Sensorimotor Thought & Infant Memory

Where it beginsPiaget's sensorimotor stage and its six substages

For roughly the first two years of life, according to Piaget, thought is action. The infant knows the world by looking, mouthing, grasping, and moving through it, and has not yet acquired the capacity to hold that world in mind as a symbol. Piaget called this the sensorimotor stage, and mapping it occupied two of his most detailed books (Piaget, 1952, 1954).

The engine of sensorimotor development, in Piaget's account, is the scheme — an organized pattern of action, such as sucking, grasping, or looking, that the infant applies to the world and gradually revises. New experience is taken in through assimilation (fitting an event into an existing scheme, as when a baby sucks a novel toy the way she sucks a nipple) and forces accommodation (revising the scheme to fit reality, as when she adjusts her mouth to a cup). The back-and-forth between the two is adaptation, and it drives the infant from reflex to representation across six substages Piaget observed closely, in part by watching his own three children.

The first substage, reflexive schemes (birth to about one month), is the raw starting material — the sucking, grasping, and orienting reflexes the newborn brings to the world, exercised and gradually made more efficient. In the second substage, primary circular reactions (roughly one to four months), the infant begins to repeat pleasurable actions centered on her own body, such as bringing a thumb to the mouth again and again; a "circular reaction" is simply a chance action that produces a satisfying result and so gets repeated. By the third substage, secondary circular reactions (about four to eight months), that repetition turns outward onto the environment — a baby who accidentally makes a hanging rattle jingle will kick or swipe to make it happen again, an early sign of interest in the effects her actions have on the world. This outward-turning is exactly the tendency Rovee-Collier's mobile studies would later harness.

The fourth substage, the coordination of secondary circular reactions (roughly eight to twelve months), is a genuine turning point: the infant now combines schemes into deliberate, goal-directed sequences, pushing an obstacle aside to grasp a toy behind it. This is the first appearance of true means–end behavior, the first reliable search for a fully hidden object — and, not coincidentally, the age at which the famous A-not-B error appears. In the fifth substage, tertiary circular reactions (about twelve to eighteen months), Piaget's "little scientist" emerges: the toddler now varies her actions deliberately, dropping a spoon from different heights to see what happens, exploring novelty through active trial and error rather than mere repetition. Finally, in the sixth substage, the beginning of representational thought (roughly eighteen to twenty-four months), the child becomes able to hold the world in mind as a mental symbol — solving problems by internal combination rather than overt fumbling, imitating an action seen hours earlier, and, by Piaget's timetable, mastering object permanence in full. It is precisely this endpoint — representation arriving only near the second birthday — that the modern evidence would push much earlier.

The big correctionInfants know more, earlier, than Piaget claimed

Piaget inferred what babies understood from what they could do — reach, search, manipulate. But motor skill lags behind knowledge, so he systematically underestimated infants. A clever method flipped this: measure what babies look at, not what they can grab.

The logic rests on a robust regularity: infants look longer at events that are novel or that violate what they expect, a tendency that lets researchers use gaze as a read-out of knowledge no reach could reveal. In the violation-of-expectation paradigm, an infant is first habituated to an ordinary event until looking time declines, then shown either a possible outcome or an impossible one that breaks a physical rule; reliably longer looking at the impossible event implies the infant represented the rule and detected its violation. Because the method asks nothing of the hands, it decouples what an infant knows from what an infant can do — precisely the confound that led Piaget's reaching-based tasks to date object permanence so late.

The classic demonstration used a hidden object. Five-month-olds who watched a screen rotate up and back like a drawbridge looked longer when it appeared to rotate through the space where a solid box had been hidden than when it stopped against the box, indicating they represented the occluded box as still present and still solid (Baillargeon, Spelke, & Wasserman, 1985). A refined version pushed the finding younger still, establishing object permanence in infants as young as three-and-a-half months (Baillargeon, 1987). On Piaget's account such infants should have no concept of the hidden object at all; on the looking-time evidence, they clearly do — they simply cannot yet organize a coordinated reach to prove it.

