For the greater part of the twentieth century, developmental psychology and behavioral neuroscience were captivated by a reductionist, pre-deterministic paradigm. Ontogenetic progression was viewed almost exclusively through the prism of linear maturation or computational symbol processing. Human infants were conceptualized as biological automata whose physical capabilities unfolded according to an innate, hardwired cortical script, or alternatively, as central processing units executing increasingly sophisticated algorithmic representations. Motor behaviors—such as reaching, rolling, sitting, and independent bipedal locomotion—were cataloged as predetermined milestones, emerging sequentially as higher cortical centers exerted inhibitory control over primitive, lower-brainstem reflex arcs. This neuromaturational orthodoxy, while descriptive, left fundamental epistemological voids: it could not adequately explain why motor development exhibits profound inter-individual variability, how novel behavioral configurations arise without a central teleological blueprint, or why motor competence fluctuates unpredictably across varying physical environments.
In the late 1970s and throughout the 1980s and 1990s, the developmental psychologist and biomechanist Esther Thelen (1941–2004), alongside key collaborators such as Linda B. Smith, initiated an epistemological revolution. Drawing upon non-equilibrium thermodynamics, theoretical physics, ecological psychology, and complex systems mathematics, Thelen formulated the Dynamic Systems Theory (DST) of motor development. DST dethroned the sovereign, prescriptive central nervous system. Instead, it proposed that movement patterns are not centrally stored commands waiting for neurological liberation; rather, they are soft-assembled, emergent phenomena arising spontaneously from the non-linear, self-organizing interactions among heterogeneous bodily subsystems, physical task constraints, and ambient environmental forces. Within this framework, no single element—whether a cortical pathway, a regulatory gene, or an external stimulus—maintains executive authority over the organism.
Thelen’s dynamic paradigm radically altered our comprehension of the developing human organism by recasting the infant body as an open thermodynamic system continually seeking homeokinetic stability amidst perpetual physical, gravitational, and morphological change. Her work dismantled long-standing dualisms separating mind from body, perception from action, and innate nature from environmental nurture. By reconceptualizing motor development as an epigenetic, non-linear landscape characterized by attractors, bifurcations, and rate-limiting components, Thelen provided a robust empirical and theoretical apparatus that unified human movement science, pediatric physical therapy, autonomous robotics, and contemporary theories of embodied cognition. This treatise provides an exhaustive, multi-disciplinary examination of the Dynamic Systems Theory of motor development, tracing its historical emergence, foundational physical axioms, seminal empirical milestones, and enduring legacy across twenty-first-century developmental science.
1. Historical Context and Epistemological Shift in Motor Development Theories
1.1 The Hegemony of Neuromaturational Models (Gesell and McGraw)
The early twentieth-century study of infant motor development was dominated by the neuromaturational framework, championed primarily by pediatrician Arnold Gesell and developmental psychologist Myrtle McGraw. Gesell’s maturational perspective posited that ontogeny recapitulates an intrinsically determined biological schedule dictated by phylogenetic heritage. Through exhaustive cinematic documentation of infant movement, Gesell formulated normative schedules of motor development, codifying sequences such as head control, trunk stabilization, reaching, creeping, standing, and independent bipedal walking. To Gesell, these developmental sequences were direct behavioral manifestations of cortical maturation, advancing in an invariant, cephalocaudal (head-to-tail) and proximodistal (trunk-to-extremity) direction. Environmental variations, sensory inputs, and physical contexts were largely relegated to the status of mere background support, incapable of altering the foundational structural trajectory mapped within the genome.
Myrtle McGraw introduced a more nuanced neurostructural perspective, yet her theoretical architecture remained anchored in a hierarchical, linear model of neural control. Through her rigorous longitudinal investigations—most famously exemplified in the twin study of Johnny and Jimmy—McGraw mapped motor development to the anatomical maturation of the central nervous system. Her classical four-phase model of behavioral acquisition conceptualized motor development as an initial dominance of subcortical and spinal reflex mechanisms, followed by a transitional phase of cortical inhibition, subsequent cortical control of deliberate movement, and finally, full integration of subcortical and cortical operations. McGraw viewed the infant brain as an evolving physical substrate wherein the ascending myelination of white matter tracts and the dendritic arborization of the cerebral cortex progressively exerted top-down, sovereign command over rudimentary motor programs.
Despite their empirical contributions to descriptive pediatric taxonomy, these neuromaturational models suffered from profound epistemological limitations. By treating the developing central nervous system as a self-contained, pre-programmed executive engine, maturational theory failed to elucidate the generative mechanisms underlying behavioral transitions. It offered descriptive narratives of when behaviors appeared, but failed to explain how novel kinematic topologies emerged from physical substrates. Crucially, the maturational paradigm proved incapable of accounting for the immense individual variability documented across diverse cultural, rearing, and physical environments. If the neuromuscular sequence was an invariant biological unfolding driven by cortical maturation, identical neural milestones should have generated universal behavioral configurations across differing somatic scales, gravitational fields, and task ecologies. The inability of maturational models to accommodate contextual adaptability and idiosyncratic trajectories exposed the theoretical inadequacy of neuro-centric determinism.
1.2 The Rise of Information Processing Approaches in Motor Control
As the cognitive revolution gained momentum in the mid-twentieth century, the neuromaturational paradigm was supplemented, and in many domains superseded, by information processing approaches to motor behavior. Rooted in cybernetics and the computational metaphor of the mind, this perspective conceptualized the human central nervous system as an advanced digital computer. Human movement was theorized to be orchestrated through central motor programs—symbolic, representational algorithms stored in long-term memory that pre-specify the spatiotemporal activation patterns, duration, and force scaling of peripheral musculature. Richard Schmidt’s generalized motor program (GMP) and schema theory attempted to inject flexibility into this representational framework by suggesting that invariant motor features (such as relative timing and relative force) were stored centrally, while variant parameters (such as overall movement duration and absolute force) were dynamically computed before execution.
These computational and information-processing paradigms relied heavily on hierarchical, top-down execution architectures. Voluntary motor performance was construed as a sequential process involving sensory input acquisition, computational encoding, central decision-making, motor programming, feedforward motor command output, and closed-loop feedback monitoring via visual, vestibular, and proprioceptive afferents. While this approach accounted for learned, discrete skills in adults, its application to infant ontogeny generated significant theoretical paradoxes. Foremost among these was the inability of computational models to resolve the famous “degrees of freedom problem” articulated by the Russian neurophysiologist Nikolai Bernstein. Bernstein observed that the human musculoskeletal system comprises hundreds of joints, thousands of muscular units, and millions of motor neurons, generating a mathematically intractable combinatorial space. A central executive system calculating individual trajectory trajectories for every degree of freedom would instantly experience computational paralysis.
Furthermore, information processing approaches suffered from an insurmountable explanatory deficit regarding the emergence of novelty—the fundamental problem of developmental biology. If the execution of coordinated motor action requires a pre-existing symbolic program, motor schema, or representational blueprint, one must ask: where does the blueprint itself come from? Positing that the infant brain creates de novo computational programs before having ever interacted with the physical environment falls prey to homunculus reasoning, requiring an internal designer to write the software. The information processing framework could not explain how a pre-reaching infant, devoid of representational motor schemas for reaching, suddenly produces an intentional manual strike, nor how an organism transitions across topologically discrete motor states without an advance cognitive template explicitly directing the physical phase shift.
1.3 Esther Thelen’s Paradigmatic Challenge and Ecological Foundations
Recognizing the foundational impasses of both neuromaturational determinism and cognitive computationalism, Esther Thelen mounted a transformative challenge to developmental science. Her theoretical revolution was forged by synthesizing two previously disparate intellectual traditions: the biomechanical insights of Nikolai Bernstein and the ecological psychology of James J. Gibson. From Bernstein, Thelen adopted the premise that movement is fundamentally an exercise in physics and biomechanics, where neural activity is merely one component interacting with inertias, gravities, reactive forces, and tissue elasticities. Movement is not dictated by the nervous system; rather, the nervous system participates in a dynamic dialogue with peripheral physical forces. From Gibson’s ecological psychology, Thelen incorporated the principle of perceptual-motor coupling and the concept of affordances—the relational opportunities for action provided by the physical environment to an embodied agent.
By blending Bernstein’s biomechanics and Gibsonian ecological realism with the non-equilibrium physical sciences, Thelen effected a conceptual shift away from static representational models toward non-linear physical system dynamics. Instead of conceptualizing development as the hardwired execution of neural programs, she defined it as a continuous process of soft-assembly. In dynamic systems, coordinated behavioral patterns do not pre-exist in the genes or the brain; they assemble spontaneously in real time through the cooperative interaction of multiple, heterogeneous bodily, neural, and environmental subsystems. Under this view, order is not imposed from above by an executive controller; it emerges organically from the bottom up through self-organizing physical dynamics.
Thelen’s empirical journey began with her meticulous naturalistic and laboratory observations of rhythmic infant stereotypies. In the late 1970s, she systematically tracked hundreds of pre-verbal infants, documenting repetitive, non-goal-directed movements—such as rhythmic leg kicking, foot rubbing, arm banging, and torso swaying. Neuromaturational theorists had dismissed these rhythmic stereotypies as purposeless, noise-like epiphenomena of an immature, incompletely myelinated motor system. Thelen, however, recognized these behaviors as spontaneous coordination phenomena analogous to the intrinsic periodic oscillations observed in coupled non-linear physical and chemical systems. She demonstrated that these stereotypies were universal, self-organizing biomechanical solutions through which infants explored their dynamic multi-limb movement spaces. These oscillations served as the intrinsic dynamic building blocks from which later, goal-directed voluntary behaviors—such as reaching, crawling, and upright walking—would be soft-assembled.
