Mentalisation, or the attribution of mental states to others, is a fundamental prerequisite for effective social interaction. It is not limited to human relations: as people increasingly engage with artificial systems in various domains of daily life, evidence suggests that mechanisms of mind attribution may come into play. In the educational domain, social robots are increasingly integrated into teaching and learning processes raising several methodological questions: how to design effective curricula integrating social robots (learning dimension), how to design effective teacher training (teaching dimension), how educational robots should be designed (technological dimension). To address these questions, it is important to understand how teachers conceptualise and interpret social robots. This research contributes to the philosophical and scientific debate on the attribution of mind to artificial systems investigating how individuals theorize about robots, integrating philosophical analysis with empirical research. The work focuses particularly on mentalistic modelling strategies: interpretative approaches through which individuals attribute cognitive capacities and mental states to robots to make sense of their behaviour. Traditionally, the attribution of mental states to robots has been studied through frameworks such as Dennett’s Intentional Stance and Theory of Mind, which explain behaviour in terms of beliefs, desires, and intentions, a strategy consistent with folk psychology at the basis of human social interaction. Moving beyond this perspective, the present work distinguishes and illuminates different styles of modelling the robotic mind that people may adopt in human-robot interaction. It introduces and outlines the characteristics of the folk cognitivist stance as a possible alternative mentalistic modelling strategy based on the theoretical vocabulary of classical cognitive science, which conceives the mind as an information-processing system organised into functional modules that elaborate representations. From a theoretical standpoint, the thesis argues that Dennett’s design stance, often treated in empirical research in human-robot interaction as a non-mentalistic perspective, can instead encompass mentalistic interpretations. Accordingly, the folk-cognitivist stance is construed as a mentalistic type of design stance. Theoretical insights have been further refined and broadened through two empirical studies. The first study investigated what possible styles people may adopt to model the minds of robots. It resulted in a taxonomy of possible ways in which people may understand and represent the mind of robots, accompanied by a codebook designed to support present and future research. The second study applied the taxonomy and codebook developed within the first study to the education domain, exploring the mentalistic modelling styles teachers adopt when interpreting robotic behaviour. Modelling styles are studied through the analysis of the explanations of robotic behaviour given by the participants, collected via semi-structured interviews. In a first stage of the analysis, explanations have been qualitatively extracted from the interviews transcripts as pairs <explanandum, explanans>. The first study employed template analysis of the explanations given by 16 participants (adults with various educational and professional backgrounds). The second study applied the taxonomy and the codebook developed in the first study to qualitatively analyse explanations given by 16 lower secondary school teachers. The results provided insights into the sociocultural embedding of cognitive science concepts, the influence of educational background on the modelling styles adopted by teachers towards robots, implications for the design of teacher education programmes and instructional approaches involving socially evocative robots, and broader considerations concerning explainable AI (XAI) in educational contexts.

La ricerca si propone di contribuire al dibattito filosofico e scientifico sull’attribuzione di mente ai sistemi artificiali, indagando come gli individui teorizzano sui robot attraverso l’integrazione di analisi filosofica e ricerca empirica. Il lavoro si concentra in particolare sulle strategie di modellazione mentalistica: approcci interpretativi attraverso i quali gli individui interpretano il comportamento dei robot attribuendo capacità cognitive e stati mentali al sistema. Tradizionalmente, l’attribuzione di stati mentali ai robot è stata studiata attraverso cornici teoriche come l’atteggiamento intenzionale (intentional stance) di Dennett e la Teoria della Mente, che spiegano il comportamento del sistema in termini di credenze, desideri e intenzioni, una strategia coerente con la psicologia del senso comune alla base dell’interazione sociale umana. Allargando questa prospettiva, il presente lavoro distingue e mette in luce diversi possibili stili di modellazione della mente robotica che le persone possono adottare. Viene introdotta e descritta una possibile strategia alternativa di modellazione mentalistica, il ‘cognitivismo (o scienza cognitiva) del senso comune’ basata sul vocabolario teorico della scienza cognitiva classica, che concepisce la mente come un sistema di elaborazione delle informazioni organizzato in moduli funzionali che trasformano rappresentazioni. Da un punto di vista teorico, la tesi sostiene che l’atteggiamento del progetto (design stance) di Dennett, spesso trattato nella ricerca empirica sull’interazione umani-robot come una prospettiva non mentalistica, possa invece comprendere interpretazioni di tipo mentalistico. Di conseguenza, il cognitivismo del senso comune è inteso come una variante mentalistica della posizione progettuale di Dennett. Gli approfondimenti teorici sono stati ulteriormente sviluppati e consolidati attraverso due studi empirici. Il primo studio ha indagato i possibili stili che le persone possono adottare per modellare la mente dei robot, portando alla definizione di una tassonomia delle diverse modalità attraverso le quali gli individui comprendono e rappresentano la mente dei robot, accompagnata da un codebook volto a supportare ricerche empiriche attuali e future. Il secondo studio ha applicato la tassonomia e il codebook sviluppati nel primo studio al contesto educativo, esplorando gli stili di modellazione mentalistica che gli insegnanti adottano nell’interpretare il comportamento dei robot. Gli stili di modellazione sono stati analizzati attraverso lo studio delle spiegazioni del comportamento dei robot fornite dai partecipanti, raccolte tramite interviste semi-strutturate. In una prima fase dell’analisi, le spiegazioni sono state qualitativamente estratte dalle trascrizioni delle interviste come coppie <explanandum, explanans>. Per il primo studio è stata condotta una Template Analysis delle spiegazioni fornite da 16 partecipanti adulti con diversi background formativi e professionali, mentre il secondo studio ha applicato la tassonomia e il codebook per analizzare qualitativamente le spiegazioni fornite da 16 insegnanti di scuola secondaria di primo grado. I risultati hanno offerto spunti di riflessione su diversi temi: l’incorporazione socioculturale dei concetti delle scienze cognitive, l’influenza del background educativo sugli stili di modellizzazione adottati dagli insegnanti nei confronti dei robot, le implicazioni per il progetto della formazione degli insegnanti e per gli approcci didattici che coinvolgono robot socialmente evocativi oltre a considerazioni più ampie sulle implicazioni nel campo della explainable AI (XAI) in contesti educativi.

