The Neonatal Imitation Experiment – Andrew Meltzoff and M. Keith Moore

The scientific understanding of human infancy underwent a seismic transformation in the final decades of the twentieth century. For over a century, classical philosophy, psychodynamic theory, and developmental psychology conceptualized the human neonate as an undifferentiated, reflexive organism lacking internal mental representations, intermodal perceptual coordination, and the capacity for deliberate social communication. The infant mind was classically framed through William James’s enduring aphorism of a “blooming, buzzing confusion”—an experiential chaos governed strictly by subcortical reflexes, primitive physiological appetites, and an inability to decouple subjective sensation from external objective reality. This prevailing doctrine reached its formal structural zenith in the genetic epistemology of Jean Piaget, who theorized that the capacity to imitate the facial actions of another person required the systematic construction of sensorimotor schemas across months of tactile and visual exploration, typically maturing only late in the first year of post-natal life.

This long-standing paradigm was fundamentally disrupted in October 1977, when developmental psychologists Andrew N. Meltzoff and M. Keith Moore published an empirical report in Science titled “Imitation of Facial and Manual Gestures by Human Neonates.” Through meticulously controlled laboratory investigations, Meltzoff and Moore demonstrated that human infants between 12 and 21 days of age—and later, neonates mere hours or minutes out of the womb—could systematically match the discrete facial gestures of an adult human model, including tongue protrusion and mouth opening. Because neonates cannot observe their own faces, the act of matching an adult’s facial movement cannot rely on direct visual guidance or conditioned operant reinforcement. The demonstration that newborns could bridge this sensory-motor divide suggested the presence of an active, innate cross-modal mapping mechanism, permanently upending behaviorist and constructivist dogmas.

The implications of this single empirical finding rippled across cognitive science, social neurobiology, philosophical psychology, and artificial intelligence. Meltzoff and Moore postulated the Active Intermodal Mapping (AIM) model, proposing that infants are born equipped with a supramodal representational space that translates visual input of human actions into corresponding motor execution programs. This early capacity for imitation provided the foundational architecture for the “Like Me” framework, illuminating how early motor matching serves as the bedrock for empathy, intentionality attribution, intersubjectivity, and Theory of Mind. This exhaustive exploration documents the historical backdrop, the exact experimental methodology, the cognitive architecture, the decades of empirical replication, the fierce meta-analytic and empirical debates, and the lasting neurobiological and philosophical legacy of Meltzoff and Moore’s seminal discovery.

1. Historical Context and Theoretical Backdrop of Infant Cognition

1.1 The Pre-1970s Consensus on Neonatal Capabilities

Prior to the late 1970s, the scientific consensus regarding the neonatal mind was defined by deep skepticism concerning the presence of organized cognitive structures, intentionality, or intermodal perceptual integration at birth. Influenced by British empiricism and late-nineteenth-century evolutionary biology, early psychologists viewed the newborn human as a physiologically immature organism whose experiential realm was fundamentally uncoordinated. In his foundational work The Principles of Psychology, William James articulated this dominant view by characterizing the infant’s first encounters with the world as a singular, overwhelming, and uncalibrated sensory inundation—a “great blooming, buzzing confusion.” Within this framework, sensory modalities such as vision, audition, olfaction, and kinesthesis operated as isolated, compartmentalized processing streams that possessed no innate baseline of structural communion.

Throughout the mid-twentieth century, classical and radical behaviorist frameworks, championed by figures such as John B. Watson and B.F. Skinner, reinforced this minimalist view of infant mental life. Behaviorism reduced neonatal behavior strictly to a repertoire of subcortical, primitive reflexes designed to facilitate survival, such as the rooting reflex, the palmar grasp reflex, the Moro startle response, and the sucking reflex. Any motor output emitted by a neonate in response to visual or acoustic stimulation was interpreted through the mechanistic lens of stimulus-response contingencies, thoroughly stripped of representational status, endogenous motivation, or communicative intentionality. The newborn was theoretically construed as a passive tabula rasa upon which environmental associations, operant conditioning, and continuous sensorimotor reinforcement gradually etched behavioral patterns.

Compounding this reductionist view was the widespread assumption that neonates lacked intermodal coordination. It was widely maintained that a newborn could not bridge the divide between distinct sensory modalities without extensive experiential training. The capacity to recognize that a tactile sensation corresponded to a visually observed object, or that an internally felt motor execution matched an external visual display, was considered an epistemological impossibility for the newborn brain. Cortical immaturity, incomplete myelination of the central nervous system, and a lack of synaptic density in associational cortices were regularly cited as neuroanatomical proof that the infant was biologically incapable of synthesizing sensory data into an integrated, cross-modal representation of the external physical or social environment.

1.2 Jean Piaget’s Stage Model of Sensorimotor Development

The pre-1970s theoretical architecture of developmental psychology was predominantly structured by the monumental constructivist framework of Jean Piaget. In his seminal texts, including The Origins of Intelligence in Children and Play, Dreams and Imitation in Childhood, Piaget established a rigorous stage model of sensorimotor intelligence that dictated the developmental trajectory of imitation. Central to Piagetian theory was the proposition that true facial imitation—which he termed the imitation of “invisible” gestures—was a complex cognitive achievement that could not possibly emerge until Stage IV of the sensorimotor period, typically occurring between 8 and 12 months of age.

Piaget drew a sharp theoretical distinction between “visible” gestures (actions the infant could observe themselves performing, such as opening and closing the hand or moving the feet) and “invisible” gestures (actions executed by bodily parts that infants cannot directly visually monitor, such as sticking out the tongue, furrowing the brow, or opening the mouth). To imitate a visible gesture, an infant could visually monitor their own limb, compare its kinematic trajectory to the visual model of the adult, and correct the motor error in real time. However, to match an invisible gesture, the child faced an intractable epistemological barrier: they could see the adult’s face, but they could never visually inspect their own facial movements. Piaget argued that this correspondence could only be established once the infant had constructed coordinated visual-tactile and tactile-kinesthetic schemas through prolonged, active manipulation of objects and observation of environmental feedback.

Furthermore, Piaget asserted that representational thought, mental symbols, and deferred imitation—the capacity to reenact a previously observed action after a significant temporal delay without continuous perceptual support—belonged exclusively to Stage VI of the sensorimotor period, emerging around 18 to 24 months of age. Prior to this, the infant was considered thoroughly solipsistic and egocentric, lacking a conceptual boundary between the self and the external world. To suggest that a newborn infant within the first weeks or hours of life could differentiate their own body from another’s body, and deliberately execute an unseen facial match, directly violated the foundational tenets of Piagetian genetic epistemology.

1.3 Emergence of Cognitive Revolution in Infancy Research

During the 1960s and early 1970s, the conceptual landscape of developmental psychology began to experience structural fissures, propelled by methodological breakthroughs that inaugurated the modern cognitive revolution in infancy research. Central to this paradigm shift was the pioneering work of Robert Fantz, who invented the visual preference and habituation techniques. Fantz demonstrated that infants, far from being trapped in an unorganized sensory blur, exhibited sophisticated visual discrimination, orienting preferentially toward patterned stimuli, complex geometric arrays, and specifically face-like schematic arrangements over uniform fields. Fantz’s work established that the visual cortex of the human newborn possessed innate architectural biases geared toward extracting salient information from the environment.

Concurrently, Scottish psychologist T.G.R. Bower launched a series of daring empirical investigations into early perceptual-motor competencies. Bower presented empirical evidence suggesting that human infants possessed early forms of perceptual constancy, depth perception, and cross-modal coordination far earlier than Piagetian orthodoxy permitted. Bower documented defensive motor responses, such as head retraction and the raising of hands, in infants presented with optical looming stimuli, indicating that babies could extract spatial and temporal information directly from the optic array. Although some of Bower’s specific experimental claims faced replication challenges, his broader empirical methodology established a vital precedent: infants possessed rich, pre-adapted evolutionary mechanisms that bypassed the need for arduous, trial-and-error sensorimotor learning.

This philosophical and empirical transition crystallized the theoretical trajectory of Andrew N. Meltzoff and M. Keith Moore. Operating during their doctoral and postdoctoral studies at the University of Oxford and later at the University of Washington, Meltzoff and Moore recognized that the prevailing constructivist and behaviorist dogmas suffered from an empirical blind spot. Researchers had consistently underestimated infant capabilities because the methodological tools deployed to test them had demanded sophisticated motor or linguistic competencies that infants simply did not possess. By devising an empirical framework centered on the infant’s natural, endogenous behavioral repertoire—specifically, the micro-movements of facial and manual anatomy—Meltzoff and Moore positioned themselves to test whether the human neonate arrived in the world pre-adapted for social connectivity.