Pre-stocked, not blankCore knowledge and where infant understanding comes from

If object permanence is present in the first months of life, it cannot have been built slowly through a year of reaching and searching, as Piaget supposed. That is the pressure the looking-time results put on his theory, and core knowledge theory is the leading answer to what should replace it. On this view, evolution has equipped the infant mind with a handful of specialized, early-emerging systems of knowledge that make the surrounding world interpretable from the start (Spelke, Breinlinger, Macomber, & Jacobson, 1992; Spelke & Kinzler, 2007).

The best-studied system concerns objects. Long before they can talk, infants treat objects as cohesive (they hang together as connected wholes), continuous (they trace exactly one connected path through space and time rather than blinking in and out of existence), and solid (two objects cannot pass through the same space at once). In a series of experiments, infants who saw a ball drop behind a screen expected it to come to rest on top of a newly inserted shelf rather than passing magically through it, and looked longer when the laws of continuity and solidity were violated (Spelke et al., 1992). Notably, these expectations concern how objects persist and move more than how they behave under gravity or inertia, which infants learn considerably later — evidence that core knowledge is a specific, principled starting point rather than a general-purpose physics engine.

Parallel systems have been proposed for number (an approximate sense of quantity that lets infants tell eight dots from sixteen), for agents (self-propelled entities that pursue goals efficiently, distinguished early from inert objects), and for space and geometry (reorienting by the shape of an enclosure, a capacity shared with other animals). Whether these amount to genuinely innate "knowledge" or to powerful early-maturing learning biases remains debated, and core knowledge is not a return to a fixed nativism: the systems are described as limited, sometimes error-prone starting assumptions that development then elaborates. But the contrast with Piaget is sharp. Where he saw a mind that constructs the object concept from scratch through action, core knowledge sees a mind that arrives pre-stocked with expectations and builds outward from them (Spelke & Kinzler, 2007).

The famous error, re-explainedA-not-B isn't what Piaget thought

Piaget's signature demonstration: hide a toy at location A, let the baby find it a few times, then hide it — in full view — at location B. Infants around 8–12 months reach back to A. Piaget read this as a fragile, incomplete object concept. Modern work says the concept is fine; something else fails.

The decisive clue is that the error depends on time. When the delay between hiding the toy at B and letting the infant reach is very brief, even eight-month-olds tend to reach correctly; stretch the delay by a few seconds and they revert to A. Tracked longitudinally, the delay an infant can tolerate before erring grows at a steady rate of roughly two seconds per month across the second half of the first year, from under two seconds at seven-and-a-half months to more than ten seconds by twelve months (Diamond, 1985). An object concept that flickered on and off would not behave like a clock. A memory that fades over a delay would — which points squarely at working memory as the limiting factor. Add to this the inhibitory problem: having been rewarded several times for reaching to A, the infant must actively suppress that now-prepotent habit to reach somewhere new, and the immature prefrontal circuitry that supports such control is not yet up to the job. On this reading the A-not-B error is a failure of memory and control, not of the object concept the infant plainly demonstrated when the toy was hidden at B in full view.

There is a genuinely competing account, and it is worth stating in its own right because it challenges the very idea that the error reflects a mental "concept" at all. The dynamic-systems theory reframes A-not-B not as a symptom of some hidden representation, weak or intact, but as the emergent product of the infant's whole reaching behavior unfolding in real time — the coupled dynamics of looking, planning, reaching, and remembering, shaped by the specific layout of the task and by the memory of prior movements in the body itself (Thelen, Schöner, Scheier, & Smith, 2001). On this view the repeated reaches to A build a motor memory that biases the next movement, and the error waxes or wanes with concrete variables — the delay, the distance between the wells, the infant's posture, even whether the infant is made to sit or stand — that a purely conceptual account has no reason to care about. The dynamic-systems reading dissolves the question "does the baby know the object is at B?" into a description of how a decision to reach is assembled, moment by moment, from perception and history. Whether one finds that satisfying or evasive, it makes the same core point as the executive-function account: the classic error is not a window onto a missing object concept.