Larghi, S (2026). How do teachers understand robots? A philosophical and empirical study of the attribution of mind to robots. (Tesi di dottorato, , 2026).

How do teachers understand robots? A philosophical and empirical study of the attribution of mind to robots

LARGHI, SILVIA
2026

Abstract

Mentalisation, or the attribution of mental states to others, is a fundamental prerequisite for effective social interaction. It is not limited to human relations: as people increasingly engage with artificial systems in various domains of daily life, evidence suggests that mechanisms of mind attribution may come into play. In the educational domain, social robots are increasingly integrated into teaching and learning processes raising several methodological questions: how to design effective curricula integrating social robots (learning dimension), how to design effective teacher training (teaching dimension), how educational robots should be designed (technological dimension). To address these questions, it is important to understand how teachers conceptualise and interpret social robots. This research contributes to the philosophical and scientific debate on the attribution of mind to artificial systems investigating how individuals theorize about robots, integrating philosophical analysis with empirical research. The work focuses particularly on mentalistic modelling strategies: interpretative approaches through which individuals attribute cognitive capacities and mental states to robots to make sense of their behaviour. Traditionally, the attribution of mental states to robots has been studied through frameworks such as Dennett’s Intentional Stance and Theory of Mind, which explain behaviour in terms of beliefs, desires, and intentions, a strategy consistent with folk psychology at the basis of human social interaction. Moving beyond this perspective, the present work distinguishes and illuminates different styles of modelling the robotic mind that people may adopt in human-robot interaction. It introduces and outlines the characteristics of the folk cognitivist stance as a possible alternative mentalistic modelling strategy based on the theoretical vocabulary of classical cognitive science, which conceives the mind as an information-processing system organised into functional modules that elaborate representations. From a theoretical standpoint, the thesis argues that Dennett’s design stance, often treated in empirical research in human-robot interaction as a non-mentalistic perspective, can instead encompass mentalistic interpretations. Accordingly, the folk-cognitivist stance is construed as a mentalistic type of design stance. Theoretical insights have been further refined and broadened through two empirical studies. The first study investigated what possible styles people may adopt to model the minds of robots. It resulted in a taxonomy of possible ways in which people may understand and represent the mind of robots, accompanied by a codebook designed to support present and future research. The second study applied the taxonomy and codebook developed within the first study to the education domain, exploring the mentalistic modelling styles teachers adopt when interpreting robotic behaviour. Modelling styles are studied through the analysis of the explanations of robotic behaviour given by the participants, collected via semi-structured interviews. In a first stage of the analysis, explanations have been qualitatively extracted from the interviews transcripts as pairs . The first study employed template analysis of the explanations given by 16 participants (adults with various educational and professional backgrounds). The second study applied the taxonomy and the codebook developed in the first study to qualitatively analyse explanations given by 16 lower secondary school teachers. The results provided insights into the sociocultural embedding of cognitive science concepts, the influence of educational background on the modelling styles adopted by teachers towards robots, implications for the design of teacher education programmes and instructional approaches involving socially evocative robots, and broader considerations concerning explainable AI (XAI) in educational contexts.
DATTERI, EDOARDO
philosophy of AI; intentional stance; theory of mind; cognitive science; explanation
philosophy of AI; intentional stance; theory of mind; cognitive science; explanation
English
10-mar-2026
38
2024/2025
embargoed_20290310
Larghi, S (2026). How do teachers understand robots? A philosophical and empirical study of the attribution of mind to robots. (Tesi di dottorato, , 2026).
File in questo prodotto:
File Dimensione Formato  
phd_unimib_878622.pdf

embargo fino al 10/03/2029

Descrizione: Silvia Larghi - PhD Thesis final
Tipologia di allegato: Doctoral thesis
Dimensione 4.89 MB
Formato Adobe PDF
4.89 MB Adobe PDF   Visualizza/Apri   Richiedi una copia

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/10281/610782
Citazioni
  • Scopus ND
  • ???jsp.display-item.citation.isi??? ND
Social impact