2. Foundational Principles of Esther Thelen’s Dynamic Systems Theory
2.1 The Concept of Soft-Assembly vs. Hardwired Motor Programs
At the center of Thelen’s theoretical framework lies the radical dichotomy between hardwired motor programs and soft-assembled actions. Traditional models assumed that functional behaviors are executed via rigid, pre-formed central representations stored within designated neural substrates. Under this hardwired view, every manifestation of a motor skill—such as bringing a hand to an object—is governed by an internal script dictating muscle contraction timings, joint angles, and torque vectors. Thelen dismantled this perspective by demonstrating that biological systems are fundamentally plastic, opportunistic, and context-dependent. The human body does not rely on rigid blueprints because its physical properties and the surrounding environment are in a state of perpetual flux.
Soft-assembly refers to the temporary, flexible, and task-specific coalescing of diverse anatomical, physiological, neural, and environmental subsystems to achieve a behavioral goal. The components that constitute a soft-assembled action are heterogeneous: they include the firing rates of central motor pools, the visco-elastic resistance of tendons, the instantaneous angle of musculoskeletal levers, the resting metabolic reserve of the infant, the frictional resistance of the support surface, the gravitational force vector, and the affective-motivational state of the child. When an infant executes an action, these heterogeneous elements converge dynamically to generate a functional kinematic outcome. Critically, this organization occurs without an executive controller or central processing unit dictating the individual trajectory of each mechanical degree of freedom.
This dynamic coalition operates on contextual fluidity rather than static representational recall. Because the assembly is soft, the constituent components can continuously configure and reconfigure based on minute alterations in environmental or internal conditions. If an infant is reaching for a toy while seated upright, the subsystem coalition involves specific thoracic stabilizing activations, particular shoulder torque generations, and specific optical feedback loops. If the same infant attempts the same reach while lying prone or submerged in water, the physical constraints alter drastically. Rather than recalculating a complex computational algorithm, the biological dynamic system immediately self-organizes an entirely novel, soft-assembled coalition of muscular forces that utilizes the immediate biomechanical affordances of the environment to fulfill the behavioral goal.
2.2 Heterochronic Development of Subsystems
A cornerstone of Dynamic Systems Theory is the principle of heterochrony—the asynchronous, non-uniform rates of maturation exhibited by the various physiological, structural, and neurological subsystems that comprise the total organism. Neuromaturational theory erroneously assumed that development proceeds in a unified, homogenous wave driven by the linear maturation of the brain. Thelen revealed that the human infant is a mosaic of independently and unevenly maturing components. Musculoskeletal mass, bone density, extensor muscle contractile strength, balance and postural mechanisms, cardiovascular endurance, peripheral neural myelination, central synaptic density, visual acuity, and motivational drive all develop along distinct, non-parallel temporal trajectories.
Because these internal subsystems mature at disparate rates, their real-time interactions do not produce a continuous, gradual improvement in overall motor performance. Instead, heterochronic development is directly responsible for the ubiquitous presence of developmental plateaus, sudden behavioral regressions, and dramatic phase transitions observed throughout early childhood. An infant may possess the neural central pattern generation necessary for reciprocal leg movements, yet remain incapable of independent standing because skeletal muscle mass has not yet achieved the mechanical threshold required to counteract gravitational acceleration. Conversely, rapid morphological growth, such as sudden weight gain, can temporarily outstrip muscular force production, causing an apparent regression or cessation of an existing motor behavior.
The concept of continuous, asynchronous interaction among independently maturing biological structures fundamentally reframes our understanding of non-uniform motor progression across different domains. Motor skills do not appear in an invariant lockstep across all children because each child possesses a unique heterochronic profile. An infant with advanced visual-perceptual subsystems but slow-maturing muscular mass will encounter and solve motor problems through an entirely different structural dynamic than an infant characterized by high physical mass, robust muscular tone, but slower sensory integration capabilities. Motor ontogeny is thus not a synchronized march along a single biological timeline, but a turbulent, polyphonic dialogue among asynchronously evolving subsystems.
2.3 The Rate-Limiting Component (Rate Limiter)
Within a dynamic system composed of heterochronically developing elements, the emergence of any new macroscopic behavioral pattern is strictly governed by the slowest developing component. In Thelen’s theoretical nomenclature, this bottleneck is termed the rate-limiting component or rate limiter. A motor behavior cannot assemble itself simply because ninety percent of the required subsystems are fully mature and functional; if a single critical subsystem has not yet reached its functional threshold, the entire macroscopic motor pattern is prevented from manifesting. The rate limiter serves as the physical or neurological anchor that holds the entire collective system in its current behavioral regime.
Rate limiters frequently manifest as biomechanical or energetic constraints rather than neurological deficits. In the acquisition of independent walking, for example, the fundamental rate limiter is rarely the neural architecture responsible for coordinated limb alternating patterns—which, as Thelen empirically demonstrated, is functionally operational months prior to onset. Instead, the rate limiters are almost universally biomechanical: namely, the lack of sufficient extensor muscle force to support dynamic single-limb stance against gravity, coupled with immature dynamic postural equilibrium mechanisms capable of actively managing the center of mass over a narrow base of support. Until extensor strength and postural equilibrium cross their critical thresholds, voluntary independent bipedal locomotion remains physically impossible, regardless of cortical intent.
Sensory-perceptual, affective, and cognitive thresholds can likewise function as powerful rate limiters. In voluntary manual prehension, the rate limiter may not be the muscular capacity to flex the digits, but rather the sensory-perceptual capacity to accurately map binocular disparity cues to intrinsic motor coordinates, or the inability of the infant to coordinate visual attention with the somatic position of the arm. The clinical and theoretical significance of the rate limiter is profound: overcoming a rate limiter is the primary catalyst that triggers qualitative motor reorganization. When the rate-limiting subsystem finally matures to its critical tipping point, it releases the systemic bottleneck, allowing the entire coalition of subsystems to instantaneously self-organize into a radically novel, qualitative motor pattern.
3. Self-Organization and Non-Linearity in Motor Skill Acquisition
3.1 Principles of Self-Organization in Open Thermodynamic Systems
To provide a rigorous mathematical and physical foundation for Dynamic Systems Theory, Esther Thelen integrated the principles of self-organization derived from non-equilibrium thermodynamics and synergetics. Drawing extensively upon the work of Nobel laureate Ilya Prigogine regarding dissipative structures and Hermann Haken‘s framework of synergetics, Thelen conceptualized the developing infant as an open thermodynamic system existing far from thermodynamic equilibrium. In open systems characterized by continuous fluxes of matter and energy, highly ordered macroscopic spatio-temporal patterns can emerge spontaneously from the local interactions of microscopic elements without the necessity of an internal or external organizer.
In physical systems, self-organization is readily observable in phenomena such as Rayleigh-Bénard convection cells: when a thin layer of liquid is heated uniformly from below, the chaotic thermal agitation of molecules suddenly self-organizes into an array of hexagonal convective rolls once a critical temperature differential is reached. There is no blueprint directing the molecules; the macroscopic order is a purely emergent property of the physical system dissipating energy. Thelen applied this exact physical logic to human ontogeny. The infant motor system is subjected to energetic, gravitational, and metabolic fluxes. Rather than succumbing to biological entropy or chaos, the physical components of the infant—neurons, muscles, soft tissues, limbs—spontaneously self-organize into stable, macroscopic behavioral patterns (such as reciprocal kicking, reaching, or crawling) to maintain stability and dissipate energetic demands under specific environmental perturbations.
Self-organization fundamentally refutes the requirement for explicit genetic or neural instructions. It demonstrates that complex biological coordination is not synonymous with complex central computation. Complex macroscopic patterns can and do emerge from remarkably simple, lower-level local interactions governed by mechanical laws, physical geometries, and energetic potentials. By adopting this principle, Thelen established that order in infant behavior is a natural thermodynamic consequence of an embodied, situated organism acting within a physical world, effectively banishing the ghost of the homunculus from the machinery of human motor development.
3.2 Non-Linear Dynamics: Sudden Shifts from Quantitative to Qualitative Changes
Biological development has historically been mischaracterized as a smooth, continuous, and linear progression. Linear systems obey the principle of superposition: output changes are directly proportional to input changes. Dynamic systems, conversely, are inherently non-linear. In a non-linear system, small, continuous, quantitative changes in a single underlying variable can accumulate silently until a critical tipping point is reached, precipitating a sudden, discontinuous, and qualitative restructuring of the entire system’s macroscopic behavior. This qualitative phase transition is known mathematically as a bifurcation point.
Thelen demonstrated that human motor development progresses through a succession of these non-linear bifurcations. An illustrative example is the transition from pre-locomotor rocking to reciprocal creeping. An infant placed in the quadrupedal position may spend weeks engaging in rhythmic, quantitative anteroposterior rocking. During this period, gradual, continuous micro-level physiological expansions are occurring: extensor muscle fibers are subtly hypertrophying, postural sway damping is marginalizing instability, and sensorimotor calibration is incrementally refining. To a traditional observer, the child appears locked in a developmental plateau. However, when extensor force production and balance control reach a critical threshold, the system undergoes a sudden bifurcation: the stable rocking pattern abruptly collapses, and the infant spontaneously organizes a qualitatively novel coordination pattern—reciprocal, diagonally cross-coupled crawling.
In the language of dynamic modeling, the variable that drives this non-linear reorganization is termed a control parameter. A control parameter does not contain the plan, map, or code for the resulting pattern; it is merely an energetic, physical, or contextual variable that, when scaled through a continuous range of values, pushes the system into regions of critical instability. Once instability is reached, the system’s internal degrees of freedom are forced to spontaneously reorganize into a new topological state. Thus, quantitative linear increments in physical parameters (e.g., muscle strength, movement velocity, environmental slope) routinely generate discontinuous, non-linear developmental breakthroughs in motor ontogeny.
3.3 Order Parameters and Collective Variables
To mathematically characterize and empirically quantify the non-linear emergence of coordinated movement, dynamic systems theorists utilize order parameters, also referred to as collective variables. When an infant moves, the degrees of freedom problem is staggering: thousands of motor units, skeletal levers, and spatial coordinates are continuously shifting. It is practically impossible to model the system by tracking every micro-variable independently. Hermann Haken’s synergetic insight, adopted by Thelen and motor scientist J. A. Scott Kelso, was that during self-organization, the high-dimensional microscopic degrees of freedom become compressed into a low-dimensional macroscopic variable that captures the collective essence of the entire coordinated action.
