Most of our life is centred around objects, from the toys with which we play as toddlers, to the food we eat, to the clothes and tools we accumulate in our adult life. Although interactions with objects depend on understanding their physical nature, the fundamental cognitive ability to quickly identify, and differentially treat, what belongs to us from what belongs to other people is needed as well. Quite predictably, disregarding the more abstract ownership status of an object – whether an object is “mine” or “yours” – leads to costly, and potentially harmful, social conflicts.

Scientific and anecdotal evidence indicates that we (humans) all share a similar understanding of ownership and – at least for most of us, most of the time – an inclination to respect it. Given the underlying mutual interest to avoid such conflicts, a multidisciplinary effort has extensively explored the “motivational” problem of why ownership norms are respected (e.g., Sugden, Reference Sugden1986/2004). But where does the domain of abstract knowledge presupposed by these norms come from?

Given its evolutionary significance and cultural universality, it has been conjectured that the domain of ownership is part of our biological endowment (Jackendoff, Reference Jackendoff2002; Stake, Reference Stake2004), perhaps in the form of domain-specific principles that orient the development of a full-fledged naïve theory (Nancekivell, Friedman, & Gelman, Reference Nancekivell, Friedman and Gelman2019). Rooted in this tradition, Boyer maintains that the set of mental representations and cognitive processes making up our “ownership psychology” is a cognitive adaptation. However, in his view, the implicit theory which is at the origins of most of our intuitions about ownership derives from (computationally) linking a cognitive system evolved to support competition over resources with one evolved to take advantage of cooperation opportunities. As a cognitive adaptation, ownership is a by-product, so to say, of more fundamental ones.

Still, recent results from several studies in cognitive science suggest that knowledge of ownership status of objects shapes multisensory and motor “output” representations of a different kind from those addressed by Boyer. It has been shown, for instance, that knowing whether a graspable object like a cup is “yours” directly modulates the potentiation of actions towards it (affordance activation; Constable et al., Reference Constable2011). In a simple grasp-to-lift task, such knowledge can alter the kinematic profile of movements in ways that suggest an automatic resistance to interact with objects owned by others (Constable et al., Reference Constable2011). More subtly, knowledge of the ownership status of objects can influence the linguistic choice of spatial demonstratives in the form of a tendency to use “this” more often to refer to objects owned by the speaker rather than by someone else (Coventry et al., Reference Coventry2014). Intriguingly, a recent study has also provided evidence that the ownership status of an object can affect the multisensory representation of the space around the body (the peripersonal space; Patané, Brozzoli, Koun, Frassinetti, & Farnè, Reference Patané, Brozzoli, Koun, Frassinetti and Farnè2021) as measured by visuotactile interaction effects, which are differently modulated when manipulating objects that belong to someone else as if they are conceived as out of reach. Finally, intriguing evidence from a somatoparaphrenic patient denying ownership of her left hand revealed that she also displayed selective disownership of objects typically associated with it (e.g., a wedding ring, a garnet ring, a watch, etc.; Aglioti et al., Reference Aglioti, Smania, Manfredi and & Berlucchi1996; for an experimental demonstration with healthy participants, see, e.g., De Bortoli Vizioli, Borghi, & Tummolini, Reference De Bortoli Vizioli, Borghi and Tummolini2020). Taken together, these studies strongly suggest that the abstract conceptual domain of ownership may in fact be, at least partially, grounded in and profoundly shaped by our sensorimotor experiences (Borghi et al., Reference Borghi, Binkofski, Castelfranchi, Cimatti, Scorolli and Tummolini2017; Scorolli, Borghi, & Tummolini, Reference Scorolli, Borghi and Tummolini2018). However, despite this mounting evidence, which kind of experiences may be sufficient to learn this conceptual domain has not been identified.

One possibility is that ownership as a conceptual domain derives from our experience of physical control over external objects – their possession – in the context of interaction with other people. The first formulation of this hypothesis is due to the developmental psychologist Lita Furby who proposed that an early understanding of ownership concepts may develop even in prelinguistic infants as a “byproduct” of their intrinsic motivation to explore, manipulate and control our immediate environment (Furby, Reference Furby1980). More specifically, she conjectured that, because of such intrinsic motivation, during their first 2 years of life, infants are under pressure to identify the objects in their environment that give rise to feelings of efficacy and personal control, and to keep them apart from those that instead thwart such feelings owing to the interference from other people. The former class of controllable objects becomes the category of objects that are understood as belonging to the self, while the latter class includes those that are not. Furby's hints at a mechanistic explanation have been influential (e.g. Pierce, Kostova, & Dirks, Reference Pierce, Kostova and Dirks2003). However, as Boyer rightly suggests, such a developmental account has been insufficiently defended on explicit theoretical ground and no computational model has so far been proposed.

Fortunately, thanks to recent advances that have started to address the mechanisms behind intrinsically motivated learning and curiosity-driven exploration using the computational framework of reinforcement learning (e.g., Gottlieb, Oudeyer, Lopes, & Baranes, Reference Gottlieb, Oudeyer, Lopes and Baranes2013), we may now have the resources to overcome these limitations. Building on this approach, we have recently proposed a neurocomputational model of concept learning based on an intrinsic motivation to acquire control (competence) over the environment (Mannella & Tummolini, Reference Mannella and Tummolini2023). Using a computationally specified process model, we have shown that a mechanism supporting the alignment in an internal representation space of the (multi)sensory and motor maps acquired during experience also supports the formation of a cross-modal categorization system with sufficient resources to re-enact its own multi-modal experiences, and, on this basis, to kick-start the formation of concepts of objects grounded in the external environment. We propose that a similar architecture may also provide the resources to formally model Furby's conjecture. By monitoring one's competence (or lack thereof), the possibility to form concepts out of these internal, metacognitive experiences of controllability (Borghi, Fini, & Tummolini, Reference Borghi, Fini, Tummolini, Robinson and Thomas2021) may be able to show that ownership can in fact be a “byproduct” of quite a different process.