Cause and Effect Toys: What a Squeak Actually Teaches
Squeeze. A squeak. Squeeze again. Water.
One action, two simultaneous outputs, identical every time. This is one of the first pieces of experimental apparatus an infant ever gets his hands on, and its specifications are better than most toys deserve.
- 01Delay approaching zero. The sound arrives while the hand is still closing, with no perceptible gap.
- 02One action, one result. Every squeeze sounds; no squeeze never sounds.
- 03Repeatable. Fifty trials in a single bath, at no cost.
- 04Multi-channel. Sound, water, palm pressure and the toy's own deformation, all from one source.
The technical name for this is contingency. And what the baby is learning is not “sound”. It is I did something, and something happened. That “I” is the earliest foundation of a sense of agency.
The turns are at 4 months and 8 months, not 6 and 12
The familiar summary — incidental discovery at 0–6 months, repeated verification at 6–12, deliberate attempts beyond 12 — points the right way but places both boundaries too late, and mischaracterises what happens in the middle.
Repeating actions on his own body — watching hands, sucking, kicking. The toy is an object that reacts, not yet something he makes react. Splashing is enjoyable, not yet “produced by me”.
Repeating an action that produced an interesting effect in the world. Squeeze → squeak gets repeated. But fascination with the effect is not the same as a causal link — at this point the connection between a specific action and its result is not necessarily conscious.
Combining actions toward a goal; means–end structure appears. Pressing the toy under water to fill it, then lifting and squeezing to spray — that is a plan, not an accident.
Varying the action to see how the result changes: squeezing harder, softer, at different angles; dropping it to test whether it floats. Piaget called this stage the young scientist.
Two corrections follow.
First, the real turns are at about 4 months and about 8 months. From 4 months, behaviour shifts from acting on the self to acting on objects. From 8 months, genuine means–end structure appears. Placing “repeated verification” at 6–12 months merges two stages of different character.
Second, and more importantly: the repetition at 4–8 months is driven by “the effect is interesting”, not by “testing a hypothesis”. The classic description of this substage is that infants are fascinated by the effects of their actions, although at this point they are not necessarily making a conscious connection between the particular things they do and the resulting consequences. Acquiring cause–effect knowledge through repeated observation of action–consequence pairs is the work of 8–12 months.
This is not pedantry; it changes what a parent should do. At 4–8 months, your job is to make the effect happen reliably. At 8–12 months, your job is to give him a goal that takes two steps. Same toy, two methods.
And the table is a sequence, not a calendar. The Jean Piaget Foundation notes substantial variation between children — and within one child across domains — in the age of reaching stage-three behaviours, with some children showing them in one domain from three to four months while another domain still sits at stage two. Read the order; do not audit the dates.
Why “it always squeaks” beats “it squeaks impressively”
There is a technical distinction here that rarely gets stated — and it decides which toy is learnable.
Perfect contingency: action and result are bound 100%. Clapping your own hands always makes a sound; kicking always moves the blanket. This is the object of primary circular reactions: the baby's own body.
Imperfect contingency: action and result are bound only “most of the time”. A squeeze usually squeaks — but not when water has entered the valve, not at that angle, not when the air path is blocked. In Watson's formulation, in such cases the behaviour is not always sufficient to cause the effect, and the behaviour is not always necessary for the effect.
One causal link, repeated six times. Nothing to disentangle: what gets learned is the link itself.
Two events, not one cause and one effect. The link dissolves in time.
Four of six attempts unanswered. What gets learned: “sometimes it works.”
Perfect contingency is learned fastest because it carries no noise. Which explains why infants are first captivated by their own hands and mouths rather than by toys: the hand is the most obedient object in the world.
That leaves the design task exposed: make an external object as obedient as a body. Every squeeze sounds. Every angle sprays. No air leak, no stuck valve, no “try again and hope”.
A toy that responds intermittently is not “more interesting”. It is harder to learn from — because it teaches that the world is random. A dying battery, a valve full of bath water, a design that only works at one angle: these are not inconveniences, they break the temporal stability of the causal chain.
Three properties, and how each fails
| PROPERTY | WHAT IT MEANS | HOW IT BREAKS | CONSEQUENCE |
|---|---|---|---|
| Immediacy | The effect follows the action with no perceptible gap | Valve lag, electronic delay, a toy that “thinks” for a second | Temporal proximity underpins causal judgement; stretch the gap and two events become two unrelated things |
| Predictability | Same action → same result, every time | Intermittent response, water in the air path, low battery | Feedback becomes noise; the child learns “sometimes it works” |
| Intensity | Enough to be noticed, not enough to drown everything else | Too loud, too bright, too fast, too much motion | Attention has a budget; strong stimulation crowds out attention to “I did that” |
The third is the most misread. The commercial instinct is louder and brighter, but what the infant needs to read is the causal relation, not the light show. A screaming toy does capture attention instantly — it captures attention to the toy, not to his own action. And the latter is what this stage is for.
