Beyond Human-Like Robots: Defining What We See and What We Build

by

Author: Denis Avetisyan

A new review clarifies the critical difference between perceiving human qualities in robots and intentionally designing them, paving the way for more accountable and effective human-robot interaction.

This paper disentangles anthropomorphism – the attribution of human traits to robots – from anthropomimesis – the deliberate incorporation of human-like features in robot design.

Despite increasing integration of robots into daily life, a consistent theoretical framework distinguishing how humans perceive human-like qualities versus how developers design them remains elusive. This paper, ‘Disambiguating Anthropomorphism and Anthropomimesis in Human-Robot Interaction’, clarifies this distinction by defining anthropomorphism as user perception and anthropomimesis as design implementation-essentially, separating ‘seeing’ humanness from ‘building’ it. Establishing this conceptual clarity is crucial for both advancing Human-Robot Interaction scholarship and fostering accountability in robot design. How will a nuanced understanding of these concepts shape the development of more effective and ethically-aligned social robots?

The Echo of Humanity: Perceiving Intent in Artificial Forms

The success of Human-Robot Interaction hinges on a comprehensive understanding of how humans instinctively interpret and react to robots. This perception isn’t merely about recognizing a machine; it’s a complex cognitive process where individuals actively assign meaning, intentions, and even emotional states to robotic entities. Consequently, researchers are focused on dissecting the factors that shape this perception – encompassing a robot’s physical appearance, its movements, its responsiveness, and the context of the interaction. Without accurately modeling these perceptual mechanisms, developers risk creating robots that are either unsettlingly alien or frustratingly unintelligible, hindering their potential for seamless integration into human lives and limiting the scope of beneficial applications.

Humans possess a remarkable propensity to imbue non-human entities – including robots – with distinctly human characteristics, a phenomenon known as anthropomorphism. This isn’t merely a whimsical tendency; it’s a fundamental cognitive shortcut that allows for quicker understanding and prediction of behavior. By projecting emotions, intentions, and even personalities onto robots, individuals can navigate interactions with greater ease, interpreting actions through a familiar human framework. The degree to which this occurs is influenced by subtle cues in a robot’s design – its movements, vocalizations, and even physical appearance – demonstrating that anthropomorphism isn’t a fixed trait, but rather a dynamic response shaped by perceived similarities to themselves. Consequently, understanding this inclination is paramount, as it directly impacts how people perceive, trust, and ultimately interact with robotic systems.

Human perception of robots isn’t a default reaction, but is actively shaped by deliberate design choices. Developers, consciously or not, embed cues – in a robot’s morphology, movement, or even its sonic emissions – that trigger the human tendency to project human characteristics. These design elements aren’t neutral; a robot with eyes, for instance, immediately invites social interaction in a way a purely functional machine wouldn’t. Furthermore, the degree to which these human-like features are emphasized – whether subtle or overt – profoundly influences how readily people accept and interact with the robot. Consequently, understanding this dynamic is paramount; responsible robot design requires a nuanced awareness of how specific features can encourage, or discourage, desired forms of human-robot engagement, ultimately impacting the effectiveness and acceptance of these technologies.

A nuanced understanding of anthropomorphism – the projection of human characteristics onto robots – and its often-overlooked counterpart, anthropomimesis, is paramount for crafting robots that interact seamlessly and ethically with people. Anthropomimesis describes the tendency of robots to elicit human-like responses, even without explicit human features; a robot moving in a predictable, goal-oriented way can trigger empathetic reactions. Responsible design necessitates acknowledging that these phenomena aren’t merely aesthetic choices, but fundamental drivers of human perception and behavior. By precisely defining and accounting for both how humans attribute qualities to robots and how robots encourage human responses, developers can move beyond superficial mimicry and create interactions that are genuinely intuitive, trustworthy, and avoid unintended social or emotional consequences, ultimately fostering a more positive and productive human-robot relationship.

