How does a newborn infant, merely hours or days old, perceive the facial gesture of an adult caregiver and execute an anatomically matching movement using their own unseen face? The Active Intermodal Mapping (AIM) hypothesis offers one of the most transformative theoretical frameworks in developmental psychology to resolve this foundational mystery of social cognition. By proposing that human infants possess an innate capacity to translate exteroceptive visual information into kinesthetic motor commands across sensory modalities, AIM radically reshaped our understanding of the neonatal mind, the emergence of self-other equivalence, and the ontogeny of social interaction.
Active Intermodal Mapping (AIM)
1. Concise Definition
Active Intermodal Mapping (AIM) is a developmental and cognitive theory postulating that neonates possess an innate, supramodal representational system enabling them to match observed visual actions of another person with their own internal proprioceptive and motor states. Formulated primarily by developmental psychologists Andrew N. Meltzoff and M. Keith Moore, the model contends that facial imitation does not require months of reinforcement, mirror self-observation, or complex associative conditioning, but rather relies on an active matching-to-target process mediated through a common geometric and somatic code.
Within this framework, imitation operates via an internal comparator loop. The infant visually encodes an adult's action, translates that visual perception into an abstract supramodal representation, generates a corresponding motor output through proprioception, and iteratively refines its movements against the perceptual target. Far from being a rigid reflex or an innate releasing mechanism, AIM models early imitation as an intentional, exploratory, and goal-directed perceptual-motor achievement.
2. Etymology & Linguistic Origin
The term Active Intermodal Mapping is composed of three conceptually distinct lexical components originating from classical Latin and adapted through the vernacular of late-twentieth-century cognitive science:
- Active: Derived from the Latin activus (pertaining to action or driving force, from agere, meaning "to do, act, or set in motion"). In this construct, it emphasizes that neonatal imitation is neither an automated subcortical reflex nor a passive conditioned response, but an active, goal-directed searching process involving continuous sensory-motor modulation.
- Intermodal: Formed from the Latin prefix inter- ("between, among") and modus ("measure, manner, mode, or method"). In perceptual psychology, modality denotes a distinct sensory channel (such as vision, audition, or proprioception). "Intermodal" (often used interchangeably with "cross-modal") denotes psychological operations that bridge disparate sensory channels.
- Mapping: Etymologically rooted in the Medieval Latin mappa mundi ("sheet or cloth of the world"), adopted into mathematics and computational cognitive science to define a structural correspondence or systematic functional projection from elements of one domain (the source domain) to elements of another (the target domain).
The compound phrase was explicitly coined by Andrew N. Meltzoff and M. Keith Moore in their seminal publications during the late 1970s and 1980s to capture the psychological process of structural geometric projection linking the visual domain of observed behavior to the internal kinesthetic domain of enacted movement.
3. Pronunciation & Grammatical Form
Pronunciation: /ˈæk.tɪv ˌɪn.tərˈmoʊ.dəl ˈmæp.ɪŋ/
Acronym: AIM (pronounced as the single syllable noun /eɪm/).
Grammatical Form: Proper noun phrase / theoretical construct. It typically functions as a singular subject or direct object in psychological and neurocognitive discourse (e.g., "The Active Intermodal Mapping model posits an innate supramodal metric…" or "Empirical observations of neonate facial matching lend support to AIM").
4. Detailed Conceptual Explanation
The fundamental conceptual problem that Active Intermodal Mapping addresses is the classical "correspondence problem" of human imitation. When an individual attempts to imitate a manual gesture, such as raising an arm or tapping a finger, the individual can monitor both the model's hand and their own hand within the same perceptual channel: the visual field. However, when an infant imitates an unseen facial gesture—such as tongue protrusion, mouth opening, or lip pursing—the child cannot see their own face. The infant has never gazed into a mirror, has had no opportunity for visual feedback of their own countenance, and has not lived long enough to acquire associative links through operant conditioning. Yet, within hours or days of birth, human infants demonstrate systematic matching of these precise facial configurations.
To explain this striking behavioral phenomenon, the AIM model proposes that infant perception does not operate in isolated, modality-specific sensory silos that must be painstakingly integrated over months of physical experience. Instead, the sensory architecture of the human neonate is structured around a common "supramodal currency." Visual inputs detailing the spatial configuration and temporal dynamics of an adult's face are processed not merely as raw luminance gradients or retinal coordinates, but are directly mapped onto abstract, organ-specific spatial coordinates. This supramodal code allows visual perceptions and proprioceptive sensations of muscle contraction and joint movement to be translated into the same underlying neurocognitive language.