This is the same conclusion we reached about water: controllable weak input outperforms uncontrollable strong input.
Five steps, not three
Model → wait → respond points the right way, but the starting point is wrong: put “model” first and it easily becomes teaching, when the child is usually already trying — just clumsily.
One hard note: receiving no return is itself stressful. Harvard's Center on the Developing Child states plainly that a persistent lack of response activates a stress response in the child. In this sequence, “return” is not a bonus. It is the floor.
Add one variable at a time
The only purely engineering recommendation here. It does not come from research; it comes from the idea of an attention budget.
A toy that simultaneously squeaks, sprays, glows, vibrates and changes shape gives the child less to read, not more — because he cannot tell which change produced which result. Add variables and causality stops being attributable.
The order should therefore be:
Five signals that the load is too high — back up one step if any appear:
The fifth is easily misread as “he doesn't like it”. Freezing is not boredom; it is overload — the same event as turning away, expressed differently. Adding more stimulation at this point is the most common mistake in the whole routine.
It teaches expectation, not logic
One thing needs stating plainly: no evidence shows that any cause-and-effect toy raises intelligence, trains logical thinking, or accelerates cognitive development. What such a toy delivers is two basic and quite limited things — my action has an effect (agency), and the world is predictable (expectation).
Both are genuinely foundational. But a foundation is not a house. Real causal reasoning, hypothesis testing and variable control come years later; what happens at this stage is only the layer underneath: I did it, and it happened.
That layer is also why a person's response matters more than a toy's feedback. A toy gives consistent physical feedback; a person gives consistent social feedback — and the second is the source of I am worth responding to. Which is why, in section five, the first variable worth adding is not a feature. It is your response.
Three sentences
First: a squeeze that squeaks is the earliest controllable experiment an infant can run — zero delay, one-to-one correspondence, repeatable, multi-channel. And “it always squeaks” matters far more than “it squeaks impressively”.
Second: the stage boundaries sit at 4 months and 8 months, not 6 and 12. The repetition at 4–8 months is the effect is interesting, not testing a hypothesis; genuine means–end structure arrives at 8–12. Same toy, two methods.
Third: you are more effective than the toy. A toy supplies stable physical contingency; you supply stable social contingency — and because no response is itself stressful, the second is not optional.
Every squeeze, the same answer
One-piece moulded, no valve to clog with bath water — so the causal chain stays intact through the fiftieth repetition.
SOURCES
- Ormrod, J. E. (2008). Human Learning, 5th ed., chapter 11 supplement — Piaget's sensorimotor substages: primary circular reactions (1–4 months); secondary (4–8 months), where infants are “fascinated by the effects of their actions, although at this point they are not necessarily making a conscious connection”; coordination of secondary reactions (8–12 months), where they “gradually acquire knowledge of cause-effect relationships” and become goal-directed; tertiary (12–18 months). nova.edu
- University of California, San Diego, Cognitive Science 101B — substage table with the same age ranges; substage 5 described as “deliberate ‘scientific’ variation of actions in order to observe effects”. ucsd.edu
- Fondation Jean Piaget, Stades 1–3 — the secondary circular reaction consists in intentionally reactivating the action that just produced an interesting result; notes substantial inter- and intra-individual variability in the age of access to stage-three behaviours. fondationjeanpiaget.ch
- Watson, M. W., Human Development (Karger) — circular reactions as self-teaching devices; secondary reactions maintained by pleasure in contingent visual and auditory reactions; secondary and tertiary reactions involve less-than-perfect contingency, where “the behavior is not always sufficient to cause the effect... and the behavior is not always necessary for the effect”. karger.com
- Center on the Developing Child, Harvard University — 5 Steps for Brain-Building Serve and Return: notice the serve; return the serve; give it a name; take turns and wait (“waiting is crucial”); practice endings and beginnings; and the note that never getting a return can be stressful for a child. developingchild.harvard.edu
- Harvard Graduate School of Education, Usable Knowledge (2017) — the same five steps restated for caregivers, with waiting as the mechanism that keeps the exchange going. gse.harvard.edu
Mold-Free Bath Toys: Why the Hole Matters More Than the Material