Welcome. This course follows a central developmental question: how does a brain that is rapidly building circuits turn everyday experience into learning, memory, and eventually routines? In this module, we begin with the behavioral evidence. Before discussing the brain circuits that support habits, we need to see how researchers establish that an infant has learned that their own action produces an effect.
By the end of this lesson, you should be able to explain how reinforcement can turn a spontaneous movement into an action an infant repeats for a particular outcome, using Rovee-Collier’s mobile paradigm as the key example. You will also distinguish this early action–outcome learning from a fully established habit.
A useful study pace is about 40 minutes: roughly 10 minutes for the first reading, 10 minutes for the second, and 20 minutes for the explanations and observations below.
From movement to action: why consequences matter
Infants move constantly: kicking, waving, vocalising, turning their head, reaching, and later crawling or banging objects. Most of these movements begin as exploratory or spontaneous behavior. A movement becomes more meaningful when the infant detects a reliable pattern:
“When I do this, something interesting happens.”
In learning theory, this is an action–outcome contingency. Formally, an outcome is contingent on an action when it is more likely after that action than when the action does not occur:
An infant does not calculate probabilities consciously, of course. But nervous systems can learn statistical regularities from repeated experience. If kicking makes a mobile move immediately and reliably, the infant may kick more. The moving mobile is then a reinforcer: a consequence that increases the future likelihood of the behavior that produced it.
This is operant conditioning. It differs from classical conditioning:
- In classical conditioning, an infant learns that one event predicts another—for example, a familiar feeding-related cue may predict milk.
- In operant conditioning, the infant learns that their behavior changes the environment—for example, “my kick makes that mobile bounce.”
The second form is especially important for the beginnings of agency: the developing sense that one’s actions have consequences.
The mobile conjugate reinforcement paradigm in a lab setting
Read this research report to see how scientists turn a simple infant behavior—kicking—into a precise test of contingency learning, reinforcement, and retention.
Begin in Section 1, “The Mobile Conjugate Reinforcement Paradigm in a Lab Setting.” Read the basic contingency, focusing on why a higher kick rate after learning is meaningful only when compared with baseline kicking. Then read the following methods passage, beginning with “Following the standard set-up…” and continuing through the description of Day 1 and Day 2 phases. Pay particular attention to why the mobile remains visible during non-reinforcement periods. In the setup description, read the immediate-feedback mechanism. Finally, read the results section immediately above Table 1. Read the group learning result, then inspect Table 1 to compare baseline with the three learning blocks and the two retention measures.
Rovee-Collier’s mobile: a compact experiment with a large implication
Carolyn Rovee-Collier and colleagues developed variants of the mobile conjugate reinforcement paradigm to study learning and memory in young infants, often around 2–6 months old.
The experimental logic is elegant. A ribbon is attached to one ankle and to a mobile above the infant. When the infant kicks the connected leg, the mobile moves. The infant is not instructed, prompted verbally, or physically guided to make the movement. The researcher asks whether the infant’s own experience changes later behavior.
| Phase | What happens | Why it matters |
|---|---|---|
| Baseline | The infant can kick, but kicking does not move the visible mobile. | Measures spontaneous kicking and controls for simply seeing the mobile. |
| Reinforcement / acquisition | Kicking the connected leg immediately makes the mobile bounce. | Tests whether the infant increases behavior that produces the interesting outcome. |
| Immediate retention | The mobile is visible again, but kicking no longer moves it. | Tests whether the learned association affects behavior after reinforcement is removed. |
| Delayed retention | The procedure is repeated after a delay, often one day. | Tests memory for the learned contingency. |
The word conjugate is important. This is not merely a reward delivered after every response. The sensory result is connected to the strength or frequency of responding: more or stronger kicking produces more mobile movement. That close, immediate mapping makes the causal structure unusually detectable for an infant.
