Abstract
Everyday social life rests on the ability to identify and treat appropriately the objects that belong to oneself from those that belong to other people. Disregarding ownership status - whether an object is “mine” or “yours” - leads to predictable and costly social conflicts. Given the relevance and ubiquity of object ownership, a multidisciplinary literature has extensively explored the “motivational” problem of why and when ownership is respected by others. However a more fundamental “cognitive” problem still needs to be systematically addressed: how do we understand that something is “mine” or “yours” in the first place? To fill this gap we adopt a neurocomputational and experimental approach. We hypothesize that the conceptual domain of ownership (1) is grounded in the experience of control over external objects and basic self-other discrimination processes and that (2) it develops as a byproduct of restrictions experienced during curiosity-driven exploration in physical and social environments (Byproduct Hypothesis). To formalize and test this hypothesis, we will realise three different computational architectures of increasingly cognitive complexity. Adopting intrinsically-motivated reinforcement learning guided by competence improvement, the first architecture will model the interactive origins of basic sensorimotor representations of objects (sensorimotor categorization). The second architecture will introduce a form of curiosity-driven exploration to enable a more abstract form of categorization grounded in the physical controllability of objects (control-based categorization). The third architecture will show how the presence of another agent physically interacting with the objects together with self-other discrimination processes are sufficient to support the acquisition of abstract concepts of “mine” and “yours” (ownership categorization). The models of control-based and ownership categorization will be tested in two series of behavioral experiments both in real and in virtual reality environments. To assess the spontaneous formation of different categories, participants will complete a visuotactile interaction task and a categorical perception task. Beside the wide interdisciplinary relevance of these scientific outcomes, uncovering the cognitive bases of object ownership has a broad technological, economic and social impact, from assistive robotics to the design and support of new virtual forms of social interaction in the upcoming Metaverse. > Objective (1): a neurocomputational model Architecture 1: sensorimotor categorization. Architecture 2: control-based categorization. Architecture 3: ownership categorization. Expected results. Sensorimotor categorization model: We expect to find a topological categorization of internal representations of objects in the sensorimotor component that a) reflects the distinction between the four categories with smooth boundaries between each other; b) is biased towards a very detailed definition of more controllable categories (a wide region of the internal space dedicated to them) and a very poor definition of less controllable objects (a narrow region of internal space dedicated to them). We also expect to find an increment in the reactivity, precision of movements and reproducibility of movements in response to controllable rather than uncontrollable objects. Control-based categorization model: we predict that the simulated agent manages to correctly discriminate controllable from uncontrollable objects. Ownership categorization model: we predict a diminished motor response for an object presented in a “your” setting, where action is performed independently of the simulated agent. > Objective (2): Experiments with humans in real and virtual-reality environments - Human experimental series 1: control-based categorization. Visual-tactile interaction task (PR and VR). Categorical perception task. Expected results: Phase 1: PRE-Exploration. Differences between object categories are expected already in this baseline: weaker visuotactile interaction (VTI) effects are expected for unreachable objects, placed outside participants’ reaching space (blue), as compared to the other reachable objects. VTI in Phase 3: POST-Exploration. We predict an increase in VTI effects when subjects start moving to reach and grasp controllable objects, but not for non-controllable objects. For kinematic variables, in the congruent condition we expect a decrease in velocity and acceleration only for controllable objects, and not for uncontrollable ones. CP in Phase 3: POST-Exploration. We expect that the control-based categorization is revealed in the categorical perception task (both in PR and VR). If on top of the four “basic” categories, the two more abstract categories of controllable and uncontrollable objects have emerged, we expect that, for instance, participants will take longer to judge pairs of “green” and “violet” objects as different (and symmetrically with pairs of “red” and “blue” objects). - Human experimental series 2: ownership categorization. PRE- & POST-Exploration. Visual-tactile interaction task (PR and VR). Reach-grasp-lift task (only for PR). Categorical perception task. Expected results: VTI in Phase 3: POST-Exploration. For the OG we expect to find a slight increase of VTI effects for objects visually inspected by the participant alone with respect to the ones seen in the presence of the other agent. For the CG we predict a significant increase of VTI effects for grasping actions executed on objects that were previously controlled, compared with objects previously controlled by the Other. We predict a significant difference in VTI measures when comparing objects only looked at (OG, alone condition) vs. objects physically controlled (CG, alone condition) by participants in the learning stage. In this task, kinematic findings should show a decrease of acceleration and velocity of the wrist. Kinematic effects in Phase 3: POST-Exploration. For the reach-grasp-lift task, we expect participants lifting the object with greater acceleration, and drawing it closer to themselves, in case of objects that they visually inspected (rather than the ones visually expected by the Other Agent): the modulation should be greater in case of objects controlled by the participant. For the objects explored by the Other we predict a rightward drift, particularly in case of Others’ physical control (Constable et al 2011). CP in Phase 3: POST-Exploration. The categorization task for the CG in the “alone”vs. “social” condition should provide further evidence for the emergence of two distinct categories of (controllable) objects, namely those controllable by self and those controllable by the other.
Dettagli del progetto
Responsabile scientifico: Claudia Scorolli
Strutture Unibo coinvolte:
Dipartimento di Filosofia
Coordinatore:
CNR - Consiglio Nazionale delle Ricerche(Italy)
Contributo totale Unibo: Euro (EUR) 85.490,00
Durata del progetto in mesi: 28
Data di inizio
05/10/2023
Data di fine:
28/02/2026
