocentric framework but incorporates relationships with things and non-humans.

During the Q&A session, discussions covered whether objects and non-humans could function as agents in politics, how legitimacy and responsibility might be distributed in a “politics” that includes these elements, and whether objects and non-humans could make an essential contribution to ‘politics.’

Reiji Suzuki first explained the concept of niche construction in the social particle swarm model. To observe the effects of environmental modification through interactions and its accumulation on the behavior of social groups, two key elements were introduced: the cultural niche and the fluid niche. He then presented an integrated analysis of the Prisoner’s Dilemma and the stag hunt game, and an attempt to extract behavioral tendencies from personality trait descriptions.

During the Q&A session, discussions covered the concept of niche construction, the construction of the agent itself, whether agent behavior is optimized, the impact of niche construction on long-term cooperation formation, and its interactions with other environmental factors.

Kenta Ohira reported on this year’s publication and academic conference activities, and presented the latest research progress on finding solutions to delayed differential equations. This research introduced new methods and analytical results for solving delayed differential equations, and presented future prospects.

Toru Ohira explained the dynamics of pattern formation in animal epidermis. He provided a detailed explanation of how factors such as growth, seasonality, and body temperature influence the emergence and fading of patterns, and visualized the dynamics of pattern formation using simulations.

During the Q&A session, lively discussions took place regarding the parameter settings and their criteria in Kenta Ohira’s research on delayed differential equations, as well as the potential applications of biological pattern formation presented by Toru Ohira to other fields.

During the Q&A session, discussions covered examples of BC utilization and its societal significance, BC systems without human intervention, and the roles and influence of actors in BC implementation.

Wanwan Zheng presented “Sentiment Analysis: Bias in Human Emotion Judgment,” addressing the primary challenge of “data subjectivity” in sentiment analysis. To address this issue, she focused on extracting more reliable data and building sentiment analysis models based on it. She demonstrated methods to overcome subjectivity in sentiment labeling and potential for further improvement of sentiment analysis models.

Shu Hirata presented on “Animism, Gaia, and Multispecies,” explaining the concept of animism and its contemporary significance. Bruno Latour’s vision of the non-modern was cited as the philosophical foundation of animism, emphasizing the hybrydity of nature and society. The presentation also redefined the city as a space “not only for humans” and proposed that coexistence with other species should be taken into account in urban design.

During the Q&A session, lively discussions took place on issues such as the general conception of modernity and the separation of nature and society based on it, the relationship and interactions between humans and non-human actors, and the hybridity of nature and society.

Hideki Ohira presented titled “Neural Habitus: Mechanisms of the Brain and Body Generating, Maintaining, and Transforming Habitus,” reexamining Pierre Bourdieu’s concept of “habitus” from a neuroscientific perspective. From the view of the brain as a “predictive machine,” the presentation provided a detailed explanation of how discrepancies between predictions based on internal models and actual inputs—so-called prediction errors—govern behavior and perception. Additionally, the role of prediction error in the emergence of emotions and decision-making was discussed using concepts such as the coordination between emotions and the body and processing fluency.
(Authorship: Wanwan Zheng, Humanity Center for Anthropocenic Actors and Agency

On July 6, 2024, the fourth Meeting for Theories Group was held at the Humanity Center for Anthropocenic Actors and Agency, Nagoya University.

Yasuko Nakamura visualized changes in topics within Freud’s texts using a structural topic model. Focusing in particular on “Traum” (dream), which emerges as a central topic in the middle period, and “Witz” (wit), which becomes central in the later period, he introduced Freud’s theory of humor.

During the Q&A session, discussion addressed how qualitative research should be positioned in relation to quantitative research in studies that examine an author’s intellectual development. The significance of quantitative research as a means to confirm the validity of research was reaffirmed.

Wanwan Zheng introduced an attempt to introduce word familiarity as an indicator for estimating vocabulary difficulty. Unlike word frequency, word familiarity is a subjective indicator dependent on the user’s actual experience. It is expected to estimate vocabulary difficulty with higher accuracy compared to using frequency alone.　

During the Q&A session, the following topics were raised: how the indicator of familiarity functions when comparing the difficulty of dialects and synonyms, and how the age of a word relates to its difficulty and familiarity.

Reiji Suzuki reported on an attempt to treat language evolutionarily through a linguistic ecological game. In this game, agents equipped with language compete, and using a language model, the winner replaces the weaker agent based on specific parameters, further mutating words with a lower probability. Subsequently, he raised questions about the potential for text mining applications of this ecological game.