Deferred imitation matters as much for what it measures as for what it shows, because reproducing a modeled act after a gap is a recall task in the same family as remembering where you left your keys: it requires bringing an absent event back to mind rather than merely recognizing it when it reappears. Follow-up work established that the memory is durable and detailed — infants imitate correctly after delays that stretch from a day to weeks, and the effect grows more robust across the second year (Meltzoff, 1988). That preverbal children can pass such a test is important precisely because they cannot yet tell us what they remember, and it sets up the puzzle that dominates the rest of this chapter: if infants encode and retain events this well, why do none of those events survive as the autobiographical memories we can later recount?

The mobile studyHow we know infants remember — Rovee-Collier's paradigm

The most elegant proof that young infants form and retain memories comes not from looking or imitating but from kicking. In Carolyn Rovee-Collier's conjugate reinforcement paradigm, a ribbon is tied from an infant's ankle to an overhead crib mobile so that the baby's own kicks make the mobile jiggle — the more vigorous the kick, the more dramatic the movement. Infants quickly learn the contingency and kick far more than at baseline, and crucially they remember it: brought back days later and shown the same mobile without the ribbon attached, a two- or three-month-old resumes kicking, demonstrating memory for the learned association (Rovee-Collier, 1999). The task is beautifully suited to infants because the response, a leg kick, is one they can already produce reliably, sidestepping the motor limitations that hobbled Piaget's search tasks.

What the mobile paradigm revealed about the structure of early memory is as striking as the retention itself. Young infants' memories are sharply context-bound: change the mobile, or even the pattern on the crib bumper that lined the playpen during learning, and the baby treats it as a new problem and fails to show retention — early memory is exquisitely tied to the cues present at encoding. Retention intervals also lengthen predictably with age, from a few days in the youngest infants to several weeks by the end of the first year. Most consequentially, a forgotten memory could be revived: a brief reminder — a glimpse of the moving mobile before the test — reinstated the memory well after it had apparently lapsed, showing that the trace had not been erased so much as rendered temporarily inaccessible. On the strength of such reactivation effects, Rovee-Collier argued for continuity in memory development, contending that the same basic memory processes operate from early infancy onward and that there is no sharp qualitative break between the "implicit" memory of babies and the "explicit" memory of older children — a pointed challenge to accounts that treat infant and adult memory as different in kind (Rovee-Collier, 1997).

The memory puzzleWhy can't you remember being a baby?

Infantile amnesia is the near-absence of episodic memories from before age 3–4. The interesting question is why, and the answer combines brain, cognition, and language.

The phenomenon is real and orderly. Asked to date their earliest memory, most adults land somewhere between three and four years old, with a near-total blank before that and a thinner-than-expected supply of memories from the years just after — a shortfall relative to what a simple forgetting curve would predict. The sharp puzzle, given Rovee-Collier's and Meltzoff's evidence, is that the shortfall cannot be blamed on a failure to encode: infants and toddlers plainly form memories, sometimes rich and durable ones. The problem lies in long-term retention and retrieval — early memories form but do not persist into later childhood in a form the older mind can access and narrate. Distinguishing encoding from retrieval failure is the key move that turned infantile amnesia from a mystery into a research problem.

The neurobiological account begins with the fact that the structures episodic memory most depends on — the hippocampus and the prefrontal cortex — are among the last to mature, so the very machinery for binding an event into a durable, retrievable trace is still under construction during the years the memories would have to be laid down. A more provocative proposal turns the developing hippocampus from a passive bottleneck into an active eraser: the extraordinary rate of neurogenesis in the infant hippocampus, the same flood of new neurons that makes early learning possible, may also destabilize and overwrite the memory traces already stored there, so that memories are lost not because they were never formed but because the circuit kept remodeling itself around them (Bauer, 2015). On this "complementary processes" view, encoding and forgetting are two faces of one immature, rapidly changing system — which is why early memories can be genuinely durable over days or weeks yet fail to survive the years-long haul into later childhood.