A central pillar of AIM is the concept of the internal comparator. According to Meltzoff and Moore, the infant does not execute a perfect, robotic replica on the very first motor attempt. Rather, observing a facial gesture activates an internal representation of the act that acts as an intentional target. The infant initiates a motor response using their intrinsic proprioceptive sense, evaluates the kinesthetic feedback of their own movement against the internal supramodal representation of the adult's demonstration, and makes self-corrective adjustments. This explains why neonatal imitation often unfolds gradually over several seconds: infants visually track the adult, halt extraneous body movements, align their facial musculature, and progressively sculpt their facial actions toward the modeled gesture through active kinesthetic trial and adjustment.
Furthermore, AIM asserts that the infant's mapping mechanism relies on an innate body scheme or proprioceptive map. Long before acquiring linguistic labels or conceptual self-awareness, the human infant has an intrinsic awareness of their own physical anatomy—a realization that they have a mouth, a tongue, and lips that are structurally homologous to the facial features of the conspecific sitting before them. This realization serves as the phylogenetic foundation for understanding other people: by recognizing structural equivalence between "my body" and "that body," the infant realizes that the other is fundamentally "like me."
5. Historical Development
Prior to the formulation of Active Intermodal Mapping, dominant twentieth-century developmental theories insisted that imitation of invisible gestures was impossible in early infancy. Jean Piaget, whose cognitive developmental stage theory set the benchmark for developmental psychology, argued in his 1951 work Play, Dreams and Imitation in Childhood that imitation develops gradually through six sensorimotor sub-stages. According to Piaget, infants in the first several months are limited to vocal contagion or simple reflexive circular reactions. He asserted that true imitation of gestures involving invisible body parts (the face) could not occur until approximately 8 to 12 months of age (Substage 4), because it required the infant to mentally correlate the tactile-kinesthetic sensations of their own face with visual perceptions of another's face—a complex cognitive milestone assumed to depend on extensive associative learning.
In 1977, developmental psychologists Andrew N. Meltzoff and M. Keith Moore shattered this consensus with their milestone study published in Science. Testing human infants between 12 and 21 days old, Meltzoff and Moore demonstrated that neonates systematically matched four distinct adult gestures: tongue protrusion, mouth opening, lip protrusion, and sequential finger movement. To preclude alternative explanations such as maternal conditioning or visual shaping, the experiments were conducted using rigorous blind-scoring methods and standardized stimulus presentations. Subsequent investigations in 1983 demonstrated that even neonates tested at an average of 36 hours old—and in some cases as young as 42 minutes post-birth—exhibited significant matching responses, definitively refuting the hypothesis that facial imitation requires months of visual-tactile associative pairing.
To provide a formal computational and theoretical architecture for these empirical findings, Meltzoff and Moore progressively developed the Active Intermodal Mapping hypothesis across several seminal theoretical papers in 1983, 1994, and 1997. They posited that neonatal imitation is fundamentally a perceptual-motor problem resolved through a dedicated neurocognitive architecture. Over the ensuing decades, AIM served as the conceptual launchpad for Meltzoff’s broader "Like-Me" framework of social cognitive development, which proposes that neonatal intermodal mapping is the foundational stepping stone for empathy, joint attention, intentionality understanding, and mature Theory of Mind.
6. Theoretical Foundations
Active Intermodal Mapping is situated at the intersection of perceptual theory, ecological psychology, motor control, and evolutionary cognitive science. It rests upon three major theoretical foundations:
First, AIM builds upon Eleanor Gibson’s and James J. Gibson’s ecological approach to perception, specifically the principle that perception and action are inextricably coupled. Rather than perceiving static sensory inputs that must be decoded by abstract logical operations, organisms directly perceive affordances and structural properties of the world. Meltzoff and Moore extended this principle by hypothesizing that human infants perceive human movement not as disjointed visual flashes, but as dynamic human transformations that directly afford motoric simulation within the observer's own somatic apparatus.
Second, the model is intimately connected to the ideomotor principle, first articulated historically by William James and expanded in modern cognitive science by Wolfgang Prinz and Bernhard Hommel under the Common Coding Theory. Common Coding posits that actions are represented, planned, and perceived using a shared, unified representational code. Perception and action planning share identical neural and cognitive representations. Within the AIM framework, the infant’s "supramodal system" is effectively an early-emerging, ontogenetically primitive common-coding matrix that links exteroception directly with motor execution.