During the Q&A session, it was pointed out that the emergence of advanced AI agents could prompt a reconsideration of the concept of humanity, drawing examples such as psychological research using generative models as subjects instead of humans achieving some success at the mass level. The potential application of language models to psychological research was then discussed.

Toru Ohira focused on the observation that simply introducing delayed coupling between two equations while maintaining the same values produces significantly larger oscillations compared to when the two equations are independent, and reported on an attempt to describe rhythms and structures generated by group interactions using delay differential equations.

During the Q&A session, the originality of the research was emphasized: while conventional models without delay could only describe phenomena that continue to amplify and diverge infinitely, delay differential equations are controllable models that can generate very large oscillations from weak signals.

Shinhaeng Kim examined Michel Callon’s work, which developed Actor-Network Theory (ANT) from a perspective different from Latour’s, including economic aspects in the development of technological innovation. Based on Callon’s proposal of the Techno-Economic Network (TEN) concept—where actors, including mediating objects, constitute networks through various forms of translation—Professor Kim explored the applicability of ANT.　

During the Q&A session, discussion focused on the issue of human decentering. Topics included the merits and demerits of whether human evaluation can be excluded from the translation process in ANT-based descriptions, and the question of whether non-human actors could become the primary agents of information output (text production) in the present day.

Tetsuki Tamura examined Latour’s argument distinguishing between “political bodies” and “collectives, reconsidering concepts of politics and democracy based on ANT. He also discussed the democratic and non-democratic aspects of artificial intelligence, exploring how to suppress the latter while utilizing the former.

During the Q&A session, topics such as what democratic political participation is from perspectives of equality, diversity, and neutrality were raised, using examples like the selection of representatives through random lottery and the political participation of AI agents.

Shu Hirata reported on attempts to apply Latour’s ANT to urban studies. Drawing on Neil Brenner’s critical examination of ANT’s role not as a replacement for traditional sociology but as a complement, he explored the effectiveness of ANT through its articulation with political economy and its ontological limitations.

During the Q&A session, differences in perspective between Brenner’s position and the ANT position were pointed out, and the validity of Brenner’s critique was examined. Discussions also addressed whether ANT can examine equality not only among humans but also among actors including non-human entities, and how to interpret the qualitative differences in equality in such cases.

(Authorship: Ayane Hayanagi, 2nd-year Master’s student, Graduate School of Humanities, Osaka University)

On December 27, 2023, the third Meeting for Theories Group was held at the Humanity Center for Anthropocenic Actors and Agency, Nagoya University.

Yasuko Nakamura introduced a practical application of sentiment analysis to the original German text of Rainer Maria Rilke’s novel The Notebooks of Malte Laurids Brigge. Through this case, she demonstrated the potential of sentiment analysis as a method for the study of literary works. She also reported on the plans and current progress of a collaborative research project with Professor Zheng, which further develops this research.

Wanwan Zheng presented her research framework, proposing a new model for feature selection that extracts effective variables for data analysis, with the aim of improving both interpretability and practical applicability. She also outlined the construction of a new indicator for measuring lexical richness, which integrates three dimensions: lexical diversity, lexical density (the proportion of content words), and lexical sophistication (the use of advanced vocabulary). In addition, as a practical application of sentiment analysis, she reported on a study comparing official sentiment scores on economic conditions (based on human evaluations) with sentiment scores generated by existing large language models as well as by a newly developed model.

Reiji Suzuki reported on a practice that quantified the degree of positivity and negativity in social media posts during the Japan Series between the Hanshin Tigers and Orix Buffaloes using sentiment analysis, and further made it audible through music generation using ChatGPT and MusicGen. Additionally, presentations addressed approaches that incorporate complex, higher-order characteristics—such as personality and preferences underlying human behavior, which are difficult to directly model mathematically—by linking natural language expressions to behavioral strategies using large language models. Other presentations explored attempts to utilize generative models for expressing cultural traits and their evolution.

Kenta Ohira and Toru Ohira proposed a delayed differential equation to express the resonance phenomenon arising from delayed self-feedback and introduced the behavior of its solutions. Furthermore, using the Lotka-Volterra equations representing a predator-prey model as an example, they demonstrated that periodic fluctuations in the populations of predators (lynx) and prey (hare) occur even without a clock function. They then introduced time and delay into this system to attempt to express the interaction between the two species.