The cognitive-social account addresses a different piece: not why traces decay, but why the memories that do survive take the particular form of a life story only after age three or four. Autobiographical memory — memory of specific events located in one's own past and understood as having happened to me — requires a self around which to organize experience and a language in which to encode and rehearse it as a tellable narrative, and both consolidate in exactly the window where reliable early memories begin (Nelson & Fivush, 2004). On this social-cultural theory the crucial catalyst is talk: as parents and children reminisce together, children learn the forms a memory takes — what counts as the who, what, where, and why of an event worth retaining — and children whose caregivers reminisce in a rich, elaborative style go on to report earlier and more detailed autobiographical memories. Consistent with the theory, the average age of earliest memory varies across cultures in step with how much, and how narratively, families talk about the past. Brain maturation and the emergence of a narrating self are not rival explanations so much as the two halves of one answer: the hardware for durable episodic memory and the self-and-language scaffolding that gives episodes their autobiographical shape come online together, and only then does the continuous personal past we take for granted begin to accumulate.

The through-lineWhat the infant mind was doing all along

Read from end to end, the chapter tells one story about method. Piaget built a monumental, largely accurate map of the sequence in which infant abilities appear — reflexes give way to intentional means–end action, which gives way to representation — but because he read the map through what infants could do, he mistook the slow arrival of coordinated action for the slow arrival of knowledge (Piaget, 1952, 1954). Switch the measure from reaching to looking, or to kicking, or to imitating after a delay, and a quicker mind comes into view: object permanence and physical expectations in the first months (Baillargeon et al., 1985; Spelke et al., 1992), durable and reactivatable memory by three months (Rovee-Collier, 1997, 1999), and representational recall well before the second birthday (Meltzoff, 1988). The classic errors survive as real phenomena but shed their old interpretations — A-not-B becomes a problem of memory, control, and embodied dynamics rather than a missing concept (Diamond, 1985; Thelen et al., 2001), and infantile amnesia becomes a problem of retention and retrieval in a still-remodeling brain and a not-yet-narrating self rather than an absence of learning (Bauer, 2015; Nelson & Fivush, 2004). The lesson generalizes past infancy: what we conclude about a mind depends heavily on the response we ask it to make.

SourcesCited in APA 7

Baillargeon, R. (1987). Object permanence in 3½- and 4½-month-old infants. Developmental Psychology, 23(5), 655–664.
Baillargeon, R., Spelke, E. S., & Wasserman, S. (1985). Object permanence in five-month-old infants. Cognition, 20(3), 191–208.
Bauer, P. J. (2015). A complementary processes account of the development of childhood amnesia and a personal past. Psychological Review, 122(2), 204–231.
Diamond, A. (1985). Development of the ability to use recall to guide action, as indicated by infants' performance on AB. Child Development, 56(4), 868–883.
Meltzoff, A. N. (1988). Infant imitation after a 1-week delay: Long-term memory for novel acts and multiple stimuli. Developmental Psychology, 24(4), 470–476.
Nelson, K., & Fivush, R. (2004). The emergence of autobiographical memory: A social cultural developmental theory. Psychological Review, 111(2), 486–511.
Piaget, J. (1952). The origins of intelligence in children (M. Cook, Trans.). International Universities Press.
Piaget, J. (1954). The construction of reality in the child (M. Cook, Trans.). Basic Books.
Rovee-Collier, C. (1997). Dissociations in infant memory: Rethinking the development of implicit and explicit memory. Psychological Review, 104(3), 467–498.
Rovee-Collier, C. (1999). The development of infant memory. Current Directions in Psychological Science, 8(3), 80–85.
Spelke, E. S., Breinlinger, K., Macomber, J., & Jacobson, K. (1992). Origins of knowledge. Psychological Review, 99(4), 605–632.
Spelke, E. S., & Kinzler, K. D. (2007). Core knowledge. Developmental Science, 10(1), 89–96.
Thelen, E., Schöner, G., Scheier, C., & Smith, L. B. (2001). The dynamics of embodiment: A field theory of infant perseverative reaching. Behavioral and Brain Sciences, 24(1), 1–34.