2. The Landmark 1977 Science Study: Experimental Design and Methodology

2.1 Sample Characteristics and Inclusion Criteria

The landmark empirical study conducted by Andrew Meltzoff and M. Keith Moore, titled “Imitation of Facial and Manual Gestures by Human Neonates,” was published in Science in October 1977. The primary objective of the experimental design was to systematically test whether human infants could match adult facial and manual actions under conditions that rigorously controlled for alternative, lower-level explanations such as environmental conditioning, experimenter bias, or generalized motor arousal. The initial participant cohort comprised healthy, full-term human infants aged between 12 and 21 days post-partum, representing an ontogenetic window wherein infants had acquired minimal experience with human faces and had received virtually no operant reinforcement for facial mimicry.

The screening criteria applied by Meltzoff and Moore were exceptionally stringent to ensure that neurological integrity and physiological homeostasis were maintained across all subjects. Infants who had experienced significant perinatal complications, birth trauma, asphyxia, or whose mothers had received high doses of central nervous system depressants during labor were methodically excluded from the sample. Furthermore, the researchers utilized the observational criteria of the Brazelton Neonatal Behavioral Assessment Scale (NBAS) to continuously track and manage the infant’s arousal state. Testing was restricted entirely to infants categorized as being in State 4: an alert, calm, and visually attentive physiological state marked by minimal extraneous somatic movement and the absence of crying or persistent fussiness.

To insulate the experimental setting from confounding environmental distractions, the testing took place within a specialized, darkened sensory laboratory. The ambient auditory and visual parameters were tightly calibrated: the testing room was visually dampened with black drapes, eliminating non-social visual cues, high-contrast edges, and dynamic ambient illumination. The infant sat securely positioned in an ergonomic, padded experimental chair positioned directly opposite the adult experimenter at a calibrated focal distance of approximately 25 to 30 centimeters—an optical distance known to correspond precisely with the optimal neonatal visual accommodation range.

2.2 Stimulus Presentation and Structural Paradigm

The structural presentation paradigm deployed by Meltzoff and Moore was deliberately structured to differentiate genuine motor matching from random reflexive responses or continuous visual-motor shadowing. The experimental battery evaluated four distinct target gestures: two invisible facial actions (tongue protrusion and mouth opening), one complex facial configuration (lip protrusion), and one visible manual gesture (sequential finger movement, characterized by the sequential opening and closing of the hand). By including both oral and non-oral movements, the researchers constructed a critical comparative framework capable of decoupling gesture-specific imitation from general motor activation.

The timing protocol utilized a standardized “burst-pause” delivery sequence designed to respect the slower processing dynamics and prolonged neuromuscular reaction latencies characteristic of the neonatal nervous system. Rather than providing continuous, rapid stimulation, the adult experimenter presented the target gesture in repetitive, rhythmic bursts lasting approximately four seconds, followed by an observational pause of equal duration. During this pause, the experimenter assumed a completely neutral, motionless facial expression, colloquially known as a “passive face,” presenting no dynamic visual cues whatsoever. This alternating presentation ensured that the infant was afforded an unhurried, structurally protected window to encode the perceptual display, map the visual percept onto internal motor programs, and execute the physical response.

To further establish that the neonatal responses were not merely immediate, low-level reflexive resonant reflexes directly entrained to the sensory input, Meltzoff and Moore introduced an innovative pacifier habituation paradigm. In this protocol, a non-nutritive pacifier was gently inserted into the infant’s oral cavity prior to the presentation of the adult gesture. The infant watched the adult execute the dynamic gesture—such as repeated tongue protrusions—while the pacifier actively suppressed and physically blocked any immediate motor execution by the infant’s tongue. Following a delay period wherein the infant was exposed to the visual display without the physical ability to reciprocate, the adult resumed the passive, neutral facial baseline, and the pacifier was removed. The researchers then monitored the infant’s subsequent oral actions in the absence of any ongoing visual model.

2.3 Blinded Video Coding and Analytical Scoring Protocols

A critical methodological advancement of the 1977 study was its multi-angle, high-speed infrared videographic recording system, constructed to eliminate experimenter expectancy effects and establish uncompromising observational objectivity. Two separate television cameras were deployed: one high-resolution infrared camera focused tightly on the infant’s face and hands, capturing micro-movements of the labial, lingual, and mandibular musculature, while a second camera was trained on the adult experimenter. The video feeds were synchronized and recorded using split-screen technology, complete with an electronic time-code display running in the corner of the visual field.

Crucially, during the subsequent analytical phase, the video records were split, and the channel capturing the adult experimenter was completely masked and removed from the analytical stream. Independent, highly trained behavioral coders scored the infant’s behavioral responses in absolute blindness; they possessed no knowledge of which specific gesture the infant had observed in any given trial, the experimental hypothesis, or the chronological sequence of the experimental conditions. The coders operated purely as objective descriptive instruments, documenting discrete physical events according to explicit, pre-established operational criteria.

The scoring methodology captured both discrete event frequencies and cumulative behavioral durations over standardized scoring epochs. To assess whether the observed behaviors constituted statistically robust, selective imitation rather than non-specific baseline activity, the data were subjected to rigorous non-parametric and parametric statistical evaluations, including Wilcoxon matched-pairs signed-ranks tests. The analytical framework demanded that an infant show a statistically significant elevation of a target behavior (e.g., tongue protrusion) exclusively in response to the homologous adult demonstration (adult tongue protrusion), and not in response to an alternate, structurally distinct adult demonstration (e.g., adult mouth opening). This within-subject differential response criterion ensured that generalized somatic activation could not account for the statistical significance of the results.

3. Target Behaviors: Operational Definitions and Experimental Conditions

3.1 Facial Gestures: Tongue Protrusion and Mouth Opening

To eliminate subjective ambiguity in behavioral classification, Meltzoff and Moore formulated exacting operational definitions for each target gesture. For the primary facial gesture of tongue protrusion, the kinematic parameters required a forward, horizontal excursion of the lingual body such that the tip of the tongue clearly crossed the boundary defined by the inner vermilion border of the lower lip. The researchers established sub-classifications for “full protrusions,” wherein the lingual musculature extended distinctly beyond the lip margins, and “partial protrusions,” characterized by visible forward movements of the tongue across the alveolar ridge that terminated prior to crossing the labial boundary. Coders utilized slow-motion playback at reduced frame rates to ensure micro-kinematic tracking of the tongue’s trajectory.

The operational criteria for mouth opening required a clear vertical mandibular displacement accompanied by an unmistakable separation of the upper and lower lips. The researchers differentiated between minor resting mouth adjustments and deliberate, full-scale mouth opening by establishing an objective threshold: the vertical aperture of the mouth had to exceed twice the baseline resting distance between the lips, and the action had to occur in the absence of yawning, deep inspirations, or localized facial grimacing associated with gastric distress. The scoring protocols tracked the latency to onset, the peak duration of the maximal aperture, and the frequency of repetitive mouth-opening cycles.

The statistical juxtaposition of tongue protrusion and mouth opening served as the primary methodological bulwark against the hypothesis that neonatal responses were merely generalized arousal artifacts. Because both tongue protrusion and mouth opening are localized within the oral region, an undifferentiated excitation model would predict that presenting either visual stimulus would cause a generalized, elevated output across all oral motor effectors simultaneously. The empirical data thoroughly refuted this artifactual explanation: adult tongue protrusions selectively elevated infant tongue protrusions while depressing or maintaining baseline rates of mouth opening; conversely, adult mouth openings selectively amplified infant mouth openings without driving upward excursions in lingual activity.

3.2 Manual Gestures: Sequential Finger Movement

To assess whether the neonate’s capacity for behavioral matching was confined strictly to the oral-facial domain or reflected a distributed, bodily perceptual-motor matching system, Meltzoff and Moore introduced manual displays into the experimental battery. The adult manual gesture consisted of a fluid, sequential finger movement characterized by the continuous, rhythmic flexion and extension of the four digits relative to the palm, proceeding sequentially from the fifth digit (the little finger) through the index finger. This complex kinetic display was presented within the infant’s visual field while the adult maintained a totally motionless facial expression.

Kinematic tracking of neonatal manual responses required high-resolution recording of digit positions and hand configurations. Coding protocols required the independent scorers to differentiate between random, uncoordinated flailing of the upper extremities and precise, structured sequential extensions and flexions of the fingers. The operational definition of an imitative manual response demanded that the infant’s hand open and close in an organized, non-random sequence, moving through discrete finger configurations that mirrored the lateral progression demonstrated by the adult model, rather than executing a simple, reflexive palmar grasp or a violent startle reflex.