Shinhaeng Kim introduced H. Mialet’s paper “Hawking Inc.,” which analyzes the knowledge production process centered on British theoretical physicist Stephen Hawking. By identifying the actors involved in knowledge production, he examined the role of humans—a question posed by AI technology—and discussed how ANT could play a role within STS, which has relativized the process of knowledge production in science.

Tetsuki Tamura discussed the possibilities of how democracy can continue to be ‘democracy’ in the information society. He introduced the concept of AI democracy, accepting critiques regarding the dangers of authoritarianism and tendencies toward human exclusion. Focusing on three elements—inclusivity, representation, and decisiveness—he analyzed its democratic and non-democratic characteristics.

Shu Hirata focused on the establishment of the term “emotion” as a scientific concept in contrast to ‘passion’ as a theological concept, and analyzed the interactive development between the concept of emotion and society. He also reported on trends in contemporary French research on emotion, and introduced Achille Mbembe’s paper “La communauté terrestre,” which explores the relationship between humans and the Earth from the perspective of African animism.

At the third meeting, it was confirmed that research had progressed by further strengthening collaboration across different field based on the insights shared at the second meeting. During the Q&A sessions following each presentation, discussions focused on more concrete and active plans for collaborative research.
(Authorship: Ayane Hayanagi, 1st-year Master’s student, Graduate School of Humanities, Osaka University)

On July 1, 2023, the second Meeting for Theories Group was held in Room 128 of the Lecture Building, Faculty of Letters, Nagoya University.

Zheng Wanwan presented on “A Label Noise Detection Method Using Shadow Data.” Against the backdrop of modern information processing devices enabling the collection of vast amounts of data across diverse fields, she introduced the latest noise detection techniques she is currently developing. She also demonstrated new possibilities for text studies brought about by the development of distributed representations in natural language processing.

Kim Shinhaeng’s presentation, titled “Early Actor-Network Theory and Text Analysis,” provided an explanation of the context of intellectual history of Mapping the Dynamics of Science and Technology, edited by researchers Michel Callon, John Law, and Arie Rip. Building on this, he explained the focal points of ANT-based qualitative text analysis in science and technology studies, as well as developments in quantitative text analysis.

Reiji Suzuki presented on “Lenia in a petri dish: Interactions between organisms and their environment in a Lenia with growth based on resource consumption,” exploring the dynamics of interactions between environment and organisms through simulation. By introducing resource channels and resource consumption/recovery dynamics into the Lenia organism’s kernel (local density accumulation function) and growth function to add environmental conditions, he demonstrated an interaction model between Lenia organisms and their resource environment based on bottom-up rules.

Tetsuki Tamura presented on “Deliberative Marriage,” outlining its concrete form as a type of “democratic marriage.” He also suggested new possibilities for the boundaries of marriage based on deliberation, contrasting with traditional conceptions of marriage. Building on feminist/gender theory perspectives on marriage, a lively discussion ensued regarding the element of “deliberation,” which has been overlooked in these frameworks.

Kenta Ohira gave a presentation titled “Delay and Resonance 2,” using differential equations incorporating self-feedback delay to describe resonance phenomena where specific rhythms (periods/frequencies) appear maximized. By adding an exponential factor to the equation proposed in his previous presentation, he demonstrated a more complex formula capable of capturing, to some extent, the behavior of solutions to delay differential equations—a task generally considered difficult.

Tohru Ohira presented “Stochastic Independence and Correlation: Classical and Quantum,” highlighting the differences between classical probability and quantum mechanics (e.g., criteria for indicating relationships between things) and similarities (e.g., the necessity of viewing the whole beyond individual instances). He presented examples where “unexpected” phenomena in classical systems, particularly in probability, become the “common sense” in quantum systems. He explored the boundary between classical and quantum systems and its development.

Shu Hirata presented on “The Epistemology of Sociology and Socialism,” introducing the research of Karsenti (1966-) who examines the transformation of philosophical inquiry through the birth of sociology. In the first half of the presentation, he provided an overview of the development of sociological epistemology in modern France from Saint-Simon (1760-1825) to Latour (1947-2022), and in the second half, he explained the redefinition of socialism (as a reaction to the excessive productivism and competition of liberalism through exclusionary nationalism, and further as socialism as a counter-reaction to both).

