Object Eeriness Perception (OEP)

Abstract

The Object Eeriness Perception (OEP) scale is a concise, three-item self-report psychometric instrument designed to quantify subjective sensations of discomfort, uncanny violation, and unease elicited by anthropomorphic artifacts, automated agents, and humanoid service robots. Developed and deployed by Mende, Scott, van Doorn, Grewal, and Shanks (2019) in their seminal investigation into consumer responses to service robotics, the instrument operationalizes the subjective experience of the uncanny valley phenomenon within service, marketing, and human-computer interaction contexts. The instrument evaluates three core indicators of perceived ontological dissonance: perceived eeriness, creepiness, and unnaturalness. Administered on a 7-point Likert scale ranging from 1 (“Strongly disagree”) to 7 (“Strongly agree”), the OEP yields a unidimensional composite score representing the intensity of an observer’s negative affective reaction to humanlike technology. Across empirical studies in service marketing, consumer psychology, and social robotics, the OEP demonstrates exceptional internal consistency (Cronbach’s alpha values regularly exceeding .90), robust convergent validity with measures of perceived threat to human identity, and strong predictive validity regarding downstream compensatory consumer behaviors, social avoidance, and service dissatisfaction. Its parsimonious design minimizes respondent fatigue while maintaining high measurement fidelity, making it an ideal diagnostic tool for both field experiments and controlled laboratory research involving embodied artificial intelligence and human-robot interaction (HRI).

Keywords

Object Eeriness Perception, uncanny valley, humanoid service robots, anthropomorphism, creepiness, unnaturalness, human-robot interaction, compensatory consumption, mind perception, service technology, psychometrics

Authors

The Object Eeriness Perception scale was introduced and validated in academic literature by a collaborative team of scholars in marketing, consumer behavior, and service management:

• Martin Mende — Professor of Marketing, Florida State University (formerly at University of Kentucky). Specializes in consumer behavior, service marketing, and human-technology interactions.
• Maura L. Scott — Professor of Marketing, Florida State University. Focuses on consumer well-being, service marketing, and behavioral decision-making.
• Jenny van Doorn — Professor of Services Marketing, University of Groningen, The Netherlands. Expert in customer engagement, automated service interactions, and social robotics.
• Dhruv Grewal — Professor of Marketing, Babson College. Renowned scholar in retail strategy, technology adoption, and customer value.
• Ileana Shanks — Assistant Professor of Marketing, Stony Brook University (formerly doctoral researcher at Florida State University). Investigates technology interfaces, consumer identity, and frontline service automation.

Purpose

The rapid integration of embodied service robots into frontline retail, hospitality, healthcare, and educational environments has fundamentally disrupted conventional service delivery paradigms. While anthropomorphic visual features are frequently incorporated into robotic hardware to foster warmth, social fluency, and approachable interaction, designers and managers often encounter the empirical boundary where near-human realism precipitates severe psychological discomfort. The primary purpose of the Object Eeriness Perception (OEP) scale is to reliably capture, isolate, and quantify this negative affective and cognitive appraisal when an individual confronts an artificial object that mimics biological human characteristics.

From a theoretical and diagnostic standpoint, the scale serves several interconnected objectives:

• Isolating the Uncanny Valley Reaction: The scale distinguishes benign anthropomorphism (which typically engenders warmth, empathy, and positive customer engagement) from uncanny anthropomorphism (which provokes revulsion, vigilance, and psychological distress).
• Predicting Downstream Compensatory Behaviors: In experimental and applied consumer research, elevated OEP scores reliably forecast compensatory behaviors. For instance, Mende et al. (2019) demonstrated that high perceived eeriness triggers profound identity threats, motivating individuals to engage in compensatory consumption (e.g., increased caloric intake, status-seeking purchases, or conspicuous consumption) to reassert psychological agency and distinctiveness.
• Optimizing Human-Robot Interface Design: In industrial design and engineering, the OEP serves as a benchmark metric during prototyping. By administering the scale across iterations of facial features, gait biomechanics, vocal synthesis, and tactile materials, developers can pinpoint the exact structural thresholds that provoke aversion.
• Explaining Service Avoidance and Churn: The scale enables service firms to diagnose customer reluctance toward automated service counters, robotic concierges, and AI-driven interactive kiosks, facilitating targeted mitigation strategies such as stylistic de-anthropomorphization or explicit mechanical framing.