The order parameter defines the relational architecture of the movement topology. In the study of inter-limb coordination—such as infant kicking, crawling, or running—the most universally recognized order parameter is relative phase ($phi$). Relative phase measures the spatiotemporal phase relationship between two oscillating limbs across a movement cycle, expressed as an angular value between 0° and 360°. When two limbs move in complete synchrony (such as the bilateral synchronous leg thrusts seen in early infancy), the relative phase is approximately 0° (an in-phase coordination state). When the limbs alternate symmetrically (such as the reciprocal kicking of an infant or bipedal walking), the relative phase hovers near 180° (an anti-phase coordination state).
By monitoring order parameters such as relative phase, researchers can quantitatively capture the precise moment a motor system loses stability and undergoes a qualitative phase shift. When an infant transitions from an in-phase bilateral leg movement to an anti-phase reciprocal stepping regime, the relative phase shifts discontinuously. Furthermore, order parameters allow researchers to evaluate the structural stability and maturity of a motor behavior. High variability in the collective variable denotes a fragile, transitional, or exploratory state, whereas low standard deviation in relative phase signifies a consolidated, stable, and resilient macroscopic attractor regime.
4. The Disappearance and Re-emergence of the Infant Stepping Reflex
4.1 The Neuromaturational Fallacy of Cortical Inhibition
To appreciate the empirical triumph of Esther Thelen’s dynamic systems framework, one must examine her seminal deconstruction of the infant stepping reflex—a pedagogical cornerstone of classic twentieth-century pediatric neurology. When a newborn infant of mere days or weeks is held vertically with their soles in contact with a flat surface, the child reliably executes brisk, coordinated, alternating steps reminiscent of mature bipedal walking. Historically, pediatric lore observed that this robust stepping behavior progressively diminishes between two and three months of age, virtually disappearing until it seemingly re-emerges toward the end of the first year as intentional, voluntary walking.
The neuromaturational canon, formulated by Gesell and McGraw and reinforced by generations of pediatricians, explained this phenomenon through the dogma of cortical inhibition. It was universally asserted that newborn stepping was an archaic, primitive subcortical reflex governed entirely by lower spinal or brainstem mechanisms. As the phylogenetically newer cerebral cortex matured, its descending corticospinal pathways allegedly developed an active inhibitory influence over these primitive subcortical circuits, systematically suppressing the stepping reflex to prepare the nervous system for advanced, voluntary cortical locomotion. The subsequent re-emergence of stepping at 10–12 months was interpreted as the triumphant realization of cortical motor commands replacing the suppressed reflex arcs.
Thelen identified glaring logical, physiological, and empirical flaws within this neuromaturational narrative. First, if stepping disappeared because descending cortical pathways were inhibiting spinal pattern generators, why did identical, alternating reciprocal limb movements persist unabated when infants were placed in a supine position? An infant lying on their back continues to kick vigorously with the exact same reciprocal kinematic pattern throughout the second, third, and fourth months of life. Second, neuroanatomical studies of the era failed to identify the emergence of specific, dedicated descending inhibitory pathways that synchronized chronologically with the precise window of reflex disappearance. The attribution of behavioral cessation purely to “cortical inhibition” was, in truth, an unfalsifiable conceptual placeholder that concealed a failure to investigate the real physical changes taking place in the infant’s growing body.
4.2 Thelen’s Seminal Biomechanical Experiments (Submersion and Weights)
Thelen approached the enigma of the disappearing stepping reflex not as a question of neuroanatomy, but as a problem of physical growth and functional biomechanics. Analyzing infant anthropometrics, she recognized that the first three months of human postnatal life are characterized by an explosive deposition of subcutaneous adipose tissue. During this rapid growth phase, the infant’s body mass increases dramatically. Crucially, the rate of fat accumulation far outpaces the development of skeletal muscle mass and contractile force generation. The infant’s legs become considerably heavier, thicker, and denser, but the muscular capacity to flex these increasingly heavy limbs against the continuous downward pull of gravity does not experience a corresponding proportional increase.
Hypothesizing that stepping disappeared because the leg muscles were simply too weak to lift the heavy limbs in an upright gravitational orientation—rather than due to cortical inhibition—Thelen conducted two revolutionary, exquisitely simple biomechanical experiments. In the first intervention, she took older infants (around 3 to 4 months of age) who had entirely ceased to display the stepping reflex in standard clinical evaluations. She submerged these infants up to their chests in tanks of warm water. The buoyant force of the water effectively countered gravitational acceleration, drastically reducing the net gravitational load on the lower extremities. Instantly and dramatically, the stepping reflex returned: submerged infants produced rapid, elegant, reciprocal alternating steps with high kinematic regularity.
To decisively corroborate the causal mechanism, Thelen conducted the reciprocal experiment. She recruited young, healthy 4-week-old infants who actively and reliably displayed vigorous upright stepping. She attached small, precisely calibrated lead-weight cuffs to their ankles, effectively simulating the future mass-to-strength ratio they would naturally encounter at three months of age. Upon the application of this mechanical load, the stepping behavior immediately ceased; the infants ceased to produce the alternating steps. When the weights were removed, stepping spontaneously resumed. These paradigm-shifting experiments demonstrated that the neural Central Pattern Generator (CPG) for stepping had never been cortically inhibited at all. The underlying neural architecture remained intact, functional, and active; the temporary disappearance of the reflex was driven entirely by a biomechanical rate limiter: the ratio of subcutaneous mass to extensor-flexor muscular strength.
4.3 The Motorized Treadmill Experiments
To further solidify her dynamic systems reinterpretation, Thelen, along with Beverly Ulrich and colleagues, subjected pre-walking infants—ranging from 1 to 7 months of age—to motorized treadmill paradigms. If stepping required mature, voluntary cortical motor programs, pre-walking infants held upright over a moving treadmill belt should have exhibited disorganized, dragging limb trajectories, as they were biologically months away from walking independently. Instead, the experimental results decisively upended traditional developmental assumptions.
When placed over a motorized treadmill moving at slow, uniform speeds, pre-walking infants instantly displayed well-coordinated, alternating bipedal stepping. The backward translation of the treadmill belt dynamically stretched the hip flexor muscles and the surrounding joint capsules of the stance limb, triggering proprioceptive stretch receptors that spontaneously initiated a kinematic swing phase, while the contralateral limb adopted a stable stance phase. To push the dynamic inquiry further, Thelen utilized split-belt treadmills, where the left and right belts were driven by independent motors running at radically different velocities (e.g., one belt moving twice as fast as the other). Pre-walking infants, completely lacking voluntary cortical walking experience, dynamically accommodated to the asymmetric physical demands: the limb on the faster belt stepped with a accelerated cycle frequency, while the limb on the slower belt adjusted its stance duration, preserving stable, anti-phase inter-limb coordination.
These split-belt experiments provided incontrovertible proof that complex, adaptive, and highly articulated motor patterns do not require top-down cortical execution. Coordinated bipedal locomotion emerged dynamically at the physical interface between peripheral biomechanics, spinal neural loops, and environmental affordances. The motor behavior was an emergent property of the organism-environment coupling. Thelen fundamentally exposed the neuromaturational fallacy: the stepping reflex had not been hidden away by cortical masters, nor was walking an intellectual cortical invention. Motor development was, at its core, the continuous, physical self-organization of an embodied system navigating physical forces.
5. The Triad of Constraints: Organism, Environment, and Task
5.1 Organismic (Individual) Constraints: Morphological and Physiological Variables
Building upon the theoretical framework of biomechanist Karl Newell, Esther Thelen conceptualized motor development as operating within an inescapable triad of constraints: Organism, Environment, and Task. The dynamic system cannot produce an infinite array of kinematic actions; its possibilities are sculpted, bounded, and channeled by these three mutually interacting categories. The first facet of this triad comprises organismic constraints, representing the internal structural, morphological, and physiological boundaries intrinsic to the individual infant.
Organismic constraints encompass structural parameters including body mass index (BMI), absolute limb lengths, segments’ moments of inertia, and the continuous downward migration of the anatomical center of mass from the thoracic to the pelvic region during the first eighteen months of life. Furthermore, organismic constraints include the physical and viscoelastic properties of the biological tissues themselves: resting muscle tone, ligamentous laxity, cross-sectional muscle area, tendon stiffness, and the contractile speed of varying muscle fiber phenotypes. Neurological and internal physiological parameters likewise act as organismic constraints: the current state of neural myelination, sensory integration thresholds, synaptic firing rates, metabolic energy reserves, and even the immediate affective, emotional, and arousal states of the developing child.
Because no two infants possess identical structural, genetic, or physiological dimensions, their internal organismic constraints are radically individualized. A stocky, heavily muscled infant faces profoundly different inertial resistances and rotational moments than an infant with lean, elongated limbs and higher joint laxity. Consequently, the mechanical solutions that emerge to solve a universal motor challenge—such as maintaining an upright posture or executing an arm reach—cannot be universal. The somatic landscape of the child establishes an individual boundary condition, meaning that motor development is fundamentally an idiosyncratic journey of discovering functional coordination states that align with one’s own unique bodily geometry.
5.2 Environmental Constraints: Gravitational and Ambient Ecology
The second component of Newell’s triad involves environmental constraints—the external, physical, ecological, and sociocultural conditions within which the developing organism is permanently situated. The most continuous, pervasive, and non-negotiable environmental constraint acting upon biological organisms is the gravitational force vector. Gravity is not merely a passive backdrop; it is an omnipresent mechanical force that exerts continuous torque across every skeletal joint. Every developmental milestone—from the neonate lifting their chin off the mattress to the toddler mastering upright equilibrium—is essentially a negotiated settlement between muscular contractile force, skeletal geometry, and gravitational acceleration.