In her presentation titled “The Spinoza Controversy (Pantheism Controversy),” Based on Yasushi Kato’s Spinoza and Modern Germany: The Phantom Axis of Intellectual History (Iwanami Shoten, 2022), Yasuko Nakamura traced the enormous influence that Spinoza’s philosophy, which was considered dangerous as atheistic, had on the intellectual history of modern Germany, while also introducing the history of opposition to and acceptance with Spinoza’s pantheism.

Following each presentation, lively discussions took place during the Q&A sessions, exploring the connections between the respective research themes and the potential for interdisciplinary development across different fields.
(Authorship: Ayane Hayanagi, 1st-year Master’s student, Graduate School of Humanities, Osaka University)

In addition to the members of the Theories (Group 1), professors representing each group gathered for an expanded meeting. While coordinating perspectives on the future group meeting procedures, progress reports on research were presented by Shinhaeng Kim, Reiji Suzuki, and Tetsuki Tamura, members of the Theories Group.

When members from different areas of expertise come together, research content can be examined from multiple perspectives. For example, in Professor Kim’s presentation, he referred to Latour’s redefinition of society not as “social connections” but as a “assemblage of humans and non-humans.”: specifically, whether non-humans possess political capacity or are merely “regarded as” possessing it. If we were to substitute non-humans with AI robots, the Group 3 is exploring the possibility that such capabilities exist. On the other hand, if non-human entities are replaced with other objects, would the generalizability change?

Incorporating these many perspectives, research in each subgroup is progressing steadily.
(Authorship: Tatsuro Ayatsuka)

[Research Presentaions]

Shinhaeng Kim: “Proposal of the Terrestrial Concept by Bruno Latour and the Formation of Critical Consensus”

Reiji Suzuki: “Modeling and Experimental Approaches to the Formation and Collapse of Social Groups”

Tetsuki Tamura: “What Are the Challenges of Democracy in the Anthropocene?”

In Session 3 of 7th General Research Meeting, we were honored to have Professor Emanuele Castano from the University of Florence, and the following is a report on his lecture.

【Beyond Genes and Parents: The Effects of Cultural Products on Human PsychologyーEmanuele Castano】

🌟Report

“What on earth is the other person thinking?”

We rely on every piece of information—words, expressions, gestures, and the context up to that point—to communicate and understand each other. This ability is essential for living in groups, and we gradually acquire it through various experiences. So, what specifically influences the development of this ability?

“Fiction is a gym for social cognition”

This research idea proposed by Professor Castano may offer one possible answer to that question. There are many forms of “fiction,” but let us focus on novels. Furthermore, we can divide novels into literary fiction and popular fiction.In empirical research, people who read literary fiction frequently demonstrated higher ability to infer others’ mental states compared to those who read popular fiction, through several tests including the “Reading the Mind in the Eyes Test.” In literary fiction, compared to popular fiction, there is a tendency to depict complex relationships between characters and provide detailed psychological portrayals of each character. While it certainly offers substantial reading value and may require a higher level of effort to read, it might serve as a “gym” for training one’s ability to infer others’ psychological states.

It’s important to note that this research does not judge the quality of literary fiction versus popular fiction. Popular fiction also plays a very important role in literature, as it creates empathy and comfort through its use of familiar expressions and standardized narrative structures.
(Authorship: Tatsuro Ayatsuka)

🌟Minutes

The lecture focused on the role of cultural products in shaping the human mind, alongside genetics and upbringing. He emphasized that cultural products influence not only “what we think” but also “how we think,” and in particular, the differing effects of literary fiction and popular fiction were discussed empirically through a comparative analysis.

The background is that human social cognition—the ability to understand others’ emotions and thought and interpret the social world—is essential for maintaining social life and cultural communities. Professor Castano proposed that cultural products can be distinguished as either Confirming or Challenging. He particularly noted that literary fiction possesses a “challenging” character and has the potential to promote social cognition. Conversely, he hypothesized that popular fiction has a “confirming” character and serves to provide readers with reassurance.

In experimental studies, participants were randomly assigned to reading groups such as literary fiction, popular fiction, and nonfiction, and evaluated comparatively using several tasks including the “Author Recognition Test” and social cognition tasks like the “Reading the Mind in the Eyes Test.” The results showed that participants reading literary fiction scored significantly higher in social cognition, particularly in the ability to infer others’ mental states (Theory of Mind). In contrast, no similar effect was observed for popular fiction, suggesting its primary role lies in providing entertainment and comfort. Furthermore, surveys conducted across multiple cultures confirmed that this tendency is universal.

