Abstract: Minimal Theory of Mind (MTM) has been regarded as a transitional form within the spectrum of the problem of other minds, situated between simple action perception and mature theory of mind. It is intended to explain the initial capacity for mindreading in infants, non-human animals, and other agents with limited cognitive resources. However, although MTM reduces the complexity of mental-state attribution, it still faces a fundamental theoretical difficulty: it presupposes that the target of cognition has already been identified as a social agent. Thus, MTM is minimal only at the level of mental-state representation, rather than at the ontogenetic point of departure for mindreading.
To address this problem, this study argues that a minimal account of understanding other minds should trace the developmental trajectory further back and begin with a more primitive capacity, namely animacy perception. Before invoking psychological concepts such as beliefs and desires, animacy perception enables individuals to distinguish objects in the environment as animate or inanimate, thereby providing the most basic object condition for subsequent other-mind-directed processing. In this sense, animacy perception constitutes a prior ontological condition for access to other minds and offers a new entry point for addressing the question left unresolved by MTM: how minded objects are first distinguished. On this basis, this study proposes a new constructive framework for the minimal form of mindreading, namely, Minimal Other Minds (MOM) based on animacy perception.
To support this proposal, this study reviews evidence on animacy perception from three aspects: visual cues, ontogenetic origins, and neural mechanisms. First, animacy perception can be triggered by both static and dynamic cues. Static cues include faces, eyes, body shapes, and texture-based appearance information. Face-like patterns, in particular, attract newborns’ attention and elicit early orienting responses in multiple species, indicating an early sensitivity to biologically relevant forms. Dynamic cues mainly include self-propelled motion, biological motion, and goal-directed motion. Even simple geometric figures may be perceived as animate if they display autonomous initiation, biological kinematic properties, or chasing-like relations. These findings suggest that animacy perception is not influenced by the perceiver’s own intentions or beliefs, but rather constitutes a highly rapid, spontaneous, and even irresistible perceptual process.
Second, the ability to distinguish animate from inanimate entities emerges early in ontogeny. Existing explanations mainly include nativist, empiricist, and integrative accounts. Nativist accounts emphasize inborn sensitivities to faces, biological motion, and self-propelled motion. Empiricist accounts stress the role of domain-general learning from environmental regularities. Integrative accounts hold that early perceptual biases provide the initial scaffold for animacy perception and are progressively extended and reorganized through experience.
Third, animacy perception relies on a distributed neural system involving the ventral occipitotemporal cortex, the lateral occipitotemporal cortex, and parietal regions associated with action understanding and goal processing. Particularly important is the neural model of animacy detection proposed by Shultz and colleagues. This model integrates three pathways responsible for static form cues, biological motion cues, and goal-directed motion cues, suggesting that animacy perception is a relatively independent processing system foundational for social cognition.
On this basis, this study further develops the MOM framework from three dimensions: process, component, and structure. In process terms, the route to other minds should begin not with mental-state attribution, but with the prior detection of whether an object is animate and socially relevant. In component terms, MOM encompasses the full content of animacy perception and serves as a foundational part of the social-cognitive framework, preparing the ground for higher-level mentalizing. In structural terms, MOM is grounded in a hierarchical object space of animacy, organized along dimensions such as being alive, resembling an animal, moving autonomously, acting independently, and being unpredictable. Correspondingly, animacy-sensitive neural pathways preferentially process this object space, together forming the external object basis and internal neural architecture of minimal social cognition.
The main contribution of this study lies in shifting the theoretical starting point of minimal mindreading from attenuated mental-state representation to prior animacy perception. By proposing MOM, this study reveals a theoretical gap in MTM at the ontogenetic level and provides a more fundamental framework for explaining how infants, animals, and other cognitively limited agents initially come to understand other minds. From this perspective, the genuinely minimal form of mindreading is not a weaker attribution of mental states, but the initial recognition that one is encountering a living, acting, and potentially minded other. More broadly, MOM helps bridge developmental psychology, comparative psychology, cognitive neuroscience, and philosophy of mind, while also offering a new reference point for research on direct social perception and computational modeling of early social cognition.

Key words: minimal other minds, animacy perception, theory of mind, mindreading, social cognition