Beyond gravity, ambient physical constraints include the specific physical affordances of the surfaces with which the infant interacts. The mechanical compliance, structural rigidity, coefficient of friction, slope, and stability of a support surface profoundly alter motor assembly. An infant who easily displays stable quadrupedal crawling on a high-friction carpeted floor may become entirely immobilized or revert to a primitive belly-slide when placed upon a low-friction polished hardwood floor or a highly compliant waterbed mattress. Ambient environmental constraints also encompass optical flow characteristics, ambient illumination, ambient temperature, and acoustic fields, all of which supply sensory information that continuously modulates the real-time stabilization of movement.
Equally critical are the sociocultural and caregiving environments that shape the infant’s ecological niche. Environmental constraints are heavily modulated by cultural parenting practices, such as routine infant carrying strategies, the traditional use of cradleboards, the temporal allocation of “tummy time” in the prone position, and the ubiquity of modern mechanical seating devices, bouncers, and walkers. For example, infants raised in cultures that practice rigorous daily stretching and suspension rituals (such as those observed in parts of West Africa and the Caribbean) display significantly accelerated walking onsets compared to Western infants who spend significant periods restrained in car seats and strollers. These sociocultural variations are not noise within the system; they are structural modifications of the environmental constraint space that dictate how, when, and where motor exploration occurs.
5.3 Task Constraints: Teleological Drives and Mechanical Goals
The final element of the dynamic triad comprises task constraints—the specific, teleological goals, physical rules, and mechanical requirements that define the purpose of an action. Movement in humans is rarely a random, unmotivated firing of motor units; it is predominantly functional, intentional, and goal-directed. The presence of a concrete behavioral objective fundamentally reconfigures the physical properties of the motor assembly. The intention to obtain a toy, to embrace a caregiver, or to escape an uncomfortable physical orientation dynamically reorganizes the kinematic and kinetic parameters of the infant’s musculoskeletal system.
Task constraints encompass the physical, geometric, and functional properties of the objects with which the infant interacts. Grasping a millimeter-sized crumb requires a finely tuned pincer configuration, precise manual deceleration, and delicate force calibration; grasping a large, slippery ball necessitates a bilateral, high-force embracing synergy. The spatial boundaries, speed demands, and energetic efficiencies of the task impose hard physical limits on the degrees of freedom that the motor system can employ. If the task requires rapid locomotion across a wide room to reach a departing parent, the task constraint actively penalizes slow quadrupedal crawling, creating an energetic impetus that forces the system toward bipedal walking.
Crucially, motor competence is not an abstract, generalizable biological status that an infant possesses in a vacuum; it is strictly an emergent product of the intersection of all three constraints. Functional competence is context-dependent. An infant may demonstrate sophisticated postural stability within a specific task context (e.g., sitting while securely supported by an array of pillows to play with a light-up toy), yet display catastrophic postural collapse when the task is altered to reaching for a distant, heavy object while seated on an unsupported bench. The dynamic systems paradigm insists that one can never evaluate, diagnose, or understand motor development by isolating the child from the task and environmental ecology in which their actions are constructed.
6. Attractor States, Phase Shifts, and Behavioral Stability
6.1 The Architecture of the Epigenetic Attractor Landscape
To conceptualize the temporal evolution, stability, and plasticity of motor behavior, Esther Thelen and her colleagues adopted the mathematical topography of the attractor landscape, an idea deeply connected to C. H. Waddington’s concept of the epigenetic landscape. In non-linear dynamic systems, the state space (the multi-dimensional mathematical space representing all possible configurations of the system’s degrees of freedom) is not uniformly accessible. Instead, the physical dynamics carve out preferred behavioral regimes known as attractor states—stable regions within the potential energy landscape toward which the system naturally gravitated and to which it tends to return following minor perturbations.
The topology of this attractor landscape is defined by its basins of attraction, which vary continuously in their depth and steepness. A deep, steep attractor basin represents a highly consolidated, stable, and energy-efficient motor state. When an infant is operating within a deep attractor basin, their movement pattern is robust, consistent, and exceptionally resistant to external disruptions; if perturbed, the system rapidly decays back to this preferred coordination state. Conversely, a shallow attractor basin signifies a flexible, fragile, and loosely assembled motor pattern. These shallow regimes are easily destabilized by internal physiological fatigue or external environmental shifts, representing emerging or transitional coordination states that have not yet achieved dynamic consolidation.
Over ontogenetic time, the entire architectural topography of the phase space evolves continuously via the interaction of genetic, biomechanical, neural, and environmental forces. What was once a deep, insurmountable attractor basin at four months of age (such as reciprocal supine kicking) gradually flattens and transforms into an expansive plateau, while new attractor basins (such as quadrupedal creeping, seated balance, and upright bipedal gait) spontaneously carve themselves into the somatic terrain. The dynamic infant navigates this shifting potential energy landscape, dynamically settling into functional attractor states that minimize energetic dissipation while satisfying the constraints of the immediate task.
6.2 Phase Shifts, Perturbation, and Critical Fluctuations
How does a biological organism escape an existing, deeply established attractor state to acquire a novel, higher-order motor skill? Thelen revealed that within a dynamic system, instability is the essential prerequisite for developmental change. In direct opposition to neuromaturational models that viewed behavioral stability as the sole hallmark of developmental maturation, Dynamic Systems Theory demonstrated that development cannot progress without the deliberate destabilization of existing motor regimes.
The transition of a system from one attractor state to another is termed a phase shift. Prior to any qualitative phase shift, the system must enter a period of heightened behavioral instability, characterized mathematically by the emergence of critical fluctuations. When an existing attractor basin begins to lose its stability—due to continuous quantitative changes in an underlying control parameter (such as increasing limb mass, rising extensor strength, or intentional drive)—the variability of the collective variable dramatically increases. The system is no longer anchored firmly to a single coordination regime; it begins to fluctuate wildly, actively exploring the boundaries of its state space.
Experimentally, this loss of stability is identified through perturbation protocols and the empirical measurement of relaxation time. If a stable dynamic system is mechanically perturbed, its relaxation time—the temporal interval required to return to the original attractor state—is nearly instantaneous. However, as the system approaches a critical bifurcation point, its relaxation time becomes progressively longer, a diagnostic phenomenon known in physics as critical slowing down. The infant’s motor behavior becomes erratic, displaying heightened trial-to-trial variance. Far from being a symptom of neurological breakdown or motor error, these critical fluctuations are the indispensable engine of developmental innovation, allowing the infant to uncouple old mechanical patterns and soft-assemble radically novel motor solutions.
6.3 Hysteresis and Non-Equilibrium Phase Transitions
A definitive hallmark of non-linear, non-equilibrium dynamic systems is the phenomenon of hysteresis. Hysteresis describes a system state where the behavioral output is not merely a function of the current control parameter value, but is dependent upon the system’s prior history and developmental trajectory. In a dynamic system displaying hysteresis, the path taken to transition from State A to State B is not identical to the reverse path taken when transitioning from State B back to State A; the critical bifurcation points occur at different parameter thresholds depending on the direction of scaling.
This dynamic asymmetry is classic in locomotion and is formalized extensively in the celebrated Haken-Kelso-Bunz (HKB) model of non-equilibrium phase transitions. When a human subject is on a treadmill whose velocity is progressively scaled upward, the transition from a walking pattern to a running pattern occurs at a specific critical velocity threshold. However, when the treadmill velocity is systematically decelerated from high speeds back to slow speeds, the transition from running back to walking does not occur at the same velocity; the system remains in the running attractor state for longer, switching back to walking at a significantly lower speed. The previous historical state of the system actively preserves dynamic stability, effectively anchoring the existing attractor basin against premature collapse.
In infant ontogeny, hysteresis explains why newly acquired motor regimes exhibit directional resistance to regression. An infant who has recently traversed the critical bifurcation from crawling to independent bipedal walking will continue to choose walking even when placed in environmental contexts that mechanically favor crawling. The experiential history of navigating the walking attractor basin fundamentally restructures the potential landscape, stabilizing the novel behavior against historical reversion. Thelen showed that dynamic hysteresis operates across multiple scales of developmental time, ensuring that as biological systems explore new coordination states, their prior behavioral histories serve as stabilizing scaffolds that preserve functional integrity amidst continuous morphological and environmental change.
7. Re-evaluating Traditional Milestones: Variability as an Engine of Change
7.1 The Pathological View of Variability vs. Dynamic Systems Interpretation
The epistemological gulf separating classical neuromaturational models from Dynamic Systems Theory is nowhere more starkly illuminated than in their contrasting interpretations of movement variability. Under the dominion of neuromaturational and information-processing paradigms, behavioral variability was pathologized. Deviations from an established kinematic average were viewed as system noise, motor execution errors, structural immaturity, or indications of neurological compromise. Standardized clinical assessments were explicitly constructed to measure how closely a child’s movement conformed to a static, normative template, with variance viewed as an undesirable index of developmental deficit.
Thelen and her contemporaries upended this perspective, arguing that movement variability is the primary source of developmental adaptability and learning. In dynamic systems, intra-individual variability is not random computational noise; it is functional, exploratory biological search behavior. An infant attempting to coordinate an arm to touch a dangling mobile does not possess an innate mathematical formula for reaching. Instead, the infant utilizes kinematic and kinetic variability to actively explore the multi-dimensional biomechanical landscape of their own body, discovering what their limbs can do within specific physical contexts.
Dynamic Systems Theory distinguishes sharply between dysfunctional, unstructured random variability (such as the uncontrolled movements seen in severe choreoathetosis) and healthy, functional exploratory variability. Functional variability is the indispensable raw material upon which developmental selection operates. Without variability, the infant would remain trapped in rigid, stereotypical motor attractors, incapable of adapting to sudden changes in terrain, load, or task constraints. Movement variability allows the biological system to test multiple muscle coalitions, sample sensory feedbacks, map dynamic joint torques, and discover energetically optimal attractor states. It is the exploratory spark that fuels self-organization.
7.2 Deconstructing the Linear Progression of Developmental Milestones
For decades, pediatric medicine has utilized standard developmental milestone charts—such as the Denver Developmental Screening Test—which present the acquisition of motor skills as a strict, unyielding, linear ladder. The infant is expected to conquer motor milestones in a universal, lockstep progression: rolling from prone to supine, sitting with support, independent sitting, quadrupedal crawling, cruising along furniture, standing, and finally, independent bipedal walking. Departure from this rigid sequence is frequently met with clinical anxiety and diagnostic scrutiny.