Third, AIM anticipated the modern neurobiology of action observation and mirroring. When Giacomo Rizzolatti and colleagues discovered the mirror neuron system in macaque monkeys and documented analogous action-perception circuits in humans, developmental researchers highlighted AIM as the functional cognitive precursor to these neural systems. While the mirror neuron system provides a putative neuroanatomical substrate (involving premotor and parietal cortical circuits) that fires both when performing an action and when observing another perform that same action, AIM remains the premier cognitive-computational model explaining how such shared representations function functionally in earliest human ontogeny.
7. Key Components, Types & Dimensions
The architecture of the Active Intermodal Mapping mechanism consists of five structural components and functional dimensions operating in real time:
- Visual Perceptual Parsing: The infant visually isolates, parses, and identifies the adult model's primary expressive organs (e.g., lips, tongue, jaw) and categorizes the dynamic vector of movement (e.g., forward extension, downward displacement).
- Supramodal Representation Formulation: The visual input is converted from retina-specific coordinates into an abstract, modality-independent metric. This representational form preserves spatial configurations, movement direction, and temporal velocity without relying exclusively on visual imagery.
- Proprioceptive Proprioceptive Feedback Monitoring: The infant continuously samples real-time somatic, kinesthetic, and muscular feedback from their own facial and oral structures, maintaining an internal map of their present somatic state.
- Active Comparator Mechanism: A cognitive comparator matches the ongoing proprioceptive feedback against the stored supramodal target representation. Discrepancies between the performed action and the perceived target are registered as sensory-motor error signals.
- Iterative Motor Shaping and Correction: In response to the error signal, the motor cortex modifies descending efferent commands to recruit facial muscles (such as the genioglossus for tongue protrusion or the orbicularis oris for mouth movements), gradually shaping the motor output until it achieves behavioral parity with the adult demonstration.
8. Examples & Illustrative Cases
The clearest empirical manifestations of Active Intermodal Mapping appear in controlled laboratory demonstrations of neonatal behavioral matching. A standard experimental scenario illustrates how AIM operates in clinical and developmental research settings:
Consider a 2-day-old infant lying comfortably in an experimental bassinet inside a dimly lit room. An experimenter leans into the infant's direct line of sight, maintaining a neutral, resting facial expression to establish a baseline. The experimenter then slowly and deliberately protrudes their tongue three times over a 20-second interval, followed by a 20-second passive face period where the experimenter's mouth remains entirely closed and still. In response to this visual presentation, the infant does not instantly eject their tongue like an automated reflex. Instead, micro-analytic video recording reveals a distinct temporal sequence: the infant fixes their gaze on the adult's face, exhibits quieting of broad trunk and limb movements (motor deceleration), and begins subtle movements of the tongue inside the oral cavity.
Over the next 10 to 30 seconds, during the passive face interval, the infant slowly parts their lips and extends their tongue past the vermilion border. If the infant initially pushes the tongue toward the side of the cheek, the infant halts, retracts the tongue, and re-extends it directly forward along the midline. This "homing-in" behavior—transitioning from crude, exploratory movements to a targeted, precise structural match—demonstrates the AIM internal comparator in action. The infant uses real-time proprioceptive feedback to detect an anatomical error, self-corrects without visual feedback of their own face, and aligns their bodily configuration with the mental representation of the adult's gesture.
A second illustrative example involves the distinction between organ identification and action selection. When an adult demonstrates a novel oral movement—such as lip pursing—infants under observation often first activate the correct organ (e.g., by moving their mouth or tongue rather than waving their hands or kicking their legs), and subsequently shape that specific organ into the precise shape modeled (compressing the lips). AIM explains this two-step process: the infant first maps the visual input to the correct somatic organ system within their internal body scheme, and then fine-tunes the geometric parameters of the action.
9. Measurement & Assessment
Because neonates cannot verbally report their internal experiences or follow instructional tasks, measuring and validating Active Intermodal Mapping requires rigorous behavioral and neurophysiological observation methodologies:
- Micro-Analytic Video Coding: The gold standard in developmental psychology involves split-screen, high-definition video recording where the experimenter and infant are filmed independently. Trained observers, strictly blinded to the gesture being presented to the infant, score neonatal movements frame-by-frame (e.g., at 30 to 60 frames per second). Coders record the frequency, duration, and amplitude of specific motor categories, such as full tongue protrusions, partial tongue protrusions, mouth openings, and head movements.
