McGarrigle and Donaldson Naughty Teddy Study (1974)

McGarrigle and Donaldson’s 1974 naughty teddy study tested whether young children fail Piaget’s conservation tasks because of how the task is run. Eighty children aged 4 to 6 took part. Far more were right when a teddy bear seemed to change the materials by accident than when the experimenter did it on purpose.

Key Takeaways

  • Aim: To test whether children fail Piaget’s conservation tasks because of how the task is run. Conservation means knowing an amount stays the same when it looks different.
  • Method: Eighty children aged 4 to 6 took part, 40 from Edinburgh nursery schools and 40 from one primary school. Each judged two situations of number, with counters, and two of length, with string, before and after a change. The independent variable was how the change was made: the experimenter did it on purpose, or a teddy bear seemed to do it by accident. The dependent variable was whether the judgement after the change was correct. Every child did both conditions (repeated measures). Half met the accident first and half the deliberate change first (counterbalancing).
  • Results: About 72% of judgements were right after the accident and about 34% after the deliberate change. Fifty of the 80 children got all four right after the accident, but only 13 after the deliberate change. Order mattered: children who met the deliberate change first scored 56.9% after the accident, against 86.9% for those who met the accident first. In the first four judgements alone, 30 of 40 children (the group that had the accident first) got all four right after the accident and only 5 of 40 (the group that had the deliberate change first) after the deliberate change. Primary children scored higher than nursery children in every condition (the paper reports no statistical test).
  • Conclusion: The authors concluded that traditional tasks underestimate what children know. They argued that children may read a deliberate action as a hint about what the question means. This is the authors’ explanation, and it remains one candidate among several. Thirty of the 80 children still did not get all four right after the accident.
  • Strengths: Same children in both conditions, so differences between children cannot explain the contrast. Tested an explanation: the usual view predicts no difference between conditions, but there was one. The authors reported the order effect openly. Other teams found the same pattern, though Light et al. found lower overall scores (Hargreaves et al., 1982; Light et al., 1979).
  • Limitations: The accident condition added teddy’s involvement, pretend alarm and a longer gap, so intention alone is not shown. The order effect means the 50-versus-13 comparison is not a clean test. Children were not asked for reasons, so a right answer may not show understanding. An adult acted out the accident, so children may have responded to the experimenter’s cues. The sample was 80 children from one city, tested before the paper was published in 1974, so the results may not generalise.
  • Ethical Issues: Parental consent: the paper does not report whether the children’s parents agreed. A way of dealing with this is to explain what the children will take part in and ask their parents if they are happy. Debriefing: the paper does not say what the children were told afterwards. A way of dealing with this is to explain afterwards what they took part in and why. The pretended accident was a mild pretence, and the task itself was harmless.
Simply Psychology Answers
Instant answers built on 19+ years of trusted expertise.

Aim

Piaget said that young children have not yet grasped conservation. This is the understanding that an amount stays the same when its appearance changes and nothing is added or taken away.

His standard test works like this. Two rows of counters, equal in number, are placed side by side. The child agrees that the rows are the same. The experimenter then pushes one row together, so it is shorter, and asks the question again.

Simply Psychology Answers
Instant answers built on 19+ years of trusted expertise.

A child who has not yet grasped conservation typically now says the longer row has more. Piaget’s theory places children who fail this test in what he called the preoperational stage. The idea is explained on the conservation page.

Simply Psychology Answers
Instant answers built on 19+ years of trusted expertise.

By 1974 there was a puzzle. Other studies suggested that children may have the knowledge needed for conservation well before they pass the standard test. Some used training, or games that avoid the task’s words.

McGarrigle and Donaldson spotted something about the standard test. The question is the same before and after the change. But in between, the experimenter changes the length of a row on purpose.

They suggested that the child may read that action as a hint. The experimenter seems to be interested in length. So the child answers about length. The words, though, ask about number.

If they were right, more children should succeed when the change looks accidental. The usual explanation is that children are swayed by how the rows look. The longer row looks bigger. That explanation predicts no difference between the two kinds of change. The rows look the same either way.