Thelen’s dynamic systems framework fundamentally deconstructs the validity of this linear developmental orthodoxy. Empirical investigations have consistently revealed that millions of typically developing infants around the globe traverse non-standard, alternative locomotor trajectories. Substantial cohorts of children entirely skip the quadrupedal crawling phase; instead, they engage in prolonged “bottom-shuffling” (hitch-sitting locomotion), belly-crawling, rolling locomotion, or transition directly from seated balance to upright bipedal cruising and independent walking. These alternative trajectories do not represent developmental pathology or neurological deficit; they are brilliant, self-organized biomechanical adaptations to specific organismic constraints (such as body weight distributions or joint stiffness) interacting with specific cultural-environmental ecologies (such as polished tile floors, specialized garments, or sleeping positioning practices).
By demonstrating that common mechanical problems have multiple valid dynamic solutions, Thelen demonstrated that rigid milestone adherence obscures the underlying systemic health of the child. Motor development is not a mono-rail train traveling through mandatory neurological stations; it is a multi-branching river navigating a complex landscape. Equifinality—the principle that a dynamic system can reach the same functional end-state from radically different initial conditions and via vastly divergent developmental pathways—is the absolute rule of human motor ontogeny. The dynamic health of a child is defined not by the speed with which they tick off normative milestones, but by their capacity to soft-assemble flexible, functional solutions within the constraints of their physical ecology.
7.3 Dynamic Selection and Neuronal Group Selection Theory (Edelman’s Integration)
To establish the biological and neural mechanism through which exploratory motor variability is channeled into stable functional actions, Esther Thelen integrated her dynamic framework with Gerald Edelman‘s Theory of Neuronal Group Selection (TNGS), colloquially known as Neural Darwinism. Edelman posited that the brain does not operate as an instruction-driven computer executing pre-written codes, but as a complex, self-organizing selective system operating on evolutionary principles within the individual lifespan.
TNGS articulates three primary tenets: primary repertoire formation (an initial anatomical exuberance of synaptic connections formed via embryonic development), secondary repertoire selection (the differential strengthening or weakening of synaptic connections driven by real-time behavioral experience), and reentry (the continuous, reciprocal signaling between distributed neural maps). Thelen recognized that Edelman’s concept of somatic selection was the precise neural equivalent of her biomechanical self-organization. As an infant executes variable, exploratory motor movements, the vast, degenerate neural architecture of the motor cortex, basal ganglia, and spinal cord generates an expansive diversity of exploratory motor commands.
When one particular soft-assembled motor coalition successfully achieves a behavioral goal—such as the infant’s hand accidentally colliding with a desired rattle—the resulting burst of visual, auditory, and proprioceptive sensory feedback acts as a powerful value signal. In accordance with Edelman’s selective mechanisms, this value signal triggers immediate synaptic consolidation within the active neuronal groups, biochemically reinforcing the specific neural-muscular pathways that produced the successful action. Through thousands of iterations of this real-time perceptual-motor exploratory loop, adaptive somatic patterns are dynamically selected, stabilized, and sculpted into deep attractor basins, while inefficient, non-functional kinematic coalitions are systematically pruned. Development is thus Neural Darwinism embodied in biomechanical action.
8. Thelen’s Empirical Methodologies: Microgenetic and Longitudinal Designs
8.1 Microgenetic Tracking of Reaching Behaviors
The epistemological shift demanded by Dynamic Systems Theory required an equally radical revolution in empirical research methodology. Traditional developmental studies relied overwhelmingly on static, cross-sectional designs—testing different cohorts of infants at 3, 6, 9, and 12 months—or coarse-grained longitudinal intervals spaced months apart. Thelen recognized that cross-sectional snapshots entirely obliterate the fluid, non-linear dynamics of developmental change, smoothing over the critical instabilities, fluctuations, and bifurcations that define ontogenetic transitions.
To capture self-organization in real time, Thelen pioneered the extensive use of microgenetic designs and dense, high-frequency longitudinal tracking. In her landmark studies of the acquisition of voluntary reaching, Thelen, Corbetta, and colleagues followed infants on a weekly, and sometimes daily, basis from approximately three weeks of age until functional, coordinated reaching was fully consolidated. The resulting data revealed that the emergence of manual prehension was not a universal, homogeneous cortical program, but a profoundly individualized, dynamic problem-solving process, brilliantly illustrated by the contrasting longitudinal trajectories of two famous research infants: Gabriel and Hannah.
Gabriel entered the study as an extraordinarily active, vigorous, and hyper-energetic infant whose natural baseline state was characterized by rapid, high-amplitude, ballistic limb oscillations. When Gabriel desired an object, his limbs flailed wildly with explosive kinetic energy. For Gabriel, the dynamic rate-limiting problem was one of damping: his challenge was to decelerate his hyperactive limbs, recruit antagonist muscle co-contractions, and stabilize his posture to allow his hand to land accurately on the target. Hannah, in stark contrast, was an exceptionally calm, placid, and hypotonic infant who moved infrequently and with very low mechanical force. Her arms rested languidly against her torso. For Hannah, the dynamic rate limiter was the complete opposite: she needed to generate sufficient active muscular power to accelerate her arm against gravity and lift it into the reach space. Gabriel and Hannah solved entirely different biomechanical problems to arrive at the same functional motor milestone—proving that there is no singular “reaching program” in the infant mind.
8.2 Kinematic, Electromyographic (EMG), and Kinetic Analyses
Thelen’s empirical methodologies were defined by their rigorous, multi-level biomechanical sophistication. Rather than relying on qualitative human observation or simple behavioral checklists, Thelen constructed advanced movement analysis laboratories that simultaneously captured the neural, kinematic, and kinetic levels of action. She integrated high-speed three-dimensional optoelectronic motion capture systems (using active light-emitting diodes or retro-reflective markers) with continuous, synchronized surface electromyography (EMG) and force-plate kinetic platforms.
This multi-level instrumentation allowed Thelen to peer beneath the superficial appearance of movement to calculate its underlying dynamic architecture. By feeding 3D marker coordinates into advanced inverse dynamics equations, her team could calculate continuous changes in joint angles, angular velocities, angular accelerations, and critically, the separation between active muscular torques (the forces generated by contracting muscles) and passive, interactive torques (the mechanical forces generated by the movement of adjacent limb segments, gravity, and inertial resistance). Her EMG analyses revealed the fine-grained firing patterns of agonist-antagonist muscle pairs, documenting the historical ontogenetic transition from primitive, rigid muscle co-contraction (stiffening the entire joint) to nuanced, energy-efficient reciprocal activation patterns.
This empirical convergence was foundational to establishing the veracity of dynamic systems. Thelen proved that superficial kinematic similarities could conceal radically divergent kinetic strategies, and conversely, that fluctuating, messy kinematic outputs were frequently driven by the emergence of remarkably elegant, underlying mechanical torque optimizations. By demonstrating that the nervous system actively exploits passive, interactive biomechanical forces—rather than having to compute and generate every ounce of force through internal metabolic energy—Thelen validated Bernstein’s classic biomechanical assertions using cutting-edge late-twentieth-century laboratory instrumentation.
8.3 Longitudinal Intra-Individual Observation Strategies
A non-negotiable methodological tenet of Thelen’s paradigm was the absolute priority of intra-individual analysis over aggregate group averaging. In standard developmental psychology, researchers routinely average the performance metrics of twenty, fifty, or one hundred infants to produce a smooth, normative developmental curve. Thelen vehemently argued that this practice commits a profound ecological fallacy: the “average infant” represented by the aggregate curve does not exist in the physical world. Group averaging mechanically washes out the sudden bifurcations, critical fluctuations, idiosyncratic strategies, and non-linear jumps that characterize every real, individual developmental trajectory.
Thelen’s methodological protocols demanded dense sampling intervals deliberately clustered around anticipated phase transition windows. If an infant showed initial signs of destabilization in their pre-reaching or pre-walking behavior, the frequency of laboratory observations was escalated to multiple times per week. Concurrently, Thelen maintained rigorous ecological validity during these high-tech laboratory experiments. Rather than strapping infants into artificial, highly constrained laboratory rigs that restricted natural movement degrees of freedom, she permitted infants to sit, lie, or move freely on their caregivers’ laps, interacting with real objects, toys, and social agents.
To mathematically detect the emergence of novel attractors and state shifts within these continuous, dense data sets, Thelen’s laboratory developed quantitative criteria rooted in dynamic systems metrics. Researchers tracked the standard deviation of order parameters, computed phase portraits (plotting position versus velocity in phase space), and calculated spatial-temporal consistency matrices across dozens of consecutive movement trials. A behavioral state shift was quantitatively operationalized not by an arbitrary calendar age, but by the mathematical collapse of an old attractor’s stability metrics, the temporary explosion of exploratory variance, and the subsequent consolidation of a new, low-variability topological attractor basin.
9. Embodied Cognition: Bridging Motor Action and Cognitive Development
9.1 Dissolution of the Cartesian Divide: Action as Cognition
Perhaps the most far-reaching intellectual legacy of Esther Thelen is her decisive role in dismantling the traditional Cartesian divide that had bifurcated Western philosophy, psychology, and cognitive science for centuries. The classical Cartesian view enforced an absolute apartheid between the “mental” and the “physical.” Cognition was conceived as an insular, disembodied realm of abstract computations, logical propositions, and centralized mental representations, while the motor system was relegated to the status of a mere physical executioner—a set of mechanical cables and pulleys carrying out the commands issued by the detached intellectual sovereign.
Thelen, alongside her co-author Linda B. Smith in their seminal 1994 monograph A Dynamic Systems Approach to the Development of Cognition and Action, argued that this Cartesian separation is fundamentally incoherent. They asserted the primacy of embodied, embedded, and enacted cognition: thinking does not occur in a vacuum prior to action; thinking is inherently, inextricably grounded in real-time bodily interaction with the physical and social environment. The physical architecture of our bodies—the fact that we have forward-facing eyes, binocular vision, bilateral symmetry, specific musculoskeletal lever arms, and must constantly negotiate Earth’s gravitational acceleration—fundamentally constitutes the very nature of human thought, categorical abstraction, and conceptual reason.