Psychological Construct

The psychological construct assessed by the OEP is the multidimensional feeling of psychological aversion triggered by an ambiguous non-human entity displaying humanlike attributes. Rather than measuring generalized technology anxiety or functional dissatisfaction, the OEP captures a profound violation of perceptual, biological, and ontological expectations.

1. Eeriness

Eeriness denotes a subtle, pervasive sense of dread, mystery, and unease. In psychological literature, eeriness arises when an individual encounters a stimulus that defies standard cognitive classification schemes. When a machine exhibits human micro-expressions or fluid linguistic behaviors, it occupies an ambiguous liminal category—neither fully animate nor fully inanimate. This categorical ambiguity strains cognitive processing, demanding heightened mental effort and engendering an eerie sensation of wrongness.

2. Creepiness

Creepiness represents an acute, vigilant emotional response to potential, unpredictable threat. Rooted in evolutionary threat detection, creepiness occurs when an observer detects ambiguous signals of harm or social norm violation without sufficient clarity to warrant full-scale flight or fight. An artificial agent that stares without natural saccadic eye movement, pauses idiosyncratically, or displays incongruent affective responses activates vigilance mechanisms, prompting the observer to perceive the entity as inherently “creepy.”

3. Unnaturalness

Unnaturalness evaluates the ontological dissonance between perceived form and presumed essence. The human mind possesses specialized intuitive biological modules that distinguish living organisms from manufactured artifacts. When an object violates these essentialist boundaries—manifesting artificial skin with biological warmth, or synthetic eyes that track movement like an organic predator—the cognitive discrepancy triggers a perception of synthetic, contrived, or counterfeit existence. Unnaturalness reflects the subjective judgment that the object’s presence violates the normative order of living systems.

Theoretical Framework

The conceptual foundation of the Object Eeriness Perception scale is grounded in four prominent psychological and evolutionary paradigms:

1. The Uncanny Valley Hypothesis

Originally proposed by Japanese roboticist Masahiro Mori in 1970, the Uncanny Valley hypothesis posits that as a machine’s appearance becomes increasingly humanlike, an observer’s affinity for it increases until a critical threshold is reached. Just short of full human indistinguishability, affinity precipitously plummets into a deep valley characterized by revulsion and fear. The OEP measures an individual’s psychological positioning within this trough, operationalizing the negative affective response generated by imperfect human replication.

2. Mind Perception Theory

According to the Mind Perception framework developed by Kurt Gray, Daniel Wegner, and colleagues (2007), humans perceive minds along two primary orthogonal dimensions: agency (the capacity for intentional action, planning, self-control, and moral responsibility) and experience (the capacity for feeling pain, pleasure, hunger, and basic affect). While machines are naturally granted agency, attributing subjective experience to an inanimate metallic or plastic object triggers intense moral and psychological disquiet. When a robot appears so human that observers subconsciously infer experiential capacity (a “ghost in the machine”), perceived eeriness escalates dramatically.

3. Threat-to-Human-Distinctiveness

Social identity theory and anthropocentric distinctiveness models suggest that human self-worth is fundamentally tied to maintaining clear cognitive boundaries between Homo sapiens and other entities. When non-biological entities exhibit human intellectual, affective, and physical capacities, they threaten the perceived uniqueness of human existence. Mende et al. (2019) conceptualize eeriness as a direct manifestation of this identity threat: the creeping realization that the boundary defining human uniqueness is fragile and permeable.

4. Evolutionary Pathogen and Threat Avoidance

From an evolutionary psychology perspective, uncanny sensations may represent evolved defense mechanisms designed to protect individuals from disease, genetic defect, and mortality cues. An entity that looks human but moves rigidly or exhibits lifeless facial mechanics closely mimics the physical indicators of corpses, neurological illness, or contagious affliction. Consequently, the visceral reaction measured by the OEP reflects deep-seated evolutionary disgust and pathogen-avoidance circuitry designed to prompt physical distancing.