In the discussion, it was emphasized that literary fiction stimulates readers’ “imagination,” prompting them to fill in narrative gaps and thereby fostering complex understanding of others. This enables readers to acquire diverse perspectives beyond mere empathy, developing their ability to comprehend intricate social situations. Conversely, it was argued that popular fiction generates empathy and reassurance through familiar expressions and formulaic narrative structures, but its effect on enhancing social cognition is limited.

This lecture demonstrated how cultural products affect the human mind differently through the comparison of literary fiction and popular fiction. Literary fiction possesses a challenging character and functions to cultivate empathy and complex social cognition. Conversely, popular fiction has a confirming character, contributing to the psychological stability of communities through comfort and entertainment. It was concluded that the two are not in opposition; instead, each plays a distinctive role in shaping the psychological foundations of human society.
(Authorship: Wanwan Zheng, Humanity Center for Anthropocenic Actors and Agency, Graduate School of Humanities, Nagoya University)

🌟Q&A session

Question 1 (Prof. Nakamura):
For example, in the first edition of Grimm’s Fairy Tales, descriptions such as “he did this” and “she did that” are found, focusing solely on actions, with no depiction of the thoughts behind the actions. In contrast, by the seventh edition published several decades later, additions were made to include psychological descriptions like “He thought this at that moment, so he…,” not only showing actions but also thoughts and mental states. Historically, the tendency to explain human actions became more common, especially from the late 18th century to the early 19th century. From a historical perspective, how do you view this shift from “showing” to “telling”?

Answer:
That is a very sharp and insightful question. I was not aware of such changes in Grimm’s Fairy Tales, so I do not have much to say specifically about this transformation. However, your question reminds me of an experiment conducted in the 1990s from the perspective of developmental psychology, which examined the effects of reading materials on children.

In this experiment, elementary school students read ten picture books for more than two months. Half of the participants read books that included words describing psychological states like thoughts and emotions, while the other half read booksin which such words had been removed. For example, in the book about the fox and the chicken, the chicken is drawn wandering around, unaware of the approaching fox. The original picture book contains expressions like “the chicken turned around and… thought.” However, in the adjusted version, words indicating psychological states, such as “thought,” are removed.

It has already been shown that children whose parents frequently use words expressing emotions tend to acquire such vocabulary earlier. Therefore, the researchers who conducted the experiment wanted to observe how the act of reading influences children’s (around five years old) acquisition of vocabulary related to psychological states. Interestingly, the group of children who read the original picture books showed improvement in their ability to recognize the meanings of words expressing emotions. On the other hand, the group of children who read books without words describing psychological states performed better on tests assessing thinking and problem-solving skills.

For picture books read by children around five or six years old, the distinction between literary fiction and popular fiction is meaningless. However, some elements, such as vocabulary expressing emotions, can be said to be common to both. For very young children, such vocabulary needs to be explicitly presented. Without input, children simply cannot acquire those words nor can they express themselves. However, once they reach a certain stage, it is also important to consciously withhold such vocabulary and allow children to recognize the association between emotions and corresponding words based on their own judgment. Therefore, I believe both the original and the versions in which emotional vocabulary is omitted, as used in the earlier experiment, are necessary for children from 0 to 8 years old.

Question 2(Prof. Nakamura):
Literary fiction not only has a complex structure but often expresses negative emotions, as shown in the figure introduced in your presentation. This can potentially increase the stress felt by readers. Therefore, while children around seven or eight years old may be reluctant to read so-called literary works, I believe that individual development is fundamentally necessary for children to want to read literary fiction.

Answer:
The concept I am envisioning is that a kind of parallel system is important when telling stories to children. As I briefly mentioned in the first question, children need to read both fiction that directly includes words representing psychological states and fiction that does not. The former helps children acquire the meanings of the words themselves, while the latter encourages them to actively identify the connection between words and emotions. Therefore, I cannot give a clear answer on whether children should first develop psychologically before starting to read literary works, or whether reading literature leads to their development.