Within this dynamic ontology, perception, action, and cognition do not operate as a sequential, feed-forward tripartite assembly line (Perceive $\rightarrow$ Think $\rightarrow$ Act). Instead, they are locked in a continuous, non-linear, recurrent loop where every motor action alters perceptual input, and perceptual input immediately alters the dynamic landscape of the acting body. The infant does not first build an abstract intellectual model of the world and then choose to act within it; rather, the infant’s continuous, exploratory physical interactions with the world are the very medium through which cognitive architectures are forged. To Thelen, the mind is not an internal mirror reflecting an external reality; it is an emergent dynamic property of an embodied organism actively making its way through the physical world.
9.2 Re-interpreting Piaget’s A-not-B Error via Dynamic Field Theory
To provide a definitive empirical challenge to disembodied cognitive theories, Thelen, Linda Smith, and their collaborators turned their attention to the crown jewel of cognitive developmental psychology: Jean Piaget’s classic A-not-B error. In this ubiquitous developmental task, an infant of approximately 8 to 10 months watches an experimenter hide a desirable toy beneath Box A. The infant reaches for and successfully retrieves the toy at Location A across several consecutive trials. Then, in the crucial test trial, the experimenter slowly and conspicuously hides the identical toy beneath Box B, in full view of the infant. Despite clearly seeing the toy placed at B, infants between 8 and 10 months overwhelmingly reach back to Location A—committing the perseverative A-not-B error.
For decades, Piaget and his cognitive successors interpreted this perseverative error as a profound, conceptual mental deficit regarding the child’s internal representational understanding of physical reality: specifically, an incomplete realization of object permanence. Later cognitive neuroscientists attributed the error to the structural immaturity of the infant’s prefrontal cortex, arguing that the infant lacks the executive working memory capacity to update the spatial location or the inhibitory control to suppress a prepotent response. In all these explanations, the error was treated as a deficit in abstract cognitive representation residing within the infant’s brain.
Thelen and Smith completely revolutionized the interpretation of the A-not-B error by demonstrating that it is not an error of internal cognitive representation, but an emergent property of a Dynamic Neural Field (DNF) integrating real-time visual attention, motor planning, and immediate bodily memory. In a series of brilliant, paradigm-shattering experiments, Thelen demonstrated that one could completely abolish or induce the A-not-B error without changing the infant’s conceptual knowledge of the toy at all, simply by manipulating low-level physical constraints. By altering the infant’s physical posture between trials—placing an infant who had reached for Location A while sitting into a standing position during the test trial—the A-not-B error instantly vanished, and infants reached accurately to Box B. Adding small wrist weights, altering the visual distinctiveness of the reaching field, or changing the infant’s immediate movement history similarly altered the behavioral outcome.
The Dynamic Neural Field model formulated by Gregor Schöner, Thelen, and Smith demonstrated that the infant’s reach is determined by continuous, non-linear activation dynamics across a motor-spatial parameter field. Every previous reach to Location A creates a lingering, localized dynamic memory trace in the physical system. When the toy is hidden at B, the sensory visual input at B must compete directly with the powerful, lingering motor activation trace at A. If the delay between hiding and reaching is long enough for the transient visual signal to decay, the deep, persistent motor attractor basin at Location A self-organizes the physical trajectory of the reaching arm. The A-not-B error is thus not a grand philosophical failure to comprehend the permanence of physical matter; it is a soft-assembled dynamic consequence of real-time sensorimotor dynamics, bodily history, and immediate perceptual cues competing within a non-linear field.
9.3 Perceptual-Motor Exploratory Loops in Spatial Understanding
The realization that motor actions actively structure cognitive development opened profound new vistas into the study of infant spatial awareness and perceptual understanding. Independent locomotion is not merely a convenient way for an infant to transport their brain from one room to another; it is a profound catalyst for catastrophic cognitive reorganization. Thelen’s work illuminated how the self-generated perceptual consequences of independent motor actions serve as the fundamental developmental engine driving spatial cognition.
This dynamic reality was vividly demonstrated in investigations utilizing the classic visual cliff paradigm, originally designed by Eleanor Gibson and Richard Walk. Historically, it was assumed that an infant’s avoidance of the deep, glass-covered drop-off on the visual cliff was an innate, hardwired biological fear that matured automatically on a preset timeline. However, developmental studies inspired by dynamic systems (notably those conducted by Joseph Campos, Karen Adolph, and David Anderson) revealed that avoidance of the cliff is intimately coupled with real-time locomotor experience. Pre-crawling infants of the same chronological age do not exhibit physiological fear (such as heart rate acceleration) when placed above the deep drop; but within weeks of acquiring independent quadrupedal crawling, infants display acute autonomic distress and resolutely refuse to traverse the transparent surface.
The generative mechanism underlying this cognitive shift is the continuous perceptual-motor exploratory loop. As an infant engages in self-generated locomotion, they are subjected to continuous optical flow—the rapid, visual streaming of ambient environmental features across the retina. Through thousands of meters of crawling, the infant’s nervous system learns to dynamically integrate this optical flow with internal vestibular and proprioceptive signals to calibrate postural equilibrium. When confronted with the visual cliff, the optical flow beneath the infant suddenly vanishes, depriving the dynamic postural system of its essential stabilizing feedback. The infant perceives the sudden loss of an affordance for support. As Karen Adolph’s meticulous slope and gap navigation studies further established, infants do not learn a single, abstract cognitive concept of “height” or “danger.” Each time an infant transitions to a radically new motor posture—from sitting to crawling, and later from crawling to cruising and walking—they must systematically re-explore and re-calibrate their perceptual affordances from scratch within the dynamic constraints of that specific somatic orientation.
10. Clinical Applications: Pediatric Physical Therapy and Developmental Intervention
10.1 Deconstruction of Traditional Interventions (e.g., NDT)
The paradigm shift engineered by Esther Thelen had profound, revolutionary ramifications for the clinical disciplines of pediatric physical therapy, occupational therapy, and neurodevelopmental rehabilitation. For the greater part of the twentieth century, pediatric intervention for neurodevelopmental disorders—such as cerebral palsy, spina bifida, and traumatic brain injury—was dominated by prescriptive, hierarchical models, most notably classical Neurodevelopmental Treatment (NDT), developed by Karel and Berta Bobath.
Classical NDT was explicitly grounded in the neuromaturational paradigm. It operated on the foundational premise that the motor deficits observed in cerebral palsy were direct, inescapable manifestations of a damaged central nervous system failing to inhibit pathological, lower-level primitive reflexes. Consequently, the traditional clinical objective was top-down normalization: the therapist acted as a direct controller, using passive manual facilitation to physically guide the child’s limbs through “normal” movement patterns while actively suppressing and inhibiting “abnormal” or “pathological” reflex behaviors. Spontaneous, atypical movement strategies generated by the child were viewed as inherently detrimental and were actively extinguished.
Dynamic Systems Theory systematically dismantled the theoretical foundations of these traditional, passive approaches. Thelen revealed that the atypical movement patterns observed in children with neurodevelopmental disabilities are not direct, unmediated readouts of an internal brain lesion. Instead, they are the child’s creative, self-organized adaptations—functional dynamic solutions soft-assembled to solve real-world behavioral tasks in the face of profound, atypical organismic constraints (such as asymmetric spasticity, compromised joint geometry, or pathologically diminished muscular force). Suppressing these functional, albeit atypical, compensatory patterns through passive manual manipulation without altering the underlying physical constraints frequently stripped the child of their only viable mechanical strategy for navigating their world. Prescriptive, repetitive, and context-free motor drills failed to achieve real-world clinical translation because they treated the nervous system in isolation from the dynamic triad of organism, environment, and task.
10.2 Dynamic Systems-Based Assessment Strategies
The clinical translation of Dynamic Systems Theory demanded an entirely novel diagnostic and assessment methodology. Rather than calculating a child’s chronological deviation from a normative milestone chart or measuring isolated passive range of motion and static reflex integrity, dynamic systems-based assessment requires a comprehensive, holistic evaluation of the child’s entire multi-subsystem ecology.
Under this dynamic clinical paradigm, the therapist’s primary diagnostic objective is to identify the individual child’s specific rate-limiting subsystems. In a child with delayed upright locomotion, the clinician does not assume an abstract “brain delay.” Instead, the therapist systematically evaluates the heterochronic components: Is the rate limiter a deficit in dynamic postural anticipatory adjustments? Is it insufficient extensor force production in the plantar flexors? Is it compromised sensory-perceptual mapping of optical flow? Or is it an affective-motivational fear of falling born of previous mechanical instability? Identifying the precise rate limiter allows for targeted, efficient clinical intervention rather than diffuse, generalized physical therapy.
Furthermore, dynamic assessment involves comprehensive attractor mapping. The clinician evaluates the depth, stability, and rigidity of the child’s existing motor regimes by systematically applying graded environmental and mechanical perturbations. A child whose motor system is trapped in an excessively deep, hyper-stable, and rigid attractor basin (such as the persistent, stereotyped extensor thrusting seen in severe spastic diplegia) requires interventions designed to destabilize this rigid regime and introduce healthy movement variability. Conversely, a child exhibiting excessively shallow, hyper-variable, and erratic motor states (such as in severe cerebellar ataxia) requires therapeutic environmental constraints that scaffold and stabilize emerging coordination patterns. By identifying naturally occurring windows of behavioral instability—where the child’s motor system is already fluctuating—clinicians can time their interventions to exploit these critical developmental bifurcation windows.
10.3 Contemporary Interventions: Task-Specific and Treadmill Training
The practical triumph of Thelen’s clinical legacy is exemplified by the global proliferation of contemporary, evidence-based pediatric interventions, most notably Task-Specific Training and Body-Weight Supported Treadmill Training (BWSTT). Directly informed by Thelen’s seminal infant treadmill investigations, researchers such as Beverly Ulrich demonstrated that infants with Down syndrome—who typically exhibit pronounced hypotonia, ligamentous laxity, and severe walking delays of up to a full year compared to typically developing peers—could achieve independent bipedal walking months earlier through early, dynamic treadmill intervention.