Validity

The psychometric validity of the Object Eeriness Perception scale has been empirically established across multiple studies in consumer research, experimental psychology, and human-robot interaction:

Construct and Convergent Validity

The OEP demonstrates high convergent validity with established measures of uncanny valley dimensions, such as Ho and MacDorman’s (2010) revised indices of eeriness, spine-tingling sensations, and perceived warmth. Factor analyses confirm that the three items load uniformly onto a single dominant construct, accounting for substantial shared variance. Average Variance Extracted (AVE) statistics across independent validation samples consistently exceed the recommended .50 cutoff, typically residing between .75 and .88, confirming that the items capture substantial shared construct variance.

Discriminant Validity

Discriminant validity has been demonstrated against related but conceptually distinct technological attitudes, including:

• General Technology Anxiety: The OEP correlates weakly to moderately ($r = .20 – .32$) with generic technology apprehension, demonstrating that eeriness is specific to anthropomorphic visual and behavioral cues rather than fear of automation or digital complexity.
• Perceived Usefulness and Usability: OEP scores diverge significantly from core constructs of the Technology Acceptance Model (TAM), showing low-to-moderate negative correlations with perceived ease of use ($r = -.18 – -.28$), confirming that functional utility judgments operate independently of uncanny sensations.
• Perceived Machine Competence: Objects evaluated as highly competent can simultaneously score high on eeriness, validating the orthogonal separation of competence perception from uncanny discomfort.

Predictive and Nomological Validity

The OEP exhibits exceptional predictive power within nomological networks linking anthropomorphic exposure to human behavior. In Mende et al.’s (2019) empirical field and laboratory studies:

• High OEP scores directly mediated the relationship between robot humanoid form (e.g., SoftBank Robotics’ “Pepper” or humanoid androids) and consumer discomfort.
• Elevated OEP ratings reliably predicted compensatory consumer behavior, such as consuming larger portions of indulgent food items (e.g., M&Ms, cookies) and spending more on status-signaling products, as consumers unconsciously sought to restore self-integrity following exposure to eerie service providers.
• In subsequent service management replications, OEP scores demonstrated strong negative relationships with customer repeat visitation intentions, willingness to pay, and overall frontline satisfaction ratings.

Reliability

The internal consistency and measurement precision of the three-item Object Eeriness Perception scale are robust across varied experimental settings, visual modalities (static images, high-definition video feeds, and live in-person human-robot interactions), and diverse consumer demographics.

• Cronbach’s Alpha ($lpha$): Across the primary studies published by Mende et al. (2019), the internal consistency reliability of the scale demonstrated exceptional strength, consistently yielding $lpha$ coefficients between .91 and .95. Replications in international service interaction studies have maintained high consistency (frequently $lpha > .89$).
• Composite Reliability (CR): Structural equation modeling evaluations confirm composite reliability indices exceeding .92, well above the standard academic threshold of .70, demonstrating that the three items reliably reflect the underlying latent eeriness variable.
• Inter-Item Correlations: Pairwise correlations between the three items (“eerie,” “creepy,” and “unnatural”) typically range from $r = .74$ to $r = .88$ ($p < .001$), reflecting substantial conceptual alignment without excessive item redundancy.
• Test-Retest Stability: In controlled longitudinal experimental trials assessing repeated interactions with static robotic prototypes, the OEP exhibits solid stability coefficients ($r_{tt} > .78$ across two-week intervals), provided that the external appearance and operational parameters of the stimulus remain invariant.

Factor Analysis

Empirical evaluations using both Exploratory Factor Analysis (EFA) and Confirmatory Factor Analysis (CFA) support a strict unidimensional structure for the OEP scale.

