Cause and Effect Baby: When "I Did That" First Clicks

Contingency vs. causation: why these are two different things, and when your baby learns both

Cause and effect baby learning begins as something surprisingly simple: noticing that two events reliably co-occur. Event A happens, event B follows. No understanding of mechanism required. Researchers call this contingency detection, and it appears as early as 6 to 8 weeks, when babies in laboratory studies show clear preference for a mobile that moves in response to their kicking versus one that moves the same amount but independently of their actions.

The classic evidence comes from a series of studies by Carolyn Rovee-Collier and colleagues beginning in 1969, in which 3-month-olds learned to kick their legs to activate a mobile attached by a ribbon to their ankle. Babies as young as 6 to 8 weeks showed the same basic learning within minutes. What they were detecting was the timing, the co-occurrence, not the causal relationship per se. They couldn't have named themselves as the agent. But the contingency was enough to drive persistent, effortful behavior.

The jump from "event A is followed by event B" to "I am causing B" is the harder cognitive milestone, and it takes considerably longer. Research by Zaadnoordijk and colleagues, published in Developmental Cognitive Neuroscience in 2020, used EEG recordings to look at what was actually happening in 3- to 4.5-month-old brains during a modified mobile task. They found that only a subset of infants at that age showed a neural signature consistent with a true causal action-effect model, a genuine internal sense that "my movement produced that outcome." The others registered contingency but not causal agency. The sense of agency, in other words, doesn't switch on as a milestone at a set age. It emerges gradually, and not all babies arrive at the same time.

This matters practically because it changes what parents can actually do. You cannot hand a baby the concept of causal agency. What you can do is create conditions under which the baby's brain has repeated, clear opportunities to test its own hypothesis. That is a different kind of play than choosing the right toy.

What the brain is doing before the behavior shows up

The brain's causal architecture comes online before it can coordinate the body to act on it. This is easier to see in the domain of visual causality than in motor causality, and a striking experiment by Leslie and Keeble in 1987 demonstrated it in 6-month-olds.

Using a habituation paradigm, the researchers showed babies a film of one object moving toward another, making contact, and the second object moving away. When the film was reversed, so that the second object appeared to launch the first (an event that violates normal causal structure), babies looked longer. The key finding was that reversing a causal film produced more recovery of attention than reversing a similar but non-causal film, where a gap in time separated contact and movement. At 6 months, something in the visual system already expects causes to precede effects directly in time, without delay.

The neural side of this story has become considerably clearer in recent years. Work by Köster, Langeloh, Michel, and Hoehl, published in NeuroImage in 2021, recorded brain activity in 9-month-olds while they watched expected and unexpected events. Unexpected outcomes produced a pronounced response in the brain's theta rhythm, sustained across two seconds and spread across the scalp. The implication is that the infant brain is actively generating predictions and then measuring the gap between prediction and reality. That gap is the signal the brain uses to update its model.

Researchers sometimes call this prediction error, and it functions, in the brain's terms, as a learning trigger. A surprise is not merely attention-grabbing; it is a moment when the brain flags: revise the model. What this means for everyday play is that activities yielding some surprises are doing more cognitive work than activities the baby has already fully mastered. A consistent, always-the-same, perfectly predictable response teaches very little after the first few repetitions.

This connects to a finding by Kidd, Piantadosi, and Aslin published in PLOS ONE in 2012, sometimes called the Goldilocks Effect, in which 7- and 8-month-olds were most likely to disengage from a stimulus that was either completely predictable or completely unpredictable, and most likely to stay engaged with stimuli at an intermediate level of complexity. Not too easy to predict, not too hard. The baby is, in a real sense, self-calibrating toward the level of challenge that maximizes learning. You can't set this threshold for the baby, but you can notice when the baby has moved past it and offer something a step more complex.

When do babies learn cause and effect: a timeline with cleaner lines

Most timelines for cause and effect baby development treat the milestone as a single step, usually placed at 4–6 months. The actual picture is a staircase, with different abilities measurable at different points using different methods.

Around 2–3 months, contingency detection is present. A baby will increase kicking to activate a foot-tethered mobile, and stop when the ribbon is detached. Watson (1972) described this as "the game," the baby's first extended experience of producing a reliable environmental response. It is probably the earliest substrate of what will later become deliberate causal agency.

Around 6 months, visual causality perception is in place. Leslie and Keeble's launching-event studies showed that 6-month-olds already represent causal structure in what they see, even when they cannot produce it themselves. The brain's causal expectations are ahead of the hands.

Around 8–9 months, intentional action-effect learning is consolidating. The CDC milestone checklist for 9 months includes specific behavioral markers: looking for an object when it is dropped out of sight, and banging two things together. These are cause-and-effect behaviors, though the CDC does not label them as such. What changes at this age is the planning component. The baby is not just registering that something followed the baby's action; the baby is constructing a sequence in advance.

By 10–12 months, babies show what might be called selective causation: they choose which action to take based on which outcome they want, and they treat familiar people as more reliable tools for producing certain outcomes than objects are. This is visible in the way 12-month-olds point to request, a behavior that assumes the other person will act as a causal agent on the baby's behalf.

The activities for a 9 month old stage is particularly rich for cause-and-effect play precisely because the planning component is newly available. Intentional repetition, doing the same thing again to see if it produces the same result, is not stubbornness or a failure of interest; it is hypothesis-testing in the most literal sense.