The manual matching data yielded theoretical significance by confirming that neonatal imitation was not merely an idiosyncratic oral phenomenon tied to primitive ingestive mechanisms. Infants demonstrated a selective and statistically significant increase in organized, sequential finger flexions specifically in response to the adult manual gesture, while exhibiting no such manual elevation when observing facial displays like tongue protrusion or mouth opening. This cross-modal dissociation provided strong empirical verification that the underlying mapping mechanism was anatomically generalized, coordinating distinct effector systems across both the facial and appendicular motor architectures.

3.3 Lip Protrusion and Novel Motor Configurations

The third facial condition evaluated in the 1977 investigations was lip protrusion—an intricate, novel motor configuration that required the infant to actively purse and push both the upper and lower vermilion borders outward, forming a distinct tubular, anterior projection of the lips without simultaneous jaw opening or tongue emergence. This gesture proved crucial because it lacked the high natural base rate of occurrence characteristic of spontaneous tongue movements or mouth adjustments, making it a critical test case against the claim that imitation was an artifact of high-frequency spontaneous motor noise.

The behavioral coding protocols systematically differentiated true lip protrusion from generalized sucking behaviors, rooting reflexes, and feeding-related oral automatisms. A genuine lip protrusion was operationally defined as an active, sustained forward protrusion of the orbicularis oris muscle structure, maintained independently of any tactile intra-oral stimulation and occurring without the rhythmic, peristaltic contractions of the cheeks that typify the nutritive sucking cycle. Coders documented both the precision of the labial morphology and the temporal stability of the held configuration.

The empirical observations of neonates attempting to replicate lip protrusion revealed a telling developmental phenomenon: infants rarely matched the complex morphology instantly in a singular, ballistic reflex. Instead, their motor responses displayed a clear trajectory of approximation. Neonates were observed initiating minor, tentative contractions of the perioral muscles, pausing, evaluating the internal sensation, and progressively adjusting the labial position across successive bursts until the configuration closely mirrored the adult demonstrator’s target posture. This gradual convergence provided empirical evidence of active proprioceptive monitoring and real-time error correction occurring within the neonatal neuromuscular system.

4. Active Intermodal Mapping (AIM): The Explanatory Cognitive Model

4.1 The Invisibility Problem and Proprioceptive Translation

The empirical demonstration that human neonates could selectively reproduce adult facial movements posed a profound epistemological puzzle that Meltzoff and Moore termed the “invisibility problem.” When an infant observes an adult protrude their tongue or open their mouth, the sensory input is entirely exteroceptive and visual: photons strike the infant’s retina, projecting an optic pattern of the adult’s moving anatomy onto the primary visual cortex. However, the infant cannot observe their own face. They possess no external visual feedback regarding their own tongue or lips, nor do they have access to a mirror to inspect whether their motor output visually matches the adult demonstrator.

To execute a facial match, the infant’s cognitive architecture must translate an exclusively visual perceptual input into an internally generated motor output monitored exclusively via kinesthesis and proprioception. Classical associative conditioning cannot account for this translation in a two-week-old infant, as the neonate has had neither the experiential history nor the systematic reinforcement schedule required to associate the visual sight of an adult’s moving mouth with the internal muscular feeling of moving their own. Direct visual guidance is physically impossible because the effector remains completely hidden from the infant’s own sight.

Meltzoff and Moore concluded that the infant must possess a proprioceptive feedback loop that functions as an internal somatic guidance system. Proprioception provides the brain with real-time, non-visual information regarding the spatial position, muscular tension, velocity, and trajectory of its own anatomical structures. To resolve the invisibility problem, the infant cognitive system must be capable of translating the externally observed visual coordinates of the other person’s face into internal proprioceptive motor commands, allowing the infant to “feel” the equivalence between their unseen facial actions and the seen gestures of the adult model.

4.2 Supramodal Representational Space

To explain how the neonatal brain bridges the gap between visual exteroception and proprioceptive motor control, Meltzoff and Moore formulated the theoretical model of Active Intermodal Mapping (AIM). The conceptual core of the AIM hypothesis is the postulation of an innate, supramodal representational space. Within this framework, sensory inputs are not permanently trapped within their modality-specific sensory silos (e.g., visual cortex versus somatosensory cortex); rather, perceptual information is automatically translated into an abstract, modality-independent representational code accessible across distinct sensory and motor domains.

When the neonate observes an adult facial movement, the visual information is parsed into an abstract structural description that captures the geometric relationships, relational transformations, and kinematic dynamics of the facial features (for example, the relative displacement of one facial part with respect to another, such as the tongue moving across the labial frame). Crucially, this abstract structural description is encoded in the exact same supramodal metric that the motor system uses to plan and execute muscle movements. Rather than requiring an intricate, learned cross-modal dictionary to translate sight into action, the sensory input and the motor plan share a common representational currency.

This formulation directly challenged the classical dual-system models of cognitive psychology, which maintained an unbridgeable architectural chasm between sensory input systems and motor output systems. The AIM model positioned the infant mind as unified at birth, possessing an innate neuro-computational architecture that naturally maps perceived actions onto executable motor programs. The supramodal representation acts as an internal comparator, enabling the infant’s central nervous system to establish an immediate, functional equivalence between the observed actions of another human being and the internally generated motor actions of the self.

4.3 Error Correction and Successive Approximations

Perhaps the most compelling empirical evidence supporting the AIM hypothesis over reflexive or automated models was the persistent observation of error correction and successive approximations in neonatal imitative responses. If neonatal imitation were merely an innate subcortical reflex—such as the patellar knee-jerk or the pupillary light reflex—the motor output would be expected to emerge in an all-or-nothing, stereotypic ballistic pattern characterized by a fixed latency and invariant kinematics. The behavioral data collected by Meltzoff and Moore revealed the opposite dynamic.

Upon observing an adult demonstrate a facial gesture, infants rarely produced an instantaneous, perfect anatomical match. Instead, the video records revealed a systematic, effortful process of motor exploration. For example, when exposed to a tongue protrusion display, an infant might first activate the mandibular muscles, slightly opening the mouth, pause, orient their gaze, subsequently execute a small intra-oral movement of the tongue, pause again, and finally execute a full-scale, horizontal lingual protrusion that matched the specific morphology of the adult model. The infant’s motor trajectory converged progressively on the target gesture across time through a process of homeostatic adjustment.

Within the theoretical architecture of AIM, this successive approximation reflects an active, goal-directed comparator process. The infant creates an internal supramodal representation of the visual target, initiates a motor command based on this representation, monitors their own physical execution via proprioceptive feedback, and compares the proprioceptive feedback against the supramodal target. Any registered discrepancy acts as a corrective error signal, prompting the infant to iteratively modify the motor output until the internal proprioceptive sensations match the structural representation of the visual model. This homing-in behavior establishes that neonatal imitation is an intentional, self-correcting cognitive achievement rather than a passive, involuntary physiological spasm.

5. The 1983 and 1989 Follow-Up Experiments: Expanding the Paradigm

5.1 Testing Newborns Under 72 Hours Old (Meltzoff and Moore, 1983)

Despite the methodological rigor of the 1977 study, theoretical critics continued to hypothesize that even the limited post-natal experience acquired within the first 12 to 21 days of life was sufficient for infants to undergo subtle operant conditioning. It was argued that mothers naturally engage in extensive face-to-face interactions, frequently mimicking their babies, which could theoretically allow infants to form associative links between their own internal motor feelings and the sight of human faces. To rigorously dismantle this post-natal socialization hypothesis, Meltzoff and Moore conducted a critical replication study published in Child Development in 1983, drastically reducing the chronological age of the experimental cohort.

The 1983 experimental cohort comprised 40 healthy neonates with a mean chronological age of only 32 hours post-birth. Most remarkably, the youngest infant in the sample was tested merely 42 minutes following delivery. These neonates had virtually zero opportunity to visually explore human facial expressions in structured interactive settings, nor had their mothers had the physical time or opportunity to condition their behavioral responses through differential reinforcement schedules. The infants were tested in a controlled hospital environment under strict laboratory protocols, fully replicating the blinded scoring procedures, infrared videography, and randomized presentation sequences of the original 1977 investigations.