In the reported study, the group’s mean kicking rate rose from about kicks per minute at baseline to in the first learning block, then to roughly and in the next two blocks. These are group-level results, not a standard every individual infant must meet. Still, they demonstrate a systematic shift: behavior was not simply occurring; it increased when it produced the moving mobile.
Researchers also use a learning criterion based on the ratio:
A ratio of at least is commonly treated as evidence that the infant increased kicking substantially relative to their own baseline. This within-infant comparison matters because babies differ greatly in their normal activity level. One child may kick energetically even before the task begins; another may be quiet but show a large proportional increase.
The design helps rule out simpler explanations:
- General arousal: perhaps the infant kicks more simply because the mobile is exciting. The non-reinforcement baseline is important here: the mobile is still visible, but kicking has no effect.
- Maturation over minutes: normal motor activity may change over time, but the sharp connection between the reinforcement phase and increased responding is stronger evidence than time alone.
- An accidental coincidence: repeated, immediate action–outcome pairings make a causal relation far more plausible than a one-off event.
The experiment does not establish that an infant has an adult-like verbal thought—“I intend to make the mobile bounce.” It does show that even very young infants can alter their behavior based on the consequences of their behavior.
Is increased kicking really goal-directed action?
This question requires careful language. Reinforcement and goal-directed action overlap, but they are not identical concepts.
At a basic level, the mobile task supports the following chain:
Initially, a kick may be just one movement among many. With repeated consequences, the infant increasingly performs that specific action because it produces a valued sensory outcome: motion, visual stimulation, perhaps sound, and novelty.
That is the beginning of goal-directed organization: behavior becomes selectively linked to an expected result.
However, developmental scientists are appropriately cautious. An increased response rate alone can be explained by a relatively simple learning mechanism: a behavior followed by stimulation becomes more likely. To make a stronger case that behavior is genuinely goal-directed, researchers seek evidence that the infant has some expectation of a specific outcome.
Three observations strengthen that interpretation.
1. The response becomes action-specific
Infants do not merely become globally more active. Research using detailed movement tracking has found increasingly specific response patterns. In some contingency tasks, older infants show selective use of the limb connected to the outcome rather than equally increasing movements of every limb.
This matters because “moving more” and “using the effective action” are different findings. The latter is closer to an action–outcome representation:
2. The infant responds when the contingency is disrupted
When reinforcement stops, infants may briefly increase responding—sometimes called an extinction burst. In the mobile task, an infant who has learned “kicking moves the mobile” may initially kick even more when the expected movement suddenly fails to occur.
That behavior is informative. It suggests that the infant is not merely passively stimulated by the mobile; they are acting in a situation where the expected consequence has been violated. Yet it remains an inference: frustration, arousal, or disengagement can differ substantially across infants.
3. The infant anticipates the outcome
At around 10 months, studies using button-press contingencies have observed anticipatory gaze shifts toward the location where the reinforcer will appear before infants press the button. Looking toward the expected outcome before acting is a stronger indicator that an action and its consequence have become coordinated.
A Quarter Century of Research on Infant Contingency Learning
This review places Rovee-Collier’s paradigm in the wider study of infant operant learning and examines what researchers can—and cannot—conclude about emerging agency.
First, read Section 1.0, “An Introduction and a Brief History of the Study of Infant Contingency Learning,” from the paragraph beginning “Some of this early literature…” through the end of the section. Notice the distinction between classical conditioning and paradigms in which an infant’s behavior has consequences. Then read the final part of Section 2.1, “Methodological Challenges and Innovations,” beginning “Research comparing different types of reinforcers…” through the paragraph ending with “real-time learning processes.” Focus on the finding that infants can learn specific movement patterns, not only become generally more active. Next, read Section 2.3, “Contingency Learning Characteristics: Examinations of Agency.” Start with the definition of emerging agency. Continue through the paragraphs on extinction and movement coordination. Finish with anticipatory looking. Keep the review’s caution in mind: a learned contingency is evidence for early agency, but it does not by itself prove adult-like intention.