How cause and effect learning builds toward language

There is a connection between early cause-and-effect experience and later language development that most parenting sources overlook, possibly because it takes some time to become visible.

The connection runs through contingency. A 2026 review by Suarez-Rivera and Tamis-LeMonda in Philosophical Transactions of the Royal Society B lays out the proposed mechanism: social contingencies, the timely responses caregivers give to infant actions, facilitate learning in several specific ways. They build and sustain infant attention, create multisensory experiences around the object or action the baby initiated, and help the baby form connections among events in memory. Over time, the baby's brain begins to map the contingency structure of social interaction the same way it maps the contingency structure of a kicking mobile: I do X, this person does Y, and the relationship is reliable.

This matters for language because verbs, specifically, describe actions and the results of actions. To learn the word "push," a baby needs a stable mental representation of pushing as an event with a predictable structure: agent, action, result. A 2024 study by Masek and colleagues in Infant Behavior & Development, following 104 infant-caregiver pairs from 6 to 18 months, found that both social attention and social contingency at 6 months predicted receptive vocabulary at 18 months. Models that included contingency outperformed models that included attention alone.

What this suggests is that responsiveness, talking to a baby during play, narrating what the baby is doing, responding to those vocalizations as if they were conversational turns, is not just warm parenting style. It is creating contingent input that the baby's language-learning system can use. If you want practical guidance on what this looks like month by month, the sensory play for babies guide covers the activity structure; the underlying learning mechanism described here is what makes it work. For an even earlier window, the how to help baby talk article covers what that responsiveness looks like in the first months of life.

Our Sensory Play Cards (0–12 months) are structured around exactly this contingency pattern, activities that respond to what the baby does, with textures, sounds, and visuals that change based on the baby's own action rather than running independently.

What cause and effect baby play actually looks like by stage

The research gives a coherent picture of what is happening at each stage, and it turns out that the activities parents are already doing, if they are tuned into the timing, are well-matched to the developmental need.

In the first two to four months, the most relevant contingencies are social and responsive. When a baby vocalizes and you respond, when the baby smiles and you mirror it back, the baby is building the first contingency models: I produce this, that follows. A 2018 AAP policy statement by Yogman and colleagues notes that back-and-forth interaction is the earliest form of play and specifically supports brain architecture for later learning. For object-based contingency at this stage, foot-activated mobiles work well because the foot is the most reliable actuator a young baby has; arm and hand movements are too variable to produce clear contingencies consistently.

From four to seven months, hand-to-object interaction becomes more reliable, and object response becomes the interesting variable. A baby at this age is not testing whether the rattle can be held; the testing is about whether the rattle's noise is related to what the baby does with it. Activities that give clear, consistent, and somewhat varied feedback work well: objects that make a sound when squeezed, surfaces that crinkle when touched, containers that produce a satisfying sound when objects are dropped in. The interest is not in the object itself but in the relationship between the baby's action and the object's response.

The baby brain development first year article covers the neurological window behind this period in more detail. One practical implication is that object complexity should stay moderate. A toy that always does the same thing loses predictive value fast. A toy that does nothing is equally useless. The sweet spot is an object that responds to the baby's action in a way that is interesting but not completely random, which is why crinkle fabric, squeaky toys, and water work better at this stage than sophisticated electronic toys.

From seven to ten months, intentional cause-and-effect behavior becomes visible in the way the baby returns to the same action repeatedly. The banging-on-the-tray phase is not noise for its own sake; it is systematic variation: bang harder, bang softer, bang with the palm, bang with the spoon. Each variation tests a slightly different hypothesis. The object permanence baby development article notes that this same experimental quality is what drives hide-and-find games; the underlying drive is the same.

At this stage, a strategically messy environment is a feature. Containers and things to fill them, objects that fit into holes, cause-and-effect toys where the relationship between action and outcome is visible (not hidden inside a box) all feed the experimental drive. The Sensory Play Cards work particularly well here because the activities are open-ended enough to tolerate the baby's own variations on the suggested interaction.

From ten to twelve months, social causation becomes the most interesting domain. The baby is testing whether specific adults can be caused to do specific things, and the answer is reliably yes. Pointing, offering, pushing objects toward a caregiver, vocalizing with clear intentionality: all of these are cause-and-effect experiments where the other person is the variable. Narrating these moments remains high-value, because the contingency between the baby's action and the caregiver's labeled response is precisely the kind of structured input the language system can begin to map.

When to mention the pattern to your pediatrician

Cause and effect development almost never presents as a discrete "missed milestone" at a scheduled visit. It shows up as the texture of engagement: whether the baby appears to understand that actions produce results, whether the baby returns to actions that worked, whether the baby uses simple social strategies to produce desired outcomes.

A few patterns are worth flagging at a routine visit. If by 9 months a baby shows no sign of intentional repetition, no banging, no dropping objects with apparent interest in the result, no eye-checking after an action to see what happened, that is a good thing to mention. If at 12 months the baby does not use any social-causal behavior (pointing, reaching toward a caregiver, vocalizing to initiate a response), that is the moment to bring it up. Neither pattern is an emergency, but both are the kind of thing a pediatrician will want to note.