The empirical findings were decisive: even infants tested within their first hours of life demonstrated a statistically significant, selective differential matching response to both tongue protrusion and mouth opening displays. The neonates responded selectively to the specific adult gesture they observed, matching the morphology of the demonstrator’s face with undeniable fidelity. By demonstrating that imitation is present essentially at the moment of birth, the 1983 study dealt a fatal blow to the hypothesis that facial matching was dependent upon post-natal associative conditioning, cementing the theoretical consensus that the capacity for intermodal social matching is an intrinsic, innate biological endowment of the human organism.

5.2 Memory and Deferred Imitation Over Temporal Delays (1989)

A primary theoretical critique raised by ethologists and radical behaviorists was that neonatal imitation, even if genuine, was merely an immediate perceptual resonant response—a form of behavioral entrainment or immediate echoic mimicry that depended entirely on the immediate, continuous presence of the sensory stimulus. To assess whether neonatal imitation was supported by internal representational storage and robust cognitive memory structures, Meltzoff and Moore designed an intricate temporal delay paradigm published in 1989 in Child Development.

In this experimental design, infants were exposed to an adult model demonstrating a specific facial gesture (e.g., tongue protrusion or mouth opening). However, during the display, the infants had a pacifier inserted in their mouths to prevent any immediate motor execution. Following the exposure phase, the adult left the room, the infant was returned to normal nursery care, and a prolonged temporal delay of up to 24 hours was instituted. The following day, the infant was placed back into the experimental environment and presented with the adult demonstrator maintaining a completely neutral, motionless passive face. No dynamic visual cues were provided during the test phase.

The results provided clear empirical proof of deferred imitation: despite the 24-hour temporal gap, the complete absence of ongoing visual modeling, and the alteration of contextual variables, the infants reliably produced the specific target gesture that had been visually presented to them the preceding day. They selectively elevated their output of tongue protrusions if they had observed tongue protrusions, or mouth openings if they had seen mouth openings. Because the visual model was entirely absent during the behavioral production, the response could not be attributed to an automatic perceptual resonance or an immediate sensory reflex. The infants had successfully encoded the visual display into memory, maintained that representational trace across a substantial delay without perceptual support, and subsequently retrieved that mental representation to direct their motor effectors.

5.3 Gesture Specificity and Novel Control Procedures

To further isolate the cognitive mechanisms driving imitation and eliminate any lingering concerns regarding non-specific motor elicitation, Meltzoff and Moore introduced a battery of sophisticated novel control conditions throughout their late-1980s and early-1990s research programs. A central innovation was the deployment of non-human, inanimate mechanical control stimuli designed to test whether the imitative response was uniquely social or merely an orientation reaction to moving physical objects.

In these control conditions, infants were presented with high-contrast, inanimate physical devices that mechanically mimicked the kinematics of human facial gestures. For instance, the researchers constructed a mechanical apparatus featuring an artificial, moving tongue-like paddle that extended and retracted horizontally from a neutral wooden oval, matching the precise spatial dimensions, speed, and rhythmic presentation cycles of the adult human tongue protrusion display. If the infant’s motor responses were driven purely by low-level physical properties—such as the visual perception of horizontal movement, dynamic contrast changes, or general mechanical motion—the artificial apparatus should have elicited the same frequency of infant tongue movements as the biological model.

The empirical results revealed an unmistakable biological selectivity: while the mechanical apparatus successfully elicited visual tracking and sustained perceptual attention, it failed completely to evoke the imitative matching response. Infants observed the moving mechanical parts with evident curiosity, but they did not systematically extend their own tongues in response. They reserved their imitative responses specifically and exclusively for biological agents—namely, actual human adults demonstrating natural facial actions. This empirical dissociation proved that the AIM mechanism was fundamentally embedded within a social perceptual matrix, specifically calibrated to identify, respond to, and communicate with fellow human conspecifics.

6. Methodological Controls and Innovations in the Experimental Paradigm

6.1 Elimination of Experimenter Bias and Expectancy Effects

The scientific credibility of Meltzoff and Moore’s empirical claims rested upon the development of experimental controls designed to eradicate experimenter expectancy effects, observational bias, and procedural artifacts. A cornerstone of this rigorous methodology was the structural decoupling of the stimulus presentation phase from the data acquisition and scoring pipelines through a dual-experimenter architecture. The experimenter who presented the facial gestures had no role in data processing, while the researchers who coded the video recordings were maintained in an absolute state of experimental blindness.

This blinding was accomplished through technical isolation of the recorded visual channels. The infrared camera feeds were electronically severed, ensuring that the primary coding monitors displayed only an extreme close-up view of the infant’s facial features, fully cropped from the chin to the forehead. Coders had no visual access to the adult experimenter’s actions, nor were they informed of the trial sequences, the hypotheses under evaluation, or the specific target gesture being demonstrated in any given epoch. The chronological order of video segments was frequently randomized prior to scoring, ensuring that raters could not deduce the progression of conditions via temporal cues.

To quantify the stability and objectivity of these observational protocols, inter-rater and intra-rater reliability assessments were systematically integrated into every published report. Video reels were independently evaluated by multiple naive scorers, and formal statistical metrics—including Cohen’s kappa coefficients for categorical events and intraclass correlation coefficients for temporal durations—were calculated. The resulting reliability metrics consistently exceeded accepted statistical thresholds (typically achieving kappa values greater than 0.85), proving that the behavioral events recorded were not subjective illusions conjured by biased observers, but highly reliable, quantifiable physiological phenomena.

6.2 Control for Generalized Arousal and Reflexive Triggers

A recurring methodological objection raised by skeptics was the generalized arousal hypothesis, which posited that infant tongue protrusion was merely an unselected, non-specific oral-motor discharge triggered by any compelling or stimulating environmental event. According to this critique, an infant presented with an engaging visual scene becomes physiologically excited, resulting in an indiscriminate elevation of oral motor activity that naive observers misinterpret as deliberate imitation.

Meltzoff and Moore confronted this hypothesis through systematic within-subject and between-subject control procedures. In their control trials, infants were presented with an array of non-facial, high-arousal sensory stimuli, including flashing strobe lights, burst sequences of white noise, and rapidly moving inanimate geometric patterns. While these intense sensory inputs successfully produced physiological signs of general arousal—such as heart-rate acceleration, limb wriggling, and wide-eyed visual tracking—they did not trigger significant elevations in infant tongue protrusion or mouth opening rates.

The definitive empirical refutation of the arousal critique came from the cross-target comparison method. In a typical balanced paradigm, infants were exposed to two structurally distinct facial displays: adult tongue protrusion and adult mouth opening. The arousal model dictates that if both displays are equally engaging, both should elicit identical generalized oral activity. However, the data revealed an unequivocal double dissociation:

• Adult Tongue Protrusion elicited high rates of Infant Tongue Protrusion, but zero statistically significant increase in Infant Mouth Opening.
• Adult Mouth Opening elicited high rates of Infant Mouth Opening, but zero statistically significant increase in Infant Tongue Protrusion.

This strict bidirectional specificity demonstrated that infant motor output was structurally matched to the precise morphology of the observed visual target, definitively ruling out generalized behavioral arousal as an explanatory mechanism.

6.3 Technological Advancements in Infant Motion Capture

The progression of Meltzoff and Moore’s empirical program over three decades closely paralleled the evolution of video capture and motion analysis technologies. In their early 1977 experiments, the researchers relied on pioneering analog videotape systems, utilizing dual-camera configurations and time-code generators that allowed micro-temporal analysis of behavioral reactions down to the single video field (one-sixtieth of a second). This fine-grained temporal resolution was vital for uncovering the subtle burst-pause dynamics and tracking the precise micro-latencies of neonatal motor recruitment.

As digital video technologies matured throughout the 1990s and early 2000s, Meltzoff and colleagues adapted standardized, computerized morphological coding schemes, most notably integrating principles from the Facial Action Coding System (FACS), adapted for infant anatomy as Baby FACS (developed by Harriet Oster and Paul Ekman). Baby FACS allowed researchers to categorize neonatal facial movements based entirely on anatomically distinct, observable muscular actions called Action Units (AUs). For instance, mouth opening was scored strictly through the activation of AU 25 (lips parting) and AU 26/27 (jaw dropping), while tongue protrusion was coded through AU 19 (tongue show). This objective muscular taxonomy eliminated descriptive subjectivity.

Furthermore, the integration of computerized frame-by-frame kinematic analysis enabled the quantification of geometric parameters, such as the exact mouth aperture ratio (the precise vertical displacement of the lips divided by horizontal mouth width) and the precise millimeter-scale distance of lingual protrusion relative to vermilion thresholds. These digital innovations transformed the study of neonatal imitation into an exact quantitative biomechanical science, dramatically reducing measurement error and establishing an empirical standard that made observational data replicable across independent laboratories worldwide.