BWSTT does not passively move the infant’s legs; rather, it places the infant in a functional, upright, weight-bearing posture supported by a harness over a moving treadmill belt. The dynamic translation of the belt supplies the external mechanical control parameter that pulls the limb into hip extension, triggering intrinsic proprioceptive feedback loops and engaging the spinal central pattern generators to self-organize alternating reciprocal stepping. The mechanical task constraint itself pulls the motor behavior out of the biological system. Over longitudinal training, this task-specific dynamic intervention progressively hypertrophies lower-extremity muscle fibers, refines dynamic postural balancing mechanisms, and drives neuroplastic synaptic consolidation within descending pathways, successfully resolving the child’s rate-limiting bottlenecks.
In modern dynamic interventions, the fundamental role of the pediatric physical therapist has undergone a radical transformation: the therapist is no longer an authoritative “controller” of the child’s movement, but an environment and task architect. Clinical intervention focuses on manipulating external constraints: modifying support surfaces, introducing novel affordances, scaling object masses and geometries, and engineering functional physical problems that the child must actively solve. Crucially, the therapist deliberately tolerates, and even encourages, movement errors and exploratory variability. It is through active somatic exploration, dynamic trial-and-error, and the physical experience of overcoming environmental perturbations that the child’s open dynamic system naturally self-organizes resilient, functional, and permanent motor solutions.
11. Contemporary Extensions: Robotics, Computational Modeling, and Motor Control
11.1 Evolutionary and Morphological Robotics
The profound conceptual implications of Esther Thelen’s Dynamic Systems Theory resonated far beyond the borders of developmental psychology, directly igniting a revolution in autonomous robotics, mechanical engineering, and artificial intelligence. Throughout the late twentieth century, mainstream robotics was dominated by the classic computational paradigm: humanoid bipedal robots (such as Honda’s early ASIMO) were designed as massive, rigid computing machines. Every single joint degree of freedom was packed with heavy servomotors governed by ultra-high-speed microprocessors executing complex inverse-kinematic mathematical equations in real time to calculate joint trajectories before a single step was taken. These computational robots were excessively energy-hungry, computationally fragile, and notoriously incapable of adapting to minor real-world floor perturbations.
In direct response to this computational impasse, roboticists such as Tad McGeer and later Rodney Brooks and Rolf Pfeifer embraced Thelen’s principles of soft-assembly and self-organization, pioneering the fields of passive dynamic walkers and morphological computation. Passive dynamic walkers are fully unpowered, mechanical bipedal machines possessing no motors, no microprocessors, no sensors, and no electrical control systems whatsoever. When placed at the top of an exceptionally shallow downward incline, these machines, utilizing nothing more than gravity, limb inertia, and the passive pendulum-like mechanics of their joint geometry, walk down the slope with an astonishingly fluid, human-like, stable bipedal gait. The locomotion is not computed in an internal processing unit; it is an emergent property of the machine’s physical morphology interacting dynamically with the gravitational field.
Morphological computation demonstrates Thelen’s fundamental assertion: the physical body itself performs computation. By engineering compliance, elasticity, and specific geometric architectures directly into the physical materials of the robot’s limbs (soft robotics), the burden of behavioral control is effectively outsourced to the physics of the body. In developmental robotics, autonomous agents are now routinely constructed to mimic infant ontogeny: these robotic systems begin with uncalibrated, chaotic, exploratory actuator firings (simulating infant motor babbling and rhythmic stereotypies), slowly discovering the non-linear dynamics of their own mechanical structures and soft-assembling stable reach, grasp, and walking attractors through continuous, self-organizing environmental interaction.
11.2 Dynamic Neural Field (DNF) Architectures
In the domain of computational neuroscience, Thelen’s dynamic principles were formalized into the powerful mathematical framework of Dynamic Neural Field (DNF) Theory, spearheaded by theoretical physicists and cognitive scientists including Gregor Schöner, John Spencer, and Wolfram Erlhagen. Moving decisively beyond static artificial neural networks, DNF models are mathematically grounded in continuous, non-linear integro-differential equations originally formulated by Shun-ichi Amari to describe the spatio-temporal activation dynamics of cortical neural populations.
A Dynamic Neural Field mathematically simulates the continuous distribution of neural activation over metric behavioral parameter spaces (such as spatial location, movement direction, or reaching velocity). The governing differential equation of a canonical dynamic neural field can be expressed as:
$$\tau \frac{\partial u(x,t)}{\partial t} = -u(x,t) + \int w(x – x’) f(u(x’,t)) dx’ + S(x,t) + h$$
In this non-linear formulation, $tau$ represents the characteristic relaxation time scale of the system, $u(x,t)$ denotes the instantaneous activation at metric position $x$ and time $t$, and $-u(x,t)$ signifies the intrinsic, continuous exponential decay toward the resting baseline level defined by the negative parameter $h$. The integral term encapsulates the foundational dynamic architecture of local excitation and lateral inhibition: the interaction kernel $w(x – x’)$ specifies that adjacent neural sites exhibiting similar behavioral parameter values mutually excite one another, while distant, competing metric sites exert powerful lateral inhibition through the sigmoidal threshold firing function $f(u)$. The term $S(x,t)$ incorporates the influx of real-time, external sensory and task inputs.
DNF architectures have provided rigorous, quantitatively validated computational models of how infants solve complex perceptual-motor problems in real time. These non-linear fields flawlessly replicate the empirical transitions observed in infant reaching, saccadic eye movements, spatial working memory, and the infamous A-not-B error. By demonstrating how transient, continuous sensory inputs interact with persistent self-sustaining activation peaks (the neural equivalent of working memory) and lingering resting-level shifts (the neural trace of behavioral history), DNF models successfully bridge the historic chasm between real-time neural population dynamics and the macroscopic execution of embodied physical trajectories.
11.3 Complex Systems Modeling of Motor Variability
Modern human movement science has expanded Thelen’s conceptualization of functional variability through the integration of complex systems modeling and non-linear time series mathematics. In contemporary biomechanics laboratories, infant movement data captured via wearable inertial measurement units (IMUs) and 3D kinematics are no longer analyzed using simple linear statistical metrics (such as mean and standard deviation). Instead, researchers deploy sophisticated non-linear analytical algorithms, including Sample Entropy (SampEn), Recurrence Quantification Analysis (RQA), and the calculation of the Largest Lyapunov Exponent ($\lambda_1$).
These advanced computational metrics quantify the exact structural complexity, predictability, and local dynamic stability of infant motor trajectories across time. Sample entropy allows researchers to measure the degree of regularity versus complexity within a physiological signal; healthy, typically developing infant motor output is characterized by a specific, optimal degree of entropy, reflecting an adaptable, flexible biological system poised dynamically between rigid, predictable order and chaotic, unstructured noise. The Largest Lyapunov Exponent quantifies the rate at which an infant’s movement trajectory diverges following an infinitesimal perturbation, providing an empirical mathematical measurement of the local dynamic stability of the underlying attractor regime.
Crucially, this non-linear analytical revolution has unlocked breakthrough capabilities in early clinical diagnostics. Cutting-edge machine learning algorithms, trained on the non-linear dynamic time-series signatures of spontaneous infant movements (such as general movements assessed in the Prechtl qualitative assessment), can now reliably classify and predict atypical neurodevelopmental trajectories—such as cerebral palsy and autism spectrum disorder—months before traditional normative clinical milestones fail to appear. By detecting subtle, systemic alterations in the mathematical structure of infant exploratory variability, complex systems modeling translates Esther Thelen’s dynamic axioms into predictive clinical technologies.
12. Epistemological Legacy and Future Directions in Developmental Science
12.1 The Paradigm Shift: From Genetic Blueprints to Relational Developmental Systems
The intellectual magnitude of Esther Thelen’s contribution to science lies in her definitive role in executing a permanent epistemological paradigm shift across developmental science. Prior to her theoretical work, the biological and psychological sciences remained hopelessly entangled in the sterile, dualistic debates of the nineteenth and twentieth centuries: Nature versus Nurture, Innate versus Acquired, Maturation versus Learning. Thelen exposed the fundamental intellectual bankruptcy of these dichotomies. In a self-organizing dynamic system, the dichotomy between biology and environment dissolves: behavior is an emergent property created at their intersection.
Thelen’s dynamic systems framework served as the foundational vanguard for contemporary Relational Developmental Systems (RDS) theory and Developmental Systems Theory (DST), championed by philosophers and developmental theorists such as Richard Lerner, Susan Oyama, and Paul Griffiths. RDS explicitly rejects all forms of biological or genetic essentialism. The genome is no longer conceptualized as an executive “blueprint” or “program” that dictates the structural unfolding of the organism. Rather, the genome is recognized as an open, responsive, and thoroughly integrated subsystem among many others. In accordance with modern molecular epigenetics, gene expression is continuously turned on, modulated, or suppressed by the real-time physical, mechanical, chemical, and behavioral activities of the organism within its ecological environment.
Ontogeny is thus recast as an open-ended, non-deterministic developmental trajectory of continuous morphological and behavioral adaptation. Human development has no predetermined endpoint, no transcendent genetic goal, and no central teleological governor. The structural form of the developing human being is perpetually co-constructed through reciprocal, bidirectional causality operating across every biological scale—from molecular transcription factors and cellular metabolisms, to musculoskeletal biomechanics, real-time perceptual-motor exploratory loops, and sociocultural ecologies. Thelen permanently shifted the foundational question of developmental science from the obsolete inquiry of what is innate versus learned, to the generative, systems-level investigation of how coordinated biological order emerges from relational complexity.