Defining the Boundaries: Anthropomorphism and Its Counterpart

Definitions of anthropomorphism vary within the research literature, encompassing a spectrum of conceptualizations. Early work, such as that by Epley et al. (2007), often defines anthropomorphism as the attribution of human characteristics – such as personality traits, intentions, or emotions – to non-human entities. Later studies, including Bartneck et al. (2009), broadened this definition to include the ascription of complete human form and behavioral patterns, moving beyond simple characteristic attribution to encompass a more holistic replication of human qualities in non-human entities. This range reflects the complexity of the phenomenon and contributes to ongoing debate regarding a unified definition.

Current research differentiates between anthropomorphism and anthropomimesis as distinct concepts. Anthropomorphism is defined as the inherent human propensity to project human characteristics, emotions, or intentions onto non-human entities. Conversely, anthropomimesis refers specifically to the deliberate incorporation of human-like features or behaviors into the design of objects, robots, or interfaces. This distinction, proposed by Shevlin (2025), clarifies that anthropomorphism is a cognitive tendency while anthropomimesis is the active implementation of that tendency through design choices.

A review of 57 studies on anthropomorphism, published between 2000 and 2020, demonstrates a considerable lack of definitional consistency within the field. The analysis identified seven distinct definitions of anthropomorphism currently employed by researchers, indicating a fragmented understanding of the concept. This inconsistency hinders comparative analysis and cumulative knowledge building, suggesting a need for greater clarity and standardization in future research regarding the definition of anthropomorphism.

Research by Fink (2012) demonstrates that anthropomorphic perception is not solely a property of the designed object, but rather emerges from the user’s cognitive processes and prior experiences. This perspective shifts the focus from inherent qualities of an object to the active role of the observer in projecting human characteristics. The attribution of human qualities is thus understood as a subjective process, influenced by individual differences, contextual factors, and the user’s propensity to seek patterns and meaning. Consequently, an object may be perceived as anthropomorphic by one user but not another, depending on their individual cognitive frameworks and interpretive biases.

Quantifying the Human Form: Metrics for Likeness

The Godspeed Questionnaire is a widely employed instrument in human-robot interaction research designed to quantify a user’s perception of an agent’s anthropomorphic qualities. It comprises a series of Likert-scale questions assessing attributes such as perceived intelligence, likeability, attractiveness, eeriness, and overall ‘human-likeness’. Responses are aggregated to generate scores for each dimension, providing researchers with a standardized metric for evaluating how closely users perceive an agent – typically a robot or virtual character – as resembling a human. The questionnaire’s validity has been demonstrated across diverse robotic platforms and user groups, making it a common benchmark for assessing anthropomorphic perception and its influence on user interaction.

The ABOT Database is a standardized resource for quantifying physical human-likeness in robots and virtual agents. It employs a scoring system based on 14 measurable characteristics, categorized as global (height, width, length) and local (e.g., head size, arm length, shin length, hand size, foot size). These dimensions are normalized by the average human measurements, and the resulting values are used to calculate an ABOT score, ranging from 0 to 1. A higher ABOT score indicates greater physical resemblance to a human; the database facilitates comparative analysis of anthropomorphic designs and provides a framework for objectively evaluating the physical human-likeness of artificial entities.

Anthropomimesis, the imitation of human characteristics in non-human entities, is categorized into distinct dimensions of replication. Aesthetic Anthropomimesis focuses on visual design, incorporating human-like features in appearance and form. Behavioral Anthropomimesis centers on replicating human actions, movements, and interaction patterns, often seen in robotics and virtual agents. Finally, Substantive Anthropomimesis involves mimicking underlying biological structures and materials, representing the most fundamental level of human replication, potentially utilizing bio-inspired materials or even synthetic organs to achieve a physiological resemblance.