7. The Piagetian Paradigm Shift: Overturning Sensorimotor Orthodoxy

7.1 Dismantling the Constructionist Timeline of Early Infancy

The empirical verification of neonatal imitation exerted an immediate, disruptive impact on twentieth-century developmental psychology by dismantling the constructionist developmental timeline formulated by Jean Piaget. Piaget’s stage theory of sensorimotor intelligence was predicated on the foundational assumption that cognitive capacities develop sequentially out of simple motor reflexes through a prolonged process of assimilation and accommodation, mediated by active environmental experience.

Piaget’s theoretical architecture mandated that the imitation of invisible gestures could not emerge until Stage IV (8 to 12 months), because the infant supposedly required nearly a year of sensorimotor experimentation to forge the cross-modal associations connecting visual percepts of others with internal kinesthetic sensations of the self. By establishing that neonates just hours out of the womb—and infants under three weeks of age—could reliably match unseen facial gestures, Meltzoff and Moore compressed this theoretical timeline by almost a full year. The assertion that an infant requires eight to twelve months of active learning to cross the bridge between vision and proprioception was empirically shattered.

This empirical falsification compelled developmental psychologists to critically re-evaluate Piaget’s observational methodology. Piaget’s conclusions had been drawn largely from informal, naturalistic diary observations of his own three children (Laurent, Lucienne, and Jacqueline). Because Piaget lacked high-speed infrared videography, split-screen technology, blind coding protocols, and the burst-pause presentation framework, subtle and slow-latency neonatal matching behaviors went entirely unnoticed. The absence of an experimental apparatus calibrated to the infant’s slow processing latencies had caused Piaget to confuse the infant’s performance limitations with a total absence of underlying cognitive competence.

7.2 Innate Knowledge Structures versus Radical Constructivism

The collapse of the Piagetian timeline catalyzed a major renaissance of nativist and neo-nativist paradigms in developmental cognitive science. Meltzoff and Moore’s empirical findings provided definitive empirical evidence that the human infant does not enter the world as a cognitive clean slate, but rather arrives equipped with innate, highly structured representational mechanisms designed to organize perceptual experience from the initial moments of life.

This theoretical stance aligned with the broader “core knowledge” frameworks later popularized by cognitive scientists such as Elizabeth Spelke and Renée Baillargeon, who demonstrated early innate competencies in physical reasoning, object permanence, and numerical estimation. However, Meltzoff advanced a nuanced variant of nativism termed “starting-state nativism.” Unlike radical nativists who argued that adult cognitive capacities are fully pre-formed and merely trigger automatically over maturation, Meltzoff proposed that evolution provides the newborn with a set of rich, specialized “starting states”—innate biological scaffolding that does not replace learning, but rather jump-starts and bootstraps subsequent experiential and social development.

Neonatal imitation, therefore, represented the pre-adapted cognitive bridge that allows the infant to immediately interface with their social environment. The capacity to translate social perception into self-action is an evolutionary adaptation that ensures the human infant is inherently pre-tuned for relational interaction. By establishing that cross-modal representational space is available at birth, the AIM model demonstrated that the human mind is inherently relational, biologically wired from its evolutionary origins to link the actions of others to the internal experiences of the self.

7.3 The ‘Like Me’ Framework: From Imitation to Social Cognition

To synthesize the developmental trajectory linking neonatal imitation to complex social intelligence, Andrew Meltzoff formulated the celebrated ‘Like Me’ framework. This theoretical model outlines how the simple capacity for bodily imitation observed in the newborn serves as the developmental engine that ultimately drives the emergence of empathy, mentalizing, and adult Theory of Mind. The model posits a progressive, three-step developmental progression:

• Step 1: Structural Equivalence at Birth. The neonate recognizes that other human beings are structural, bodily equivalents to the self. When the infant visually perceives an adult moving an anatomical part, the AIM mechanism matches that visual pattern onto their own homologous bodily effector. The infant experiences an immediate, non-verbal realization: “That entity looks like me, and moves like me.”
• Step 2: Subjective First-Person Experience. As the infant matures across the first months of life, they continuously experience internal subjective states—intentions, desires, emotions, and attentional orientations—that reliably accompany their own self-generated bodily actions. The child learns that choosing to reach for an object is accompanied by an internal desire, or that crinkling their face is accompanied by an internal emotional state.
• Step 3: Attributing Internal States to Others. Because the child has already established that others are “Like Me” in their physical acts (Step 1), and because the child knows that their own bodily acts are driven by internal mental states (Step 2), the child uses this self-knowledge as an analogical projection tool to infer the hidden mental states of others. When they see another person reach for an object, frown, or look toward a spatial target, they infer that the other person is experiencing internal intentions, sadness, or visual interest similar to their own.

The ‘Like Me’ framework positions neonatal imitation as the foundational bedrock of human intersubjectivity. Rather than developing Theory of Mind through abstract logical deduction or late-emerging verbal instruction, children build their comprehension of other minds from the ground up, rooted in the visceral, embodied experience of bodily matching that was present at birth.

8. Primary Counterarguments, Skepticism, and Replication Challenges

8.1 The Innate Releasing Mechanism (IRM) Hypothesis

Despite the acclaim surrounding Meltzoff and Moore’s publications, their theoretical interpretation was challenged by ethologists and evolutionary biologists who sought to explain the phenomenon without invoking complex cognitive concepts like “supramodal representations” or “intentional matching.” Foremost among these alternative accounts was the Innate Releasing Mechanism (IRM) hypothesis, which rooted its explanation in classical Lorenzian ethology.

Proponents of the IRM hypothesis argued that human tongue protrusion in response to an adult face was an evolutionary Fixed Action Pattern (FAP)—a primitive, subcortical reflex released automatically by a specific, high-contrast visual sign stimulus. Within this paradigm, an adult face sticking out its tongue represents an unconditioned optical trigger that automatically disinhibits a subcortical motor circuit in the infant’s brainstem, causing an involuntary thrust of the lingual musculature. Some researchers hypothesized that this was an evolutionary remnant of an ancient ingestive or suckling adaptation, designed to clear the airway or prepare the oral cavity to latch onto the maternal nipple.

Meltzoff and Moore vigorously refuted the IRM model by emphasizing the profound morphological and temporal flexibility documented in their experiments. If tongue protrusion were an involuntary FAP, it should fire predictably in an all-or-nothing fashion immediately upon presentation of the releaser, and it should fail to explain why neonates selectively imitate multiple diverse gestures, including mouth opening, lip protrusion, and sequential finger movements. The existence of deferred imitation across 24 hours, the gradual trajectory of error-correcting successive approximations, and the failure of mechanical devices with identical kinematics to release the response comprehensively demonstrated that neonatal imitation could not be reduced to an ethological reflex.

8.2 Moses Anisfeld’s Meta-Analyses and Critical Re-evaluations

The most persistent and statistically rigorous critique of the neonatal imitation literature was mounted by developmental psychologist Moses Anisfeld in two extensive meta-analytic reviews published in 1991 and 1996. Anisfeld conducted exhaustive quantitative syntheses of dozens of published empirical replication attempts across multiple independent laboratories, arriving at a profoundly skeptical conclusion that divided the developmental psychology community.

Anisfeld claimed that when all published data were aggregated, only a single gesture demonstrated a statistically reliable, repeatable elevation across diverse laboratories: tongue protrusion. The statistical evidence for the neonatal imitation of mouth opening, lip protrusion, finger movement, head turning, and emotional expressions was, according to Anisfeld’s calculations, statistically weak, erratic, and largely indistinguishable from experimental noise. Furthermore, Anisfeld argued that the apparent elevation of tongue protrusion was not a selective imitative match, but rather a non-specific behavioral manifestation of generalized neonatal excitement.

Meltzoff and Moore countered Anisfeld’s meta-analyses by demonstrating that his conclusions were skewed by the uncritical inclusion of replication studies that had committed serious methodological departures from the original 1977 and 1983 protocols. Many of the failed replications aggregated by Anisfeld had utilized extremely brief exposure times, had failed to employ the burst-pause delivery technique, had scored behaviors while the adult was still rapidly moving their face (thereby producing sensory interference), or had neglected to carefully control the infant’s physiological arousal state using Brazelton criteria. Meltzoff argued that when experimental protocols strictly adhered to the original procedural methodology, the imitative matching of both tongue protrusion and mouth opening remained an entirely robust, replicable scientific fact.