- Infrared Facial Motion Tracking: Contemporary researchers employ high-speed infrared motion capture cameras and tiny reflective stickers or computational markerless tracking (using deep learning tools like DeepLabCut) applied to the infant's facial landmarks. This provides kinematic metrics, such as movement velocity, trajectory curvature, and acceleration profiles, allowing quantitative analysis of motor correction loops.
- Surface Electromyography (sEMG): By placing miniature non-invasive surface electrodes over specific facial muscle groups—such as the orbicularis oris (mouth closure/pursing) and the mentalis or suprahyoid muscle groups (associated with tongue and jaw movement)—investigators can record subtle sub-threshold muscular activations that occur when an infant views an adult gesture, even before an overt behavioral movement is visible.
- High-Density Electroencephalography (EEG): Neurodevelopmental researchers utilize high-density infant EEG caps to track sensorimotor mu rhythm desynchronization (typically in the 6–9 Hz band in human infants). Mu rhythm attenuation over central scalp locations during both the observation and execution of facial gestures provides empirical evidence of shared neural substrate activation, reflecting intermodal somatic mapping.
10. Applications & Practical Significance
The practical and translational significance of the Active Intermodal Mapping framework extends far beyond infancy research into clinical, therapeutic, educational, and computational domains:
In clinical neurodevelopmental pediatrics, AIM-derived assessments of neonatal imitation provide an early behavioral window into the integrity of the infant's central nervous system. Deficits or abnormalities in neonatal sensorimotor mapping and imitative responsivity can serve as early biomarkers for atypical neurological development, perinatal asphyxia, or subcortical-cortical communication disorders. Because AIM taps into basic cross-modal integration, early tracking of these behaviors aids in the prospective detection of neurodevelopmental vulnerabilities long before language or gross motor delays can be clinically evaluated.
In the study of autism spectrum conditions, AIM provides a vital baseline for understanding how early social-communicative pathways deviate. A core characteristic of autism is disrupted social reciprocity, motor imitation difficulties, and atypical intermodal integration. Researchers investigating early infant siblings of children with autism utilize AIM paradigms to observe whether deficits in cross-modal mapping, motor mirroring, or social contingent responding are present during the first months of life, laying the groundwork for early preemptive relational interventions.
In artificial intelligence, biomimetic robotics, and computational cognitive science, the architectural principles of AIM are routinely applied to solve the robotic correspondence problem. Developing autonomous humanoid robots that can acquire novel skills from human demonstrations requires the robot to map visual data from an external human actor onto its own internal kinematics and joint actuators. Engineers implement synthetic AIM algorithms featuring supramodal state-spaces, forward-inverse internal models, and predictive comparator loops to enable rapid robotic learning by imitation.
11. Research & Empirical Evidence
Over four decades of empirical investigation have evaluated the validity, reliability, and biological boundaries of Active Intermodal Mapping. Landmark studies by Meltzoff and Moore (1977, 1983, 1989) initially documented neonatal imitation across diverse sample cohorts, rigorously showing that infants did not imitate indiscriminate oral movements when shown control stimuli (such as flashing lights or opening hands), but responded specifically and categorically to matched human facial stimuli. Further supporting evidence emerged from independent laboratories globally, including findings by Field et al. (1982), who demonstrated that newborns could discriminate and imitate emotional facial expressions (happy, sad, and surprised faces).
Subsequent longitudinal studies demonstrated that early neonatal intermodal matching directly predicts subsequent cognitive and social competencies. Longitudinal work by Heimann (1989, 2001) revealed that infants who demonstrated robust neonatal imitation during the first weeks of life exhibited superior verbal abilities, sustained attention, and advanced communicative capacities at 12 and 18 months of age. These findings suggest that the internal mapping mechanism posited by AIM is not an isolated developmental anomaly, but an important foundational engine for longitudinal socio-cognitive growth.
Neuroimaging research has provided crucial complementary evidence. Studies examining infant cerebral hemodynamic and electrophysiological responses have confirmed that viewing human facial gestures activates motor regions in the infant brain. Saby, Meltzoff, and Marshall (2013) utilized electroencephalography to show that 14-month-old infants display localized sensorimotor cortex somatotopy: observing another person perform an action with their hand or foot activates the specific regions of the infant's sensorimotor cortex corresponding to the infant's own hand or foot. This demonstrates that human neural architecture contains somatotopically organized cross-modal linkages precisely matching the core postulates of AIM.