A change planned by the experimenter cannot be a real accident. So the authors made one seem accidental. They used a small teddy bear, operated by the experimenter, that seemed to mess up the game. Children readily treat toys as if they act on purpose.

These aims can be put as two questions.

  • The Effect: Do more children judge correctly when a naughty teddy seems to change the materials by accident than when the experimenter changes them on purpose?
  • The Reason: Does the experimenter’s deliberate action lead children to misread the question?

Method

Design and Variables

In outline, each child was shown counters or string and gave a judgement (McGarrigle & Donaldson, 1974). Then the materials were changed, and the child judged again. The change was made in one of two ways.

Every experiment has two kinds of variable. The independent variable is what the researcher changes. The dependent variable is what the researcher measures.

  • Independent Variable: How the change was made. The experimenter did it directly (the intentional condition), or teddy seemed to do it while “spoiling the game” (the accidental condition).
  • Dependent Variable: Whether the child’s judgement after the change was correct. A child counts as a conserver if all four judgements after the change were right.
  • Design: Every child took part in both conditions. This is called a repeated-measures design.

The labels used here (repeated measures, independent and dependent variable) and the two questions in Aim are ours, not the paper’s.

The children were split into two groups, balanced for sex and age. Group I did the accident judgements first. Group II did the deliberate ones first.

This is called counterbalancing. It spreads any effect of order evenly across the conditions.

Within each group, half the children did the number tasks first and half the length tasks first. For an exam, name the way the change was made as the independent variable. Name whether the judgement was correct as the dependent variable.

Simply Psychology Answers
Instant answers built on 19+ years of trusted expertise.

Participants

The sample is the group of people who take part. Eighty children took part, with a mean age of 5 years 4 months. The youngest was 4 years 2 months and the oldest 6 years 3 months.

Forty attended Edinburgh nursery schools. Forty attended a local state primary school.

The primary children were older, with a mean age of 5 years 10 months, against 4 years 10 months for the nursery children. There were equal numbers of boys and girls in each sample.

Materials and the Four Situations

The materials were counters, lengths of string, and a small teddy bear 3 inches tall. A box hid the bear.

Each child met four situations, two of number and two of length. The child was asked a question before and after the change.

  • Number, Equal: Four red and four white counters in two equal rows. One row was pushed together to make it shorter. The right answer is that the rows are still the same.
  • Number, Unequal: Rows of four and five counters. The child was asked which row had more. The right answer is the row of five.
  • Length, Equal: Two 10-inch strings side by side. One was bent into a curved shape, a crescent. The right answer is that they are still the same length.
  • Length, Unequal: A 10-inch string and an 8-inch string. The child was asked which was longer. The right answer is the 10-inch string.

The Procedure

At the start, only the box was on the table. The experimenter lifted out the teddy bear and showed him to the child. He explained that teddy was very naughty and might escape from his box to “mess up the toys” and “spoil the game”.

  • Intentional Condition: Teddy did not make the change. After the first judgement the experimenter said “Now watch…”. He rearranged the materials with one direct action and asked again.
  • Accidental Condition: After the first judgement, the experimenter pretended to be alarmed (“Oh! look out, he’s going to spoil the game”). He moved teddy over the materials, so they looked messed up by chance.

The experimenter always changed the material furthest from the child, to give teddy more time before the child could step in. He gave no feedback. The same material was changed in both conditions.

Four times in the accident condition, the child put the materials back before answering. That situation was restarted with the child asked to hold teddy “prisoner”, so the change stayed in place.

The children were not asked to explain their answers. The authors thought that asking would mean more complex exchanges between experimenter and child. These might have affected later answers.

Results

The Headline Results

Children were right far more often when the change seemed accidental.

  • After the Accident: About 72% of judgements were correct (71.9%).
  • After the Deliberate Change: About 34% were correct (33.7%).
  • Conservers: 50 of the 80 children got all four judgements right after the accident. Only 13 did after the deliberate change.