8.3 The Sucking and Arousal Hypothesis (Susan Jones)

Further empirical skepticism emerged from a series of studies conducted by developmental psychologist Susan S. Jones during the 1990s and early 2000s. Jones proposed an alternative mechanism known as the Sucking and Arousal Hypothesis, which challenged the core assumption that infant tongue movements bore any intentional, representational relationship to the adult’s face.

In an influential series of experiments, Jones demonstrated that human infants would dramatically increase their rates of tongue protrusion when presented with totally non-social, inanimate visual displays, provided those displays were sufficiently interesting and visually stimulating. When infants were exposed to bright flashing perimeter lights, brightly colored spinning toys, or music-accompanied geometric patterns, their tongue protrusion rates elevated to levels matching or exceeding those elicited by adult human faces. Jones posited that tongue protrusion is an exploratory, physiological component of the orienting reflex—a motor accompaniment of focused visual attention and state transitions tied to the suppression of nutritive sucking.

Jones asserted that when an infant is shown an adult face sticking out its tongue, the infant does not match the gesture through an abstract supramodal comparator; rather, the dynamic, protruding pink lingual body acts as a uniquely captivating visual target that grabs the infant’s visual attention, precipitating an exploratory arousal burst that naturally manifests as oral tongue movement. While this critique underscored the necessity of utilizing non-social control stimuli and analyzing multiple motor effectors, it struggled to explain why infants in Meltzoff and Moore’s studies systematically differentiated mouth opening from tongue protrusion, or why manual sequential finger movements were selectively matched by the infant’s own hands.

9. The 2016 Oostenbroek et al. Longitudinal Study and Contemporary Debates

9.1 Design and Findings of the Comprehensive Longitudinal Study

The academic debate surrounding neonatal imitation reached a dramatic modern climax in 2016 with the publication of a massive, longitudinal study in Current Biology by Janine Oostenbroek, Virginia Slaughter, and their colleagues. Titled “Comprehensive Longitudinal Study Challenges the Existence of Neonatal Imitation,” the investigation was widely billed as the definitive, fully powered empirical test of the neonatal imitation phenomenon, tracking a substantial cohort of 106 human infants longitudinally across four separate developmental time points: 1 week, 3 weeks, 6 weeks, and 9 weeks of age.

Oostenbroek and colleagues tested an extensive battery of 11 distinct adult gestures—including facial expressions (tongue protrusion, mouth opening, happy face, sad face), manual displays (finger opening, index finger pointing), and vocal/auditory sounds (ee-sound, click sound)—using an experimental protocol designed to evaluate the presence of behavioral matching across early infancy. The researchers deployed sophisticated statistical linear mixed-effects modeling to detect whether infants systematically matched the specific target gestures shown by the adult demonstrator above baseline control levels.

The findings of the Oostenbroek et al. study were stark: the researchers found no evidence whatsoever of selective neonatal imitation. Infants produced tongue movements, mouth openings, and vocalizations, but their motor behaviors were structurally unrelated to the specific gestures displayed by the human model. For example, infants were just as likely to stick out their tongues in response to an adult opening their mouth, expressing sadness, or pointing a finger as they were to an adult protruding their tongue. The authors concluded that neonatal imitation is a scientific artifact that does not exist in nature; instead, they claimed that true imitation is a learned social behavior that develops gradually over the course of the first year of life as a product of social scaffolding and maternal imitation.

9.2 Meltzoff et al.’s Detailed Rebuttal and Methodological Critiques

The publication of the Oostenbroek study triggered an immediate, forceful rebuttal from Andrew Meltzoff, M. Keith Moore, and an international coalition of developmental cognitive neuroscientists. In a comprehensive critique published in Developmental Science (2018), Meltzoff and colleagues demonstrated that despite its impressive sample size, the Oostenbroek study was compromised by fatal methodological deviations from established protocols, which fatally undermined its empirical validity.

Meltzoff and colleagues documented that Oostenbroek et al. had committed an acute protocol failure regarding stimulus exposure time and testing design:

• Stimulus Saturation and Fatigue: Oostenbroek et al. presented 11 disparate adult behaviors sequentially to 1-to-9-week-old neonates within a single, continuous testing session. Meltzoff pointed out that subjecting fragile, easily exhausted neonates to a battery of 11 consecutive, highly stimulating demonstrations induces rapid sensory fatigue, habituation, and behavioral fussiness, thoroughly violating the Brazelton State 4 requirement necessary for infant cognitive engagement.
• Inadequate Exposure and Scoring Windows: The experimental timing deployed by Oostenbroek severely curtailed the infant’s processing windows. Infants were provided brief exposure periods and immediate scoring windows that completely failed to accommodate the slow neuromuscular reaction latencies characteristic of neonatal intermodal mapping.
• Methodological Baseline Artifacts: The statistical design compared the frequency of infant gestures during demonstration against an inappropriate composite baseline constructed from the pooled responses to all other 10 non-matching stimuli, diluting statistical power and mathematically masking genuine matching effects.

Meltzoff emphasized the core scientific maxim of procedural fidelity: in developmental psychology, altering the critical temporal pacing, sensory parameters, and arousal conditions of a delicate perceptual paradigm does not constitute an objective replication; rather, it introduces disruptive variables that actively suppress the expression of the target phenomenon.

9.3 Current State of the Consensus in Developmental Psychology

Today, the developmental psychology and cognitive science communities remain actively divided regarding the ontological status of neonatal imitation, reflecting a fundamental philosophical schism between nativist and constructivist worldviews. One faction of researchers, heavily aligned with modern constructivism and associative learning theory, maintains that human infants learn to imitate exclusively through social interaction during post-natal ontogeny. Within this framework, early infant matching is viewed as an illusion created by high baseline rates of spontaneous oral movements, non-specific arousal reactions, and observational bias on the part of parents and researchers.

The opposing faction, supported by nativist frameworks, evolutionary developmental biologists, and cognitive neuroscientists, maintains that the foundational capacity for cross-modal social matching is genuine, hardwired, and reproducible when investigated under methodologically valid laboratory conditions. They point to the vast corpus of successfully replicated independent studies published across five decades as incontrovertible proof that neonates possess an innate, starting-state intermodal mechanism.

Increasingly, the modern consensus is moving beyond the binary question of whether neonatal imitation categorically “exists” or “does not exist,” shifting toward a nuanced, conditional paradigm that investigates the specific environmental, neurobiological, and individual parameters that moderate early social matching. Researchers now recognize that early imitation is not an automatic, invariant mechanical response, but a delicate, highly context-dependent social dialogue that is extraordinarily sensitive to the infant’s arousal state, maternal bonding history, testing environment, and individual differences in neurodevelopmental maturation.

10. Neurobiological Correlates: Mirror Neurons and Neural Resonance

10.1 The Discovery of the Mirror Neuron System (MNS)

In the early 1990s, the theoretical plausibility of Meltzoff and Moore’s Active Intermodal Mapping model received dramatic, unanticipated empirical corroboration from cognitive neuroscience through the discovery of the Mirror Neuron System (MNS). A research team led by Giacomo Rizzolatti, Leonardo Fogassi, Vittorio Gallese, and Luciano Fadiga at the University of Parma, utilizing single-cell microelectrode recordings in macaque monkeys (Macaca nemestrina), discovered a unique class of visuomotor neurons in the ventral premotor cortex (area F5) and the inferior parietal lobule.

These specialized neurons possessed a remarkable property: they fired both when the monkey executed a specific, goal-directed motor act (such as grasping a peanut with a precision grip) and when the monkey passively observed a human experimenter execute an identical motor act. The mirror neuron system directly linked visual perception with motor execution at the cellular level. Subsequent neuroimaging and magnetoencephalographic investigations confirmed the existence of an equivalent, highly developed frontoparietal mirror neuron network in the human brain, encompassing the inferior frontal gyrus (Broca’s area), the premotor cortex, and the rostral inferior parietal cortex.

The discovery of mirror neurons provided the long-sought biological substrate for the theoretical architecture postulated by Meltzoff and Moore decades earlier. The mirror neuron network provided a concrete neural mechanism capable of executing supramodal translation. By directly mapping observed visual input onto the homologous motor representations stored in premotor and parietal cortices, the mirror system provided a functional explanation for how the brain naturally bypasses the “invisibility problem,” establishing a common neural currency connecting self and other.

10.2 Infant EEG, Mu Rhythm Desynchronization, and Neural Tracking

To establish whether this mirror resonance system operates during human early infancy, developmental cognitive neuroscientists led by Peter J. Marshall and Andrew N. Meltzoff deployed infant electroencephalography (EEG) to examine the neural dynamics of the sensorimotor cortex in very young infants. Their research focused specifically on the suppression or desynchronization of the infant sensorimotor alpha rhythm, conventionally termed the mu rhythm.