12.2 Unresolved Questions and Theoretical Critiques
Despite its theoretical dominance and empirical victories, Dynamic Systems Theory has faced substantive criticisms, conceptual debates, and enduring unresolved challenges within developmental psychology and theoretical neuroscience. A primary theoretical challenge revolves around the immense mathematical difficulty of formalizing comprehensive, multi-scale computational models that seamlessly span from the microscopic firing rates of millions of individual neurons, through continuous muscular torque kinetics, up to macroscopic sociocultural behaviors. While dynamic systems has generated exquisite mathematical models of discrete, localized tasks (such as bipedal stepping or reaching), a unified, predictive mathematical framework capable of capturing the entirety of human ontogeny remains an aspirational horizon.
Furthermore, persistent methodological critiques have targeted the operationalization and precise empirical measurement of dynamic constructs in human subjects. While concepts such as “attractor depth,” “potential landscapes,” and “critical fluctuations” are mathematically crisp in controlled physical systems (like fluid dynamics or coupled mechanical oscillators), their rigorous quantification in squirming, non-stationary, and highly fatigue-prone human infants is notoriously fraught with measurement noise. Skeptics have questioned whether the dynamic vocabulary of “attractors” and “bifurcations” occasionally serves as an elegant, sophisticated descriptive metaphor rather than a truly predictive, falsifiable mechanical model.
Another profound theoretical battleground involves the outright rejection of mental representation. Radical dynamicists, following Thelen’s most provocative assertions, argued that the concept of internal representation should be completely banished from cognitive science—asserting that the physical body and the environment, through continuous, coupled sensorimotor action, serve as their own best representations. However, mainstream cognitive neuroscientists and evolutionary psychologists counter that this radical anti-representational stance struggles to adequately explain higher-order, offline human cognition: tasks such as abstract counterfactual reasoning, long-term mental time travel, retrospective memory consolidation, and symbolic linguistic creativity, which occur explicitly in the complete absence of immediate perceptual-motor action. Reconciling the self-organizing elegance of rapid, online soft-assembly with the undeniable existence of enduring, offline cognitive architectures remains one of the greatest uncompleted theoretical tasks of modern cognitive science.
12.3 The Continuing Evolution of Dynamic Systems in the 21st Century
As developmental science navigates the twenty-first century, Esther Thelen’s dynamic systems vision is experiencing an unprecedented, technology-driven renaissance. The convergence of miniaturized, wearable biosensors, non-invasive portable functional neuroimaging (such as functional near-infrared spectroscopy, fNIRS), and Ecological Momentary Assessment (EMA) protocols has finally provided the technological apparatus required to realize Thelen’s microgenetic vision at a planetary scale. Researchers can now continuously track the real-time, high-frequency movements, heart rates, gaze trajectories, and neural activations of infants within their natural home ecologies across months of ontogenetic time, generating massive, multi-dimensional time series datasets analyzed via advanced complex systems algorithms.
Moreover, the core dynamic principles of soft-assembly, rate-limiters, and non-linear phase shifts have transcended their original biomechanical confines, expanding into universal theoretical paradigms across every domain of human development. Contemporary developmental scientists deploy Thelen’s dynamic systems apparatus to model the non-linear emergence of infant emotional regulation, the sudden, explosive phase shifts of vocabulary acquisition in early childhood, the turbulent attractor reorganizations of adolescent social dynamics, and the non-linear breakdowns of cognitive control in senescent neurodegenerative disorders. Thelen’s theoretical architecture has proved to be a fractal, scale-invariant epistemology.
Ultimately, Esther Thelen’s enduring vision stands as the bedrock of the contemporary 4E Cognitive Science movement—the unified philosophical and empirical consensus that human cognition is profoundly Embodied, Embedded, Enacted, and Extended. By liberating the developing child from the mechanistic confines of pre-programmed genetic scripts and disembodied computational algorithms, Thelen restored the living, physical body to its rightful place at the center of the developmental universe. She taught the world to view the developing infant not as a passive vessel waiting for biological instructions, but as an active, exploratory artist of physics—an embodied, self-organizing dynamic system continually forging order, competence, and mind from the turbulent, beautiful physical realities of our world.
Conclusion
The Dynamic Systems Theory of motor development, as formulated and empirically substantiated by Esther Thelen, represents one of the most profound paradigms shifts in the history of developmental science. By integrating Nikolai Bernstein’s biomechanics, James J. Gibson’s ecological psychology, and Ilya Prigogine’s non-equilibrium thermodynamics, Thelen effectively shattered the century-long hegemony of neuromaturational determinism and computational information processing. Her work demonstrated that human motor behavior is not a pre-programmed, top-down execution directed by an imperial central nervous system, nor is it a standardized sequence of invariant, genetically determined milestones. Instead, motor coordination is fundamentally an emergent, soft-assembled property arising dynamically and opportunistically from the real-time, non-linear self-organization of the infant’s heterochronically maturing body, the specific teleological demands of the task, and the continuous physical forces of the ambient environment.
Through her legendary empirical investigations—most vividly exemplified by the deconstruction of the disappearing stepping reflex via water submersion and motorized treadmills, and the radical reinterpretation of Piaget’s A-not-B error through Dynamic Neural Field modeling—Thelen proved that development progresses through non-linear phase shifts, bifurcations, and the critical exploration of movement variability. In doing so, she permanently eradicated the Cartesian divide that artificially separated mind from body and action from cognition. Thelen repositioned the human body as an active, physical participant in the generation of its own intelligence, establishing that thinking itself is grounded in sensorimotor exploration. Today, her dynamic framework continues to inspire revolutionary breakthroughs across pediatric physical therapy, autonomous and soft robotics, computational neuroscience, and embodied cognitive science, ensuring that her legacy remains a luminous, guiding beacon for all who seek to understand the exquisite, emergent complexity of human life.
References
- Adolph, K. E., & Hoch, J. E. (2019). Motor development: Embodied, embedded, enculturated, and enabling. Annual Review of Psychology, 70, 141–164. https://doi.org/10.1146/annurev-psych-010418-102836
- Amari, S. (1977). Dynamics of pattern formation in lateral-inhibition type neural fields. Biological Cybernetics, 27(2), 77–87. https://doi.org/10.1007/BF00337259
- Bernstein, N. A. (1967). The co-ordination and regulation of movements. Pergamon Press.
- Campos, J. J., Anderson, D. I., Barbu-Roth, M. A., Hubbard, E. M., Hertenstein, M. J., & Witherington, D. (2000). Travel broadens the mind. Infancy, 1(2), 149–219. https://doi.org/10.1207/S15327078IN0102_1
- Edelman, G. M. (1987). Neural Darwinism: The theory of neuronal group selection. Basic Books.
- Gesell, A. (1946). The ontogenesis of infant behavior. In L. Carmichael (Ed.), Manual of child psychology (pp. 295–331). John Wiley & Sons.
- Gibson, J. J. (1979). The ecological approach to visual perception. Houghton Mifflin.
- Haken, H. (1983). Synergetics: An introduction. Nonequilibrium phase transitions and self-organization in physics, chemistry, and biology (3rd ed.). Springer-Verlag. https://doi.org/10.1007/978-3-642-88338-5
- Kelso, J. A. S. (1995). Dynamic patterns: The self-organization of brain and behavior. MIT Press.
- McGraw, M. B. (1943). The neuromuscular maturation of the human infant. Columbia University Press.
- Newell, K. M. (1986). Constraints on the development of coordination. In M. G. Wade & H. T. A. Whiting (Eds.), Motor development in children: Aspects of coordination and control (pp. 341–360). Martinus Nijhoff. https://doi.org/10.1007/978-94-009-4460-2_19
- Pfeifer, R., & Bongard, J. (2006). How the body shapes the way we think: A new view of intelligence. MIT Press. https://doi.org/10.7551/mitpress/3585.001.0001
- Prigogine, I., & Stengers, I. (1984). Order out of chaos: Man’s new dialogue with nature. Bantam Books.
- Schmidt, R. A. (1975). A schema theory of discrete motor skill learning. Psychological Review, 82(4), 225–260. https://doi.org/10.1037/h0076770
- Schöner, G., & Spencer, J. P. (Eds.). (2015). Dynamic thinking: A primer on dynamic field theory. Oxford University Press. https://doi.org/10.1093/acprof:oso/9780199300563.001.0001
- Thelen, E. (1979). Rhythmical stereotypies in normal human infants. Animal Behaviour, 27(3), 699–715. https://doi.org/10.1016/0003-3472(79)90006-X
- Thelen, E. (1984). Learning to walk: Ecological demands and phylogenetic constraints. Advances in Infancy Research, 3, 213–260.
- Thelen, E., & Fisher, D. M. (1982). Newborn stepping: An explanation for a “disappearing” reflex. Developmental Psychology, 18(5), 760–775. https://doi.org/10.1037/0012-1649.18.5.760
- Thelen, E., Fisher, D. M., & Ridley-Johnson, R. (1984). The relationship between physical growth and a newborn reflex. Infant Behavior and Development, 7(4), 479–493. https://doi.org/10.1016/S0163-6383(84)80007-7
- Thelen, E., & Smith, L. B. (1994). A dynamic systems approach to the development of cognition and action. MIT Press. https://doi.org/10.7551/mitpress/2534.001.0001
- Thelen, E., Corbetta, D., Kamm, K., Spencer, J. P., Schneider, K., & Zernicke, R. F. (1993). The transition to reaching: Mapping intention and intrinsic dynamics. Monographs of the Society for Research in Child Development, 58(3), i–79. https://doi.org/10.2307/1166074
- Thelen, E., Schöner, G., Scheier, C., & Smith, L. B. (2001). The dynamics of embodiment: A dynamic field theory of infant perseverative reaching. Behavioral and Brain Sciences, 24(1), 1–34. https://doi.org/10.1017/S0140525X01003910
- Ulrich, B. D., Ulrich, D. A., Angulo-Kinzler, R. M., & Chapman, D. D. (2001). Sensitivity of infants with and without Down syndrome to treadmill perturbations. Journal of Motor Behavior, 33(4), 347–360. https://doi.org/10.1080/00222890109601918
- Waddington, C. H. (1957). The strategy of the genes: A discussion of some aspects of theoretical biology. Allen & Unwin.