The authors summed this up as nearly three-quarters correct against one-third. They said the conclusion held with a looser rule than four out of four.

Order Mattered

The percentages differed by group.

GroupAfter the accidentAfter the deliberate change
Group I: accident first86.9%40.0%
Group II: deliberate change first56.9%27.5%

Doing the deliberate change first affected the later answers. This is called an order effect. Group II did worse after the accident than Group I did. The gap was too big to put down to chance.

So the authors looked at each group’s first four judgements only. Group I met the accident first, and 30 of 40 children were conservers. Group II met the deliberate change first, and only 5 of 40 were.

These were different children, not the same ones twice.

Age and the Children’s Own Words

The primary-school children scored higher than the nursery children in every condition.

So scores still rose with age in this sample. The paper reports no statistical test of the age difference.

Some Group II children seemed to treat teddy’s mess as if it were aimed at changing the length of a row. One child chose the unbent string as longer “because you bent that one doing it and it’s the shortest one”.

Why Did the Accident Help?

Reading the Experimenter’s Hints

In the early stages of learning language, young children work out what a speaker means partly by watching what the speaker does.

When the experimenter changes a row’s length on purpose, the child behaves as if the question were about length.

When the change seems accidental, there is no such hint. The child can answer about number.

In the string task, the child compared the unbent string with some feature of the crescent, usually the distance between its sides, when the experimenter shaped it. The child could compare the right feature when teddy shaped it.

This is the authors’ explanation. It remains one candidate among several.

Why Not Perception or Timing?

In both conditions the child faced the same misleading arrangement.

The authors judged that an explanation based on how things look could not explain the difference by itself.

The gap between the first and second judgement was about 30 seconds in the accident condition and 3 to 5 seconds in the deliberate one.

A separate study with a small number of children found no sign that this explained the better results.

Conclusion

McGarrigle and Donaldson drew four conclusions.

  1. Standard Tasks Underestimate Children: Traditional ways of testing conservation seriously underestimate what children know.
  2. Success Depended on the Change: Most of these children were conservers after the accident but mostly were not after the deliberate change.
  3. It Is About Interpretation: The results need an explanation in terms of how the child interprets the experimenter’s words and actions.
  4. Look at the Setting: More attention should be paid to the interaction between adult and child before judging a child’s competence.

The study does not say that children do not develop.

Thirty of the 80 still did not get all four right after the accident. The primary children also did better than the nursery children in every condition.

Strengths

The study has four main strengths:

  1. Same Children, Both Conditions: Differences between children cannot explain the contrast. Order was counterbalanced, so no condition always came first.
  2. Honest About Order: The authors reported the order effect openly. That is why they also analysed the first four judgements.
  3. Tested an Explanation: The usual explanation predicts no difference between conditions, because the rows look the same. The study found a difference.
  4. Repeated by Others: Other teams found the same pattern with other materials and children, though Light et al. found lower overall scores (Hargreaves et al., 1982; Light et al., 1979).

Limitations

Before working through each one, here is a one-line summary of the main limitations:

  1. More Than Intention Differed: The accident condition added teddy’s involvement, pretend alarm and a longer gap.
  2. Order Effect: Children who met the deliberate change first did worse after the accident than children who met the accident first.
  3. Right Answers, No Reasons: Children were not asked to explain, so a right answer may not show real understanding.
  4. A Staged Accident: The accident was acted out, and children may have responded to the experimenter’s cues.
  5. Small Sample: Eighty children from one city, tested before the paper was published in 1974.
  6. Development Remains: The study questions how failure is read, not whether children develop.

More Than Intention Differed

The accident condition added teddy’s involvement, pretended alarm and a longer gap. The gap was 30 seconds, against 3 to 5 seconds for the deliberate change. The authors ran a separate small study on the gap. They did not separate the other differences.

So the study cannot show that intention alone made the difference. A simple “teddy makes it a fun game” explanation is also possible.

Later variants help a little.

A monkey also worked, and a change that merely seemed irrelevant worked best (Hargreaves et al., 1982). That points to the meaning of the change, not to teddy himself. The abstract does not say whether the gap was controlled.