In human infants, the mu rhythm reflects synchronized electrophysiological oscillations centered over the central rolandic scalp regions, overlying the primary motor and somatosensory cortices. When an infant is at rest, this rhythm demonstrates high amplitude and synchronized firing. However, when the infant executes an active bodily movement—such as moving their own hand or reaching with their arm—the mu rhythm drops sharply in amplitude, exhibiting pronounced event-related desynchronization (ERD), reflecting active cortical recruitment of the sensorimotor networks.

Marshall and Meltzoff demonstrated that when young human infants passively observe an adult execute a manual or facial action, their central sensorimotor mu rhythm exhibits the exact same pattern of significant desynchronization observed during physical execution. Most remarkably, high-density EEG somatotopic mapping revealed that this mirror activation is anatomically localized: observing an adult move their hand activates the hand representation of the infant’s motor strip, whereas observing an adult move their foot or face selectively desynchronizes the homologous foot or face regions of the infant’s sensorimotor cortex. This neurophysiological evidence confirms that the infant brain naturally mirrors the actions of others, providing objective, non-behavioral proof of early sensorimotor coupling.

10.3 Comparative Perspectives: Primate Neonatal Imitation

The evolutionary roots of early social imitation were further illuminated by groundbreaking comparative primatology investigations that expanded Meltzoff and Moore’s paradigm across species boundaries. In a landmark 2006 study published in PLOS Biology, Pier Francesco Ferrari, Leonardo Fogassi, Giacomo Rizzolatti, and their colleagues demonstrated the presence of robust neonatal imitation in rhesus macaque monkeys (Macaca mulatta).

Ferrari and his team tested infant macaques within the first week of life, demonstrating that baby monkeys selectively matched the facial gestures of human and conspecific models, reliably replicating tongue protrusion and lip-smacking—a fundamental rhythmic social affiliative gesture in the macaque communicative repertoire. Concurrently, pioneering research conducted by Masako Myowa-Yamakoshi and colleagues at the Primate Research Institute of Kyoto University revealed parallel neonatal imitation capacities in infant chimpanzees (Pan troglodytes), who reliably matched tongue protrusion and mouth opening displays during their initial weeks post-partum.

These comparative evolutionary findings revealed an intriguing phylogenetic phenomenon: in non-human primates, neonatal imitation is a transient, ephemeral evolutionary adaptation. Infant macaques and chimpanzees exhibit rich facial matching during their first days of life, but the behavior spontaneously wanes and disappears entirely within a matter of weeks, typically terminating by the end of the first month. In contrast, in the human lineage, neonatal imitation does not disappear; rather, it is assimilated and structurally transformed into the rich, lifelong, flexible socio-cognitive system that underpins cultural learning, communicative language, and human cultural transmission.

11. Intersubjectivity and the Social Dimensions of Neonatal Imitation

11.1 Primary Intersubjectivity: Colwyn Trevarthen’s Formulations

While Andrew Meltzoff prioritized the cognitive and representational dimensions of neonatal imitation, Scottish psychobiologist Colwyn Trevarthen formulated an intrinsically relational, dialogic interpretation of early infant behavioral matching. Trevarthen positioned neonatal imitation as the primary empirical manifestation of what he termed primary intersubjectivity—an innate, human-specific communicative motive that drives infants to seek mutual social resonance, affective sharing, and rhythmic attunement with other human beings from birth.

Trevarthen argued that neonatal imitation should not be conceived merely as a cold, isolated motor gymnastics exercise wherein an infant mechanistically calculates physical spatial coordinates. Rather, it represents the foundational prototype of human conversation—a “proto-conversation.” In naturalistic interactions, the infant’s imitative matching is saturated with emotional prosody, temporal musicality, and relational reciprocity. The infant does not simply copy an adult’s gesture; they engage in an active, reciprocal dance of communicative sharing, alternating roles between active expresser and attentive receiver.

Within this relational paradigm, the Active Intermodal Mapping mechanism functions as the physiological and neurobiological medium through which two separate subjectivities establish mutual communicative contact. Neonatal imitation represents the infant’s declaration that they are not merely an isolated sensory processor, but a communicative partner participating in a shared intersubjective space. By synchronizing their bodily actions with those of their caregivers, neonates establish early affective attunement, cultivating the emotional bonds that form the bedrock of human socialization and relational development.

11.2 Daniel Stern’s Interpersonal World of the Infant

The psychiatric and psychoanalytic significance of early imitation was profoundly expanded by developmental psychoanalyst Daniel Stern in his transformative work The Interpersonal World of the Infant. Stern utilized Meltzoff and Moore’s empirical findings to directly dismantle traditional Freudian and Mahlerian psychoanalytic theories that characterized early infancy as a phase of “normal autism” or symbiotic undifferentiation.

Traditional psychoanalysis asserted that the newborn infant possesses no boundary between the self and the mother, living in a narcissistic, oceanic delusion of absolute oneness. Stern demonstrated that the reality documented by neonatal imitation experiments proved the exact opposite: to imitate another person’s face, the infant must possess an emergent sense of self and an emergent sense of an “other.” The infant must recognize that the visual target exists “out there,” located on another independent bodily agent, and that they must deploy their own, distinct bodily effectors to match that external display.

Stern built upon this to develop his foundational theory of affect attunement—the cross-modal matching of internal feeling states between caregiver and child. For Stern, affect attunement is the direct psychological descendant of neonatal imitation. Just as the neonate maps an external visual form onto an internal proprioceptive motor program, the growing infant and their caregiver map dynamic emotional intensities, rhythmic temporal contours, and feeling states across different sensory modalities. A mother matches her baby’s joyful vocal pitch with a rhythmic bobbing of her head, or matches the baby’s excited arm movement with an affectionate vocal crescendo. This cross-modal emotional communion provides the structural scaffolding for secure attachment, emotional self-regulation, and interpersonal intimacy.

11.3 Identity Tracking and Person Recognition

In subsequent empirical investigations conducted throughout the 1990s, Meltzoff and Moore uncovered a crucial functional application of neonatal imitation: its role in person recognition and identity tracking. Infants in natural environments are confronted with a challenging perceptual problem: people leave the room, change their clothes, alter their hairstyles, and transform their emotional expressions. How does the infant determine whether a returning person is the same individual they interacted with previously?

Meltzoff and Moore designed an experiment wherein an adult sat before an infant and executed a specific facial gesture (e.g., sticking out their tongue). The adult then became completely motionless, adopting a neutral, passive face. A few moments later, the first adult silently stepped aside and was replaced by a completely new adult, who also sat with an identical, neutral passive face. The infant observed the new person intently, looking from face to face. In this ambiguous social scenario, what did the infant do?

The empirical records revealed that the infant would look directly at the new adult and systematically execute the exact gesture that the first adult had demonstrated—thrusting their tongue forward. Meltzoff and Moore discovered that the infant was utilizing imitation as an active behavioral probe—a communicative test designed to ask the existential social question: “Are you the person who made that face?” The infant used imitation as an interrogative social tool to verify personal identity across spatial and temporal interruptions. Imitation was revealed to be a powerful vehicle for early social epistemology, allowing the young human mind to track individual persons and establish continuous relational histories with specific caregivers.

12. Contemporary Legacy and Enduring Impact on Developmental Science

12.1 Clinical Applications: Autism Spectrum and Developmental Disorders

The insights generated by Meltzoff and Moore’s research program have provided foundational frameworks for clinical developmental psychopathology, particularly in unraveling the etiology of and designing early screening protocols for Autism Spectrum Disorder (ASD). A primary clinical hallmark of early autism is a significant, pervasive impairment in spontaneous motor imitation, joint attention, and social reciprocity during early infancy and toddlerhood.

Atypicalities in the Active Intermodal Mapping mechanism and dysregulations in underlying mirror neuron and sensorimotor networks are frequently documented in neurodivergent populations. When an infant struggles to automatically map observed social actions onto their own internal proprioceptive representations, the entire ‘Like Me’ developmental cascade is profoundly disrupted. The intuitive, effortless realization that other people are intentional social agents sharing an internal world becomes difficult to form, often precipitating the profound social communicative challenges that characterize the autism spectrum.