A crucial distinction: early action learning is not yet a habit
In everyday language, we might call a repeated infant behavior a “habit.” In neuroscience and learning research, that would be premature.
A useful contrast is:
| Feature | Goal-directed action | Habit |
|---|---|---|
| Main relationship learned | Action valued outcome | Situation/cue response |
| Why the behavior occurs | The outcome is expected or wanted | The response has become automatic in that context |
| Sensitivity to outcome | If the outcome loses value, behavior should decline | Behavior can persist even when the outcome changes |
| Infant example | Pressing a toy button while looking toward the expected light | Automatically reaching toward a familiar object in a familiar routine |
The strongest laboratory test of goal-directedness is outcome devaluation: after an action has been learned, make its outcome less desirable and see whether the action declines. If behavior declines specifically because its outcome is no longer valuable, that supports a genuine action–outcome relationship.
Such tests are difficult with young infants. They cannot explain what they expect, and their motivations, attention, motor abilities, and emotions all change quickly. The research review therefore makes a valuable distinction: infant contingency learning clearly shows that actions can be reinforced, but fully demonstrating goal-directed behavior in the strict technical sense requires additional evidence.
So the most accurate conclusion is:
Rovee-Collier’s paradigm shows the initial formation of action–outcome learning—an essential precursor to goal-directed action and, later, to habits.
In the next lesson, we will examine how extensive repetition can gradually shift control away from the expected outcome and toward more automatic cue–response patterns.
What this looks like at eight months
Your son is beyond the age of the classic ankle-and-mobile studies, but the underlying learning principle is very active at eight months. His expanding motor repertoire gives him more ways to discover controllable effects.
You may see action–outcome learning when he:
- bangs an object and repeats the movement after it makes a sound;
- drops an object and watches for its fall, sound, or your reaction;
- presses, pulls, shakes, opens, or closes a safe toy feature repeatedly;
- vocalises, pauses, and vocalises again after a caregiver responds;
- makes a gesture toward a familiar routine and watches for the expected next event.
The important observation is not simply repetition. Look for a pattern of action, outcome, renewed action, especially when the action becomes more targeted or when he looks toward the expected effect.
A practical, science-aligned parenting implication is to provide safe, immediate, and reasonably predictable feedback during play. A securely designed cause-and-effect toy, a spoon tapping safely on a tray, or a caregiver responding to a vocalisation all allow the infant to detect contingencies. The aim is not to “train” him constantly; it is to leave room for his own exploration and let his actions sometimes produce clear, understandable effects.
Two cautions are equally important:
- Do not recreate the mobile experiment at home with strings, ribbons, or attachments. Cords and loops can pose entanglement or strangulation risks. The value of the experiment is its logic, not its equipment.
- Do not interpret every repetition as a deliberate plan. Infants repeat behavior for many overlapping reasons: sensory pleasure, motor practice, novelty, social engagement, fatigue, or emotion. One episode is not a diagnosis of intention; a consistent contingent pattern is more informative.
Key takeaways
Reinforcement occurs when an outcome increases the future probability of the behavior that produced it. In Rovee-Collier’s mobile paradigm, infants learn that kicking causes a mobile to move, then increase their kicking above baseline.
The experiment is powerful because it compares reinforcement with non-reinforcement while keeping the mobile visible, and because it measures both immediate behavior change and later retention. Specific limb movements, persistence after a contingency is removed, and anticipatory looking provide increasingly strong evidence that infants are learning action–outcome relations.
Still, contingency learning is best viewed as the foundation of goal-directed action, not proof of adult-like intention or a fully formed habit. Next, we will build on this distinction by examining how repetition and the basal ganglia help shift actions from outcome-sensitive choices toward cue-triggered habits.
Can't find a good explanation? Sign up and we'll make it for you
Sign up