The Order Effect

Group II’s accident score was lower than Group I’s, 56.9% against 86.9%. Earlier experience of the deliberate change seems to have dragged it down.

The 50-versus-13 comparison uses the same children twice. The cleaner comparison, 30 of 40 against 5 of 40, uses different groups of children.

The paper does not say how children were put into groups, beyond balancing sex and age. Both comparisons support the effect. Neither is a perfectly clean test.

Right Answers, No Reasons

The authors did not ask for explanations. So a child counts as a conserver here by choosing the right answer four times. Piagetian researchers often asked for reasons too. Some would call this a looser test.

Guessing is an unlikely explanation, because four right in a row would happen only about 1 time in 36 (our calculation). But a right answer cannot show how the child got it.

Neilson et al. (1983) tested this. They ran a similar procedure with 128 children and asked half of them to explain. The teddy effect appeared again. Overall, 80% of judgements were right after the accident and 34% after the deliberate change.

But few right answers came with a logical reason. Only about a quarter did, after the accident.

Once a right answer also needed a logical or empirical reason (mostly counting), the gap between the conditions was no longer too big to put down to chance. This held on all four tasks.

They did not say the children lacked reasoning (Neilson et al., 1983).

More Simply Psychology Answers
Instant answers built on 19+ years of trusted expertise.

A Staged Accident

Ecological validity means how well findings apply to real life. Each child was tested alone by an adult. The accident was acted out.

A true accident cannot be built in, the authors admit. So the children had to believe a performance. It may not show how children think about amounts in everyday life. That is our reading.

Demand characteristics are cues that suggest to participants what is expected. The paper does not say whether the experimenter knew the hypothesis. So effects of the experimenter cannot be ruled out.

There is an irony here. The study says cues from the experimenter shape children’s answers, but its accident condition is itself a set of cues from the experimenter.

Small Sample, One City, One Era

Generalisability means how far results apply beyond the people, place and time studied. The sample was small and local. It was 80 children from Edinburgh nursery schools and one state primary school, tested before the paper was published in 1974.

A sample that does not represent everyone is called sample bias.

The paper says nothing about social class, ethnicity or language background. So we cannot say how widely the effect applies. The replications had samples of 60 to 80 children (Hargreaves et al., 1982; Light et al., 1979).

Development Remains

The strongest claim the data support is that the standard task underestimates conservation. It is a claim about how failure is read, not about whether development happens.

Thirty children still did not get all four right after the accident.

The primary children outperformed the nursery children in every condition. The authors argued that traditional procedures may underestimate children’s abilities. They did not say that differences between ages are an illusion.

Ethical Issues

The accident was staged, so the accident condition involved a mild pretence towards very young children. The task itself was harmless. The paper does not report whether the children’s parents agreed or what the children were told afterwards.

  • Informed Consent: Informed consent means agreeing to take part after being told what will happen. The paper does not report whether the children’s parents agreed.
  • Debriefing: Debriefing means explaining the study afterwards. The paper does not say what the children were told.
  • Deception: Deception means misleading participants. The pretended accident was a mild pretence. Calling it deception is our reading.

AQA credits these ways of dealing with such issues.

  • For Consent: Explain what people will take part in and ask if they are happy. If they are under 16, ask their parents too.
  • For Debriefing: Afterwards, explain what they took part in, why, and how the results will be used.
  • For Deception: Tell participants as soon as possible if they were lied to.

Exam Questions

What Examiners Look For

AQA’s specification names this study. It is in Paper 1, section 3.1.3, Development. It lists “McGarrigle and Donaldson’s ‘naughty teddy study'” beside “Hughes’ ‘policeman doll study'”. Both come after “Reduction of egocentricity, development of conservation”.

AQA has set the study in Paper 1 twice as a describe-and-evaluate question. It also set a question on conservation itself.