Clinically, this basic science has been translated directly into transformative early intervention therapies, most notably within Naturalistic Developmental Behavioral Interventions (NDBIs) such as the Early Start Denver Model (ESDM) and Reciprocal Imitation Training (RIT). In these therapeutic frameworks, trained clinicians and parents reverse the traditional pedagogical dynamic: instead of demanding that the autistic child imitate the adult, the adult systematically and lovingly mimics every motor action, vocalization, and gaze fixation produced by the child. This reciprocal, responsive imitation acts as a potent social catalyst, reigniting the child’s awareness of self-other equivalence, triggering neural sensorimotor resonance, and fostering dramatic improvements in spontaneous communicative language, eye contact, and relational engagement.

12.2 Evolutionary Robotics and Artificial Intelligence

Beyond human psychology, the computational architecture formalized in the Active Intermodal Mapping model has exerted a substantial, lasting impact on the fields of developmental robotics, cognitive artificial intelligence, and embodied machine learning. Roboticists attempting to design autonomous humanoids face an engineering bottleneck known as the “robotic correspondence problem.”

The correspondence problem is the exact robotic equivalent of the human infant’s invisibility problem: when an artificial agent observes a human human demonstrator execute an action (e.g., reaching, grasping, or walking) through digital camera sensors, how does the robot’s artificial neural network translate that external, pixel-based optical array into the complex, multi-joint actuation commands required to move its own mechanical limbs, which possess radically different kinematics, motor gearings, and degrees of freedom? Traditional brute-force programming approaches require millions of manual mathematical calculations, rendering the robotic system brittle, rigid, and incapable of adapting to novel real-world environments.

To overcome this limitation, computational roboticists have directly implemented Meltzoff and Moore’s AIM architecture into artificial neural networks. By equipping developmental robots with an artificial supramodal representational space that establishes abstract geometric and kinematic equivalences between observed visual actions and internal motor motor programs, robots can achieve autonomous, real-time imitation learning. Artificial agents utilize AIM-inspired error-correcting feedback loops to engage in successive approximations, learning complex physical tasks—from manipulating delicate tools to navigating dynamic social terrains—entirely through spontaneous, embodied imitative learning modeled after the human newborn.

12.3 Epistemological Reflections: Reconceptualizing Human Nature

Ultimately, the enduring legacy of the 1977 Meltzoff and Moore neonatal imitation experiment resides in its radical philosophical reconceptualization of human nature. For centuries, the Western intellectual tradition—from René Descartes’ isolated, solipsistic cogito to the twentieth-century computational paradigm that framed the mind as an isolated, disembodied information processor—conceptualized the human individual as fundamentally solitary, disconnected, and alienated from birth, struggling to bridge an unbridgeable chasm between the subjective self and the objective world.

Meltzoff and Moore’s empirical breakthrough revolutionized this epistemological baseline. By proving that human infants enter the world pre-equipped with an active, innate capacity to recognize conspecifics as “Like Me,” their work established that human beings are fundamentally, structurally, and biologically social organisms from their very first breath. We do not begin our lives as isolated, solipsistic thinkers who eventually learn to socialize through cold, transactional logic or societal coercion; rather, we arrive in the world wired for connection, pre-tuned to seek the faces of others, to mirror their movements, and to discover our own minds through the reflection of those who nurture us.

Five decades after its publication, the 1977 Science paper stands as a monumental milestone in the history of cognitive science. By looking deeply into the faces of newborn babies and daring to ask whether they could look back with genuine understanding, Andrew Meltzoff and M. Keith Moore permanently transformed our understanding of early development, illuminating the biological foundations of human empathy, social understanding, and intersubjectivity.

Conclusion

The scientific journey ignited by Andrew Meltzoff and M. Keith Moore’s 1977 breakthrough fundamentally altered the landscape of modern psychology and cognitive science. What began as a series of experiments investigating whether newborns could protrude their tongues or open their mouths in response to adult demonstrations evolved into a comprehensive paradigm shift that revolutionized our understanding of infant cognition, social neuroscience, and the nature of human intersubjectivity.

By establishing that human neonates can match unseen facial gestures, Meltzoff and Moore systematically dismantled the prevailing constructivist consensus of the twentieth century, compressing the developmental timeline of cross-modal representation by nearly a year. Their Active Intermodal Mapping model provided an elegant, cognitively robust theoretical framework that bridged visual perception and proprioceptive motor control through an innate supramodal representational space, prefiguring the neurobiological discovery of mirror neuron networks by decades.

While the phenomenon of neonatal imitation continues to generate scholarly debate, methodological contention, and empirical cross-examination, its historical and conceptual impact remains profound. Through the ‘Like Me’ framework, Meltzoff demonstrated how early sensorimotor matching provides the foundational developmental engine that drives the emergence of empathy, social attunement, and Theory of Mind. In an intellectual tradition that long viewed the infant as a passive, reflexive tabula rasa, Meltzoff and Moore revealed the human newborn to be an active, communicative participant in the social world—biologically pre-adapted from the very start of life to connect, to communicate, and to discover the self in the mirror of the other.

References

• Anisfeld, M. (1991). Neonatal imitation. Developmental Review, 11(1), 60–97. https://doi.org/10.1016/0273-2297(91)90003-Y
• Anisfeld, M. (1996). Only tongue protrusion modeling is matched by neonates. Developmental Review, 16(2), 149–161. https://doi.org/10.1006/drev.1996.0007
• Bower, T. G. R. (1977). A Primer of Infant Development. W. H. Freeman.
• Ferrari, P. F., Visalberghi, E., Paukner, A., Fogassi, L., Rizzolatti, G., & Suomi, S. J. (2006). Neonatal imitation in rhesus macaques. PLOS Biology, 4(9), e302. https://doi.org/10.1371/journal.pbio.0040302
• James, W. (1890). The Principles of Psychology. Henry Holt and Company.
• Jones, S. S. (1996). Imitation or exploration? Young infants’ matching of adults’ oral gestures. Child Development, 67(5), 1952–1969. https://doi.org/10.2307/1131604
• Jones, S. S. (2009). The development of imitation in infancy. Philosophical Transactions of the Royal Society B: Biological Sciences, 364(1528), 2325–2335. https://doi.org/10.1098/rstb.2009.0045
• Marshall, P. J., & Meltzoff, A. N. (2015). Body representations in the infant brain: Conducting the musical score of social development. Developmental Science, 18(6), 875–891. https://doi.org/10.1111/desc.12328
• Meltzoff, A. N. (2007). The ‘like me’ framework for recognizing and becoming an intentional agent. Acta Psychologica, 124(1), 26–43. https://doi.org/10.1016/j.actpsy.2006.09.005
• 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.198.4312.75
• Meltzoff, A. N., & Moore, M. K. (1983). Newborn infants imitate adult facial gestures. Child Development, 54(3), 702–709. https://doi.org/10.2307/1130058
• Meltzoff, A. N., & Moore, M. K. (1989). Memory in infancy: Deferred imitation in 2- to 3-month-old infants. Child Development, 60(4), 864–876. https://doi.org/10.2307/1131027
• 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
• Meltzoff, A. N., Murray, L., Simpson, E., Heimann, M., Nagy, E., Nadel, J., Pedersen, E. J., Ferrari, P. F., & Subiaul, F. (2018). Re-examining neonatal imitation: A response to Oostenbroek et al. (2016). Developmental Science, 21(6), e12609. https://doi.org/10.1111/desc.12609
• Myowa-Yamakoshi, M., Tomonaga, M., Tanaka, M., & Matsuzawa, T. (2004). Imitation in neonatal chimpanzees (Pan troglodytes). Developmental Science, 7(4), 437–442. https://doi.org/10.1111/j.1467-7687.2004.00364.x
• Oostenbroek, J., Suddendorf, T., Nielsen, M., Redshaw, J., Kennedy-Costantini, S., Davis, J., Clark, S., & Slaughter, V. (2016). Comprehensive longitudinal study challenges the existence of neonatal imitation. 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. W. W. Norton & Company.
• Piaget, J. (1952). The Origins of Intelligence in Children. International Universities Press.
• Rizzolatti, G., Fadiga, L., Gallese, V., & Fogassi, L. (1996). Premotor cortex and the recognition of motor actions. Cognitive Brain Research, 3(2), 131–141. https://doi.org/10.1016/0926-6410(95)00038-0
• Stern, D. N. (1985). The Interpersonal World of the Infant: A View from Psychoanalysis and Developmental Psychology. Basic Books.
• Trevarthen, C. (1979). Communication and cooperation in early infancy: A description of primary intersubjectivity. In M. Bullowa (Ed.), Before Speech: The Beginning of Interpersonal Communication (pp. 321–347). Cambridge University Press.