12. Cultural & Cross-Cultural Considerations
A core postulate of Active Intermodal Mapping is that the supramodal matching mechanism represents an innate, phylogenetically evolved property of the human brain rather than a product of cultural conditioning. To examine this claim, cross-cultural developmental researchers have evaluated neonatal imitation across diverse ecological settings, spanning industrialized Western metropolitan contexts, rural agricultural communities, and traditional indigenous foraging groups.
Studies conducted across various non-Western societies—such as rural cohorts in Nepal, indigenous groups in southern Africa, and communities across East Asia—have consistently documented neonatal matching of tongue protrusion and mouth opening when standardized testing methodologies are maintained. These cross-cultural findings substantiate the hypothesis that the basic sensory-motor translation architecture underlying AIM is universal among human populations, providing a shared bio-behavioral starting point for all human neonates regardless of cultural ecology or parenting paradigms.
However, cultural parenting styles exert a substantial downstream influence on how this innate mapping mechanism develops over ontogeny. In cultures that prioritize continuous face-to-face dyadic vocal and visual interaction (often characterized as "distal" parenting styles, common in middle-class Euro-American contexts), early facial imitation is often amplified and reinforced by parents who imitate their infants back. In contrast, in societies emphasizing "proximal" parenting styles (involving constant physical carrying, skin-to-skin touch, and co-sleeping, typical in many sub-Saharan African and traditional agricultural communities), cross-modal attunement often centers more heavily on tactile-vestibular synchronization and shared postural rhythms. Thus, while the underlying neurological architecture for Active Intermodal Mapping appears universal, the primary modalities and cultural pathways through which this intermodal competence is scaffolded diverge significantly across human societies.
13. Criticisms, Debates & Limitations
Despite its profound influence, the Active Intermodal Mapping hypothesis has remained a focal point of intense academic debate and methodological dispute in developmental science. Skeptics and alternative theoretical camps have challenged AIM on both empirical and theoretical grounds:
The primary empirical challenge concerns the replicability and scope of neonatal imitation. In a comprehensive critique, Anisfeld (1991, 1996) conducted systematic meta-analyses of published studies and argued that among all the gestures tested, only tongue protrusion reliably met empirical criteria for statistical significance across labs. Anisfeld contended that tongue protrusion might not represent true intentional, supramodal imitation at all, but rather an innate, non-specific exploratory reflex or oral arousal response: when an infant views an interesting or stimulating adult face, the infant exhibits oral excitement, which manifests simply as extending the tongue.
This challenge escalated dramatically with a major longitudinal study published by Oostenbroek et al. (2016) in Current Biology. Testing over 100 infants across four time points (1, 3, 6, and 9 weeks of age) on 11 distinct gestures, the authors reported finding no evidence of specific imitation, concluding that neonatal imitation is an experimental artifact and that infants are equally likely to produce tongue protrusions to non-matching adult gestures, objects, or emotional expressions. Meltzoff and colleagues robustly countered this critique, publishing extensive methodological rebuttals showing that the Oostenbroek et al. study suffered from profound procedural design flaws—including presenting adult gestures for brief intervals insufficient for neonatal processing, coding behaviors during ongoing model presentation rather than during the passive-face response window, and using highly unusual statistical categorization criteria that diluted matching effects.
On a theoretical level, radical constructivist and associative learning theorists, notably Cecilia Heyes (2001, 2016), have challenged the necessity of an innate supramodal mechanism. Through the Associative Sequence Learning (ASL) model, Heyes argues that the correspondence between visual perception and motor execution can be rapidly acquired through general learning mechanisms—specifically through early associative pairing. Heyes suggests that infants acquire these visual-motor links through optical reflections, through parents mimicking their infants back (providing immediate visual feedback of the infant's own motor outputs), and through general sensorimotor plasticity, rendering the hypothesis of an innate, specialized supramodal comparator redundant.
14. Related Terms & Distinctions
To ensure conceptual clarity, Active Intermodal Mapping must be rigorously differentiated from related psychological, neurological, and developmental constructs:
- Associative Sequence Learning (ASL): An empiricist alternative to AIM which argues that imitation relies on standard sensorimotor associative learning (Hebbian plasticity) acquired through experience, rather than an innate supramodal code.