  • June 2021, Question 17 (6 marks): Describe and evaluate the study. Three marks are for describing and three for evaluating.
  • June 2021, Question 16 (4 marks): Describe Piaget’s concept of conservation, using a child called Ibrahim.
  • June 2023, Question 18 (9 marks): Describe the study, then evaluate its research method. Four marks are for describing and five for evaluating.

The examiners’ report on June 2023 gave this advice.

  • Describe All Four Parts: Give the aim, method, results and conclusion. Some students who labelled their margin A, M, R and C seemed to find it helped them include all four.
  • Do Not Stop at the Method: Several students described the method in great detail. They left out the results and conclusion.
  • Know the Research Method: Do not confuse the research method with the experimental design or sampling methods. Those answers earned no evaluation credit.
  • Link Evaluation to the Method: Evaluating the study without linking it to the research method earned at most 1 evaluation mark.
  • Evaluation Is Half the Marks: Roughly half the marks in a question of this kind are for evaluation.

The real questions are quoted below. The practice questions after them are our own, in the style of the exam. The “Imagine” questions use invented data.

Real Question 1: June 2021, Question 17 (6 marks)

Question: “McGarrigle and Donaldson investigated conservation in the ‘naughty teddy study’. Describe and evaluate this study.” [6 marks]

Model answer (ours): The aim was to test whether young children can conserve when a change to the materials looks accidental. Eighty children aged 4 to 6 judged number and length before and after a change.

In one condition the experimenter changed the materials on purpose. In the other, a naughty teddy bear seemed to do it by accident.

About 72% of judgements were right after the accident and about 34% after the deliberate change. The primary-school children scored higher than the nursery children. So traditional tests may underestimate what children know.

One strength is that the study challenged Piaget’s view of when conservation develops.

One weakness is that 30 of the 80 children still did not get all four right after the accident. A second is that the older children came from one primary school and the younger ones from nursery schools. So education may explain the age difference.

  • Mark scheme: Marked by levels, with 3 marks for description and 3 for evaluation. Methodological issues are credited if they are not generic.
  • Description Credited: The aim was to test whether young children can conserve when changes look accidental. Eighty children aged four to six judged two identical rows of counters.
  • Description Credited: The teddy accidentally spread out one row, and the children were asked again. Over 60 per cent were right, and older children did better. So children under seven can conserve.
  • Evaluation Credited: It challenged Piaget’s view that conservation does not develop until about seven. A replication found lower results, though still better with the teddy.
  • Evaluation Credited: Over 30 per cent still failed to conserve. The older children came from one primary school and the younger from nursery schools, so education could be an extraneous variable.
  • Evaluation Credited: Some children may not have noticed the change. The setting was strange and the experimenter an unfamiliar adult.

Real Question 2: June 2023, Question 18 (9 marks)

Question: “McGarrigle and Donaldson investigated the development of conservation in the ‘naughty teddy study’. Describe this study. Evaluate the research method used in McGarrigle and Donaldson’s study.” [9 marks]

Model answer (ours): McGarrigle and Donaldson investigated conservation. In Piaget’s test the experimenter changes a row of counters on purpose. The authors thought this might make children misread the question.

Eighty children aged 4 to 6 took part. Each judged number and length before and after a change. In one condition the experimenter changed the materials on purpose. In the other, a naughty teddy bear seemed to do it by accident.

About 72% of judgements were right after the accident and about 34% after the deliberate change. Fifty of the 80 children got all four right after the accident. Only 13 did after the deliberate change. The authors concluded that traditional tasks underestimate what children know.

AQA treats the research method as a laboratory experiment with tight control. The paper itself does not say where it was run.

Every child met both conditions, with the same materials and questions. This removes many extraneous variables. So the researchers can be more confident that how the change was made affected the result.

The procedure was standardised. The experimenter gave no feedback. This means the study can be repeated, and the reliability of the findings can be checked with other children.

But the tasks were artificial. An adult moved counters, acted out an accident with a teddy bear and asked questions. This can reduce validity. It is hard to generalise the findings to a normal setting.