The Shadow of Familiarity: Navigating the Uncanny Valley

The pursuit of increasingly realistic robots inadvertently courts a peculiar phenomenon known as the Uncanny Valley. This concept posits that as a robot’s appearance and movements become more human, positive emotional responses from observers increase – but only up to a point. Beyond that threshold, even minor deviations from perfect human realism – a slightly unnatural gait, subtly vacant eyes, or imperfect skin texture – trigger a sense of unease, revulsion, and even fear. This dip in affinity isn’t simply a matter of aesthetics; it suggests a deeply ingrained psychological mechanism where the brain flags these near-human entities as potentially unsettling anomalies, perhaps activating threat-detection systems or eliciting feelings associated with disease or death. Consequently, designers must carefully balance the desire for realism with the risk of falling into this ‘valley’, where subtle imperfections become profoundly disturbing.

The successful integration of robots into human society hinges on a delicate balance between familiarity and realism, a challenge addressed through the principles of anthropomorphism and anthropomimesis. Anthropomorphism, the attribution of human characteristics to non-human entities, and anthropomimesis, the imitation of human movements and behaviors, are crucial for fostering acceptance, yet these approaches are fraught with risk. While some degree of human-likeness can enhance relatability, exceeding a certain threshold – or implementing it imperfectly – can trigger a negative emotional response. Poorly executed facial expressions, unnatural gaits, or inconsistencies between appearance and behavior introduce unsettling discrepancies, leading to feelings of unease and even revulsion. Consequently, designers must carefully consider the extent to which they imbue robots with human qualities, prioritizing believable execution over simply striving for photorealistic accuracy; a subtly stylized or clearly artificial appearance may, paradoxically, prove more palatable than a flawed attempt at perfect replication.

Effective robotic design necessitates a detailed comprehension of human perceptual and emotional responses to increasingly realistic machines. Investigations into this area reveal that subtle deviations from expected human characteristics – in movement, appearance, or behavior – can trigger feelings ranging from discomfort to outright revulsion, a phenomenon linked to our innate ability to detect anomalies. Designers must therefore move beyond simply replicating human features; instead, they require a nuanced approach that considers the psychological impact of each design choice, carefully balancing realism with elements that reinforce a perception of ‘otherness’ or clearly signal non-human identity. This involves understanding how humans attribute intention and emotion to robots, and proactively mitigating potential negative reactions through careful attention to detail and a deep understanding of human social cognition.

The study dissects the subtle, yet critical, difference between perceiving human characteristics in robots-anthropomorphism-and designing those characteristics into them-anthropomimesis. This distinction isn’t merely semantic; it carries significant weight for establishing accountability in human-robot interaction. As G.H. Hardy observed, “The essence of mathematics lies in its simplicity,” and a similar principle applies here. By clearly defining these concepts, the research seeks to simplify the complex landscape of HRI, allowing for a more rigorous examination of how human-like designs influence perception and, ultimately, responsibility. The article suggests that a nuanced understanding of both concepts is crucial for navigating the ethical considerations inherent in increasingly sophisticated robotic systems; time, as the medium in which these systems operate, demands this clarity.

What Lies Ahead?

The disentangling of anthropomorphism and anthropomimesis, as this work demonstrates, is less a resolution than a refinement of the questions. The tendency to project human qualities onto non-human entities isn’t diminished by understanding how those qualities are deliberately embedded in design; rather, the locus of responsibility shifts. Technical debt accumulates in these designs, mirroring the natural erosion of initial intent. A robot built to appear empathetic does not possess empathy, and the failure to recognize this isn’t a bug, but a predictable state.

Future inquiry must address the temporal dynamics of this perceived likeness. Uptime – a fleeting phase of temporal harmony where a robot successfully performs as intended – is not sustainability. As designs age, the seams between intention and execution fray, revealing the artifice beneath. The challenge isn’t to eliminate anthropomorphism – an inherent facet of perception – but to design for its inevitable reinterpretation over time.

Ultimately, the field edges towards a consideration of accountability. When a human-like robot fails, to whom – or what – is the failure attributed? The designer, the programmer, the user, or the system itself? This isn’t merely a technical problem, but a philosophical one, forcing a reckoning with the boundaries of agency and the illusion of control in increasingly complex systems.

Original article: https://arxiv.org/pdf/2602.09287.pdf

Contact the author: https://www.linkedin.com/in/avetisyan/

See also:

2026-02-11 12:55