- Mirror Neuron System (MNS): The specialized neuroanatomical circuit (primarily within the ventral premotor cortex, inferior parietal lobule, and superior temporal sulcus) that discharges both during action observation and execution. While AIM provides the functional and psychological computational model of cross-modal translation, the MNS is frequently conceptualized as a biological neural substrate underlying such actions.
- Innate Releasing Mechanism (IRM): An ethological concept referring to an automated, stereotyped subcortical reflex pattern triggered inevitably by an external sign stimulus. AIM explicitly distinguishes itself from IRMs by showing that neonatal imitation involves an active, error-correcting, goal-directed process with flexible temporal onset.
- Ideomotor Theory: A broader cognitive paradigm proposing that performing an action is mediated by activating mental representations of the perceptual effects of that action. AIM represents an ontogenetic application and extension of ideomotor principles to infant cross-modal facial matching.
- Delayed / Deferred Imitation: The capacity to reproduce a demonstrated behavior after a substantial temporal delay (hours, days, or weeks). Neonatal imitation under AIM can operate across brief delays (e.g., during the passive-face interval or after an intervening pacifier period), but deferred imitation of complex novel multi-step object tasks emerges later in infancy (around 6 to 9 months).
- Like-Me Hypothesis: Meltzoff's developmental framework which builds directly upon AIM, proposing that the innate ability to map others' actions onto the self allows the child to progressively deduce that other people possess minds, feelings, and intentions identical to their own.
15. Summary / Key Takeaways
Active Intermodal Mapping remains one of the foundational theories of developmental cognitive science. Its major empirical and theoretical takeaways can be summarized across five central points:
- The Core Premise: Newborn infants possess an innate neurocognitive ability to translate visual perceptions of another person's facial actions into their own internal kinesthetic and motor commands without requiring prior mirror experience or associative training.
- The Supramodal Metric: The sensory architecture of the neonate operates using an abstract, modality-independent code that allows seamless translation across the visual and proprioceptive sensory domains.
- The Comparator Process: Neonatal imitation is an active, goal-directed matching-to-target process. The infant compares proprioceptive feedback against an internal target representation, progressively adjusting their facial movements toward parity with the observed gesture.
- Theoretical Paradigm Shift: AIM decisively overturned the classical Piagetian view that deferred and facial imitation cannot occur before late infancy (8–12 months), fundamentally redefining the newborn as an active social and communicative agent from birth.
- Foundational for Social Cognition: Cross-modal somatic mapping serves as the ontogenetic foundation for the "Like-Me" framework, providing the critical structural bridge that allows human beings to develop empathy, joint attention, and Theory of Mind.
In conclusion, the Active Intermodal Mapping hypothesis serves as a vital bridge between perception, motor execution, and social development. By showing how the human newborn can translate an external visual scene into an internal proprioceptive act, AIM demonstrates that we are born inherently tuned to connect with other minds. Long before the emergence of spoken language, deliberate logical reasoning, or formal symbolic culture, human beings navigate the interpersonal world through an innate, embodied bridge that links self to other.
References
- Anisfeld, M. (1991). Neonatal imitation. Developmental Review, 11(1), 60–97. https://doi.org/10.1016/0273-2297(91)90003-Y
- Field, T. M., Woodson, R., Greenberg, R., & Cohen, D. (1982). Discrimination and imitation of facial expressions by term neonates. Science, 218(4568), 179–181. https://doi.org/10.1126/science.7123230
- Heyes, C. (2001). Causes and consequences of imitation. Trends in Cognitive Sciences, 5(6), 253–261. https://doi.org/10.1016/S1364-6613(00)01661-2
- Meltzoff, A. N., & Moore, M. K. (1977). Imitation of facial and manual gestures by human neonates. Science, 198(4312), 75–78. https://doi.org/10.1126/science.897687
- Meltzoff, A. N., & Moore, M. K. (1997). Explaining facial imitation: A theoretical model. Early Development and Parenting, 6(3‐4), 179–192. https://doi.org/10.1002/(SICI)1099-0917(199709/12)6:3/4<179::AID-EDP157>3.0.CO;2-R
- Oostenbroek, J., Suddendorf, T., Nielsen, M., Redshaw, J., Kennedy-Costantini, S., Davis, J., Clark, S., & Slaughter, V. (2016). Comprehensive longitudinal evaluation challenges the existence of neonatal imitation in humans. Current Biology, 26(10), 1334–1338. https://doi.org/10.1016/j.cub.2016.03.047
- Piaget, J. (1951). Play, dreams and imitation in childhood. Heinemann.