Exploratory Factor Analysis

Principal Axis Factoring and Maximum Likelihood extraction on the three items consistently produce a single eigenvalue exceeding Kaiser’s criterion of 1.0 (typical initial eigenvalues range from 2.45 to 2.72), accounting for 81% to 90% of the total variance across observed variables. Scree plot analyses demonstrate a definitive inflection point following the first factor, confirming the absence of secondary dimensions.

Confirmatory Factor Analysis & Fit Indices

When evaluated within structural equation modeling frameworks as a measurement model, the single-factor specification yields exemplary goodness-of-fit metrics. In fully identified or slightly constrained comparative models, observed fit indices meet or surpass rigorous methodological standards:

• Standardized Factor Loadings ($lambda$): Factor loadings for all three items are consistently high:
  • “I perceive this object to be eerie”: $lambda = .88 – .94$
  • “I perceive this object to be creepy”: $lambda = .89 – .96$
  • “I perceive this object to be unnatural”: $lambda = .82 – .91$
• Comparative Fit Index (CFI): Values consistently exceed .98.
• Tucker-Lewis Index (TLI): Values consistently exceed .97.
• Root Mean Square Error of Approximation (RMSEA): Estimates fall consistently below .05 (with 90% confidence intervals spanning .000 to .068).
• Standardized Root Mean Square Residual (SRMR): Values consistently measure below .025.

These psychometric properties confirm that the three manifest variables reliably indicate a unified psychological reaction.

Instrument / Measurement Tool

• Instrument Name: Object Eeriness Perception (OEP)
• Construct Assessed: Subjective perception of eeriness, creepiness, and unnaturalness elicited by anthropomorphic and robotic entities
• Administration Format: Self-administered paper-and-pencil or computer-assisted web interview (CAWI)
• Number of Items: 3 items
• Administration Time: Approximately 30 to 60 seconds
• Target Population: Consumers, adults, and experimental participants in human-robot interaction, marketing, and cognitive psychology research
• Response Format: 7-point Likert scale (1 = Strongly disagree, 7 = Strongly agree)
• Scoring Procedure: Scores are calculated as either the arithmetic mean or the unweighted sum of the three items:
  • Mean Scoring: Range 1.00 to 7.00 (Total Sum / 3)
  • Sum Scoring: Range 3 to 21
• Reverse-Scored Items: None; all three items are scored in a direct, positive direction.
• Interpretation: Higher scores reflect greater perceived eeriness, acute discomfort, and heightened uncanny valley response toward the evaluated object.

Permissions & Fee and Test Year

The Object Eeriness Perception scale was introduced in 2019 in the Journal of Marketing Research by Martin Mende, Maura L. Scott, Jenny van Doorn, Dhruv Grewal, and Ileana Shanks. As an academic measurement tool published in peer-reviewed literature, the scale is accessible for non-commercial scientific inquiry, university-based teaching, and scholarly research without direct licensing fees. Researchers employing the scale in empirical investigations are expected to cite the foundational publication (Mende et al., 2019) according to standard academic citation practices. For commercial applications, proprietary consumer testing, or inclusion in commercial software platforms, permissions should be secured through the copyright holder, the American Marketing Association (AMA) / SAGE Publications.

References

• Gray, H. M., Gray, K., & Wegner, D. M. (2007). Dimensions of mind perception. Science, 315(5812), 619–619. https://doi.org/10.1126/science.1134475
• Ho, C. C., & MacDorman, K. F. (2010). Revisiting the uncanny valley theory: Developing and validating an alternative to the Godspeed indices. Computers in Human Behavior, 26(6), 1508–1518. https://doi.org/10.1016/j.chb.2010.05.015
• Mende, M., Scott, M. L., van Doorn, J., Grewal, D., & Shanks, I. (2019). Service robots rising: How humanoid robots influence service experiences and elicit compensatory consumer responses. Journal of Marketing Research, 56(4), 535–556. https://doi.org/10.1177/0022243718822827
• Mori, M. (1970). The uncanny valley. Energy, 7(4), 33–35. Translated by MacDorman, K. F., & Kageki, N. (2012). IEEE Robotics & Automation Magazine, 19(2), 98–100. https://doi.org/10.1109/MRA.2012.2192811

Items of the Scale