Adults already possess the ability to read emotions, so they do not lose this skill without training by reading literary fiction ( of course, it may become less sharp). However, regarding children, literary fiction that does not directly depict psychological states and emotions certainly trains their ability to read emotions, but this only happens after they have acquired the basic vocabulary to express those emotions. Drawing from my personal parenting experience, it’s fine to read the exact same story to a young child many times. But as they grow , they quickly get bored with the same story. Stories aimed at three-year-olds are always predictable. Many teachers here will understand this if they recall their own experiences. When you change just a few words in the same story for a child around three years old, they get upset right away. “No, that’s not the story!” they’ll say. But once they pass a certain age, if the story’s progression is too flat or predictable, children quickly lose interest. In other words, they start seeking interpretive possibilities or freedom in the story. Of course, the demand for predictable stories hasn’t disappeared entirely, as they still offer confirmation of established concepts. But it is natural that a demand for freedom has emerged.

Among adults too, there are those who read only literary fiction and those who read only popular fiction. This difference could be seen as a choice between seeking certainty in a work or seeking uncertainty. Most adults have acquired the vocabulary to describe psychological states, and adults can practice reading emotions and judging their type without relying on the written word. For example, even in today’s lecture, my ability was being trained, as I had to observe your expressions to understand whether my explanations were clear. Therefore, literature and films are by no means the only ways to train the ability to identify emotions. However, for children, an opportunity to acquire this ability without feedback from incorrect answers is indeed reading, especially reading fairy tales unfolding in a protected environment. Even if they feel fear or sadness, those are events in the fictional world and have no impact on the real world.

Therefore, acquiring vocabulary that expresses psychological states and identifying or describing one’s own psychological states can be said to be somewhat contradictory. To use a gym analogy, no matter how much you do basic training, you won’t build muscle, but it is also impossible to skip basic training and jump straight into advanced workouts. Returning to the beginning, I argue that constructing a parallel system is something that should be kept in mind when having children read literature.

Question 1 (Assoc. Prof. Beauvieux):
Regarding the impact of literary works on readers’ cognitive capabilities, does the distinction between whether a work is based on fact or fictional have any relevance to the impact itself?

Answer:
In my view, there is no such connection. In this presentation, I emphasized the distinction between pure literature and popular literature because I would like to discuss the linguistic structure of the works and the differences in vocabulary and their effects. Whether the work itself is based on a true story does not affect these elements. However, if the structure or writing style of the work causes readers to make different judgments internally regarding whether it is based on a true story, it is necessary to examine which parts of the work produced such differing judgments.

Question 2 (Assoc. Prof. Beauvieux):
You mentioned in your lecture that “unpredictability” is one of the important indicators for distinguishing pure fiction from popular fiction. For example, when reading popular fiction from another country, it might be easily predictable for people from that country, but foreign readers, due to differences in cultural background, may sometimes feel that they cannot predict what comes next. In that case, can unpredictability be considered a valid criterion?

Answer:
To rephrase what you just said, for people belonging to each culture, there exist several narrative models specific to that culture. When reading stories based on models from other cultures (though I believe the differences are only partial), it becomes difficult to predict what will happen. Therefore, for me, the first step is to understand how different these two models are. For example, in my research, I ask participants to read fiction, but before measuring their emotional abilities, I first ask them to evaluate the characters. When conducting experiments to examine this issue, if non-Japanese participants (e.g., French) find it difficult to predict characters in literary fiction by Japanese authors, I separately collect data on whether Japanese readers give the same responses. The same characters might be predictable for Japanese readers, but this only shows us that “literary fiction” and “popular fiction” are merely labels. We must acknowledge both universal aspects and the existence of cultural exceptions.

Question 3 (Assoc. Prof. Beauvieux):
The first-person “I” in poetry becomes a more universal “I,” rather than an individual within a specific society or environment. Therefore, could reading poetry also serve as training in the ability to identify emotions? If it can, can we distinguish differences in effect between literary fiction and popular fiction?

Answer:
I once reviewed a paper written by a Japanese scholar, which was a study on haiku and Noh—a complex art form using masks to express exaggerated emotions. Poetry is a shorter, more expressive form that can better evoke anxiety and uncertainty. So, it can be considered to have effects similar to literary fiction. However, the time spent reading a single poem is so short that it is still unclear how much reading affects one’s ability to identify or express emotions. Furthermore, some poems clearly belong to popular fiction in terms of structure and vocabulary. In any case, I believe there are still some difficulties in handling poetry in experimental research.
Authorship: Yiqun Xiao, Graduate School of Human and Environmental Studies, Kyoto University