  • Mark scheme: Levels 7 to 9, 4 to 6 and 1 to 3. Four marks are for describing and five for evaluating the research method.
  • Description Credited: The same points as June 2021 above: aim, 80 children, counters, the accidental teddy, over 60 per cent right, older children better, and the conclusion.
  • Evaluation Credited: A laboratory-based study under highly controlled conditions. Control removes extraneous variables, so the researcher can be sure the independent variable affected the dependent variable.
  • Evaluation Credited: Standardised procedures allow replication and a check on reliability with different groups.
  • Evaluation Credited: Artificial tasks, such as an adult moving counters and asking questions, can reduce validity. High control makes behaviour artificial and findings hard to generalise.
  • Linking Rule: Evaluation with no link to the research method earns at most 1 evaluation mark.

Real Question 3: June 2021, Question 16 (4 marks)

Question: “Briefly describe Piaget’s concept of conservation. Refer to Ibrahim in your answer.” [4 marks]

In the mark scheme, Ibrahim is four. He thinks a tall, thin glass of juice holds more than a short, wide glass of the same juice.

  • Mark scheme: 2 marks for knowledge and 2 for applying it to Ibrahim.
  • Credited: Conservation means knowing the quantity stays the same when the appearance changes. Piaget said children develop it at about seven.
  • Applied: Ibrahim cannot yet conserve, because he is in the pre-operational stage.

Where the Mark Scheme and the Paper Differ

The mark schemes are short. Some credited points differ from what the paper reports.

  • Over 60 Per Cent: This fits 50 of 80 children (62.5%) with all four right. The mean of correct judgements was 71.9%. AQA does not say which measure.
  • The 16 Per Cent: Some notes give 16%. That almost certainly comes from 13 of 80 children (16.25%) after the deliberate change, our inference.
  • Over 30 Per Cent Failed: This fits 30 of 80 children (37.5%) who did not get all four right after the accident. “Failed to conserve” is the mark scheme’s reading.
  • Nursery and Primary: This matches the paper. Forty children came from Edinburgh nursery schools and 40 from one state primary school.
  • Lower Results: The mark scheme names no study. Light et al. (1979) found a lower overall level of conserving judgements than the original.
  • Not in the Paper: As we read it, the paper says nothing about whether children noticed the change, or whether the setting was strange or the adult unfamiliar.
  • The Venue: As we read it, the paper does not state where the children were tested. AQA calls the study laboratory-based.

What should you write? The mark-scheme points are credited, so you can use them. Offer the points the paper does not report as possible weaknesses. Say “may” or “could”.

Check the Details in Your Notes

Revision notes often differ from the paper. Check these against the paper.

  • Teddy: Some revision notes call him a glove puppet, but the paper describes a small teddy bear operated by the experimenter.
  • Materials: Counters and string, not sweets or beads.
  • Tasks: Number and length, not number only.
  • Conditions: Every child did both, in different orders.
  • Date: The paper is dated 1974. Some sources say 1975.

Practice Question 1: Calculate a Percentage (2 marks)

Imagine a study, with invented numbers that are not McGarrigle and Donaldson’s. A researcher tests 20 children after the experimenter changes a row of counters on purpose. Seven are right. The same 20 children are tested after a toy seems to change a row by accident. Sixteen are right.

Question: Calculate the percentage of children who were right in each condition. Show your working. [2 marks]

Model answer: deliberate change, 7 ÷ 20 × 100 = 35%. Accident, 16 ÷ 20 × 100 = 80%.

  • Mark scheme: 1 mark for each correct percentage, with working shown.

Practice Question 2: Design an Experiment (6 marks)

Question: Design an experiment to test whether children do better when a change to counters seems accidental. [6 marks]

Model answer: I would use two groups of four-year-olds. The independent variable is how the change is made. One group sees the experimenter push a row together. The other sees a toy do it by accident.

The dependent variable is whether the child says the rows are still the same in number.

Everyone gets the same counters, question and words. There is no feedback. This is a standardised procedure: it is the same for every child. It makes the test fair, because only the way the change is made differs.

I expect more children to be right after the accident. The experimenter’s action gives no hint that the question is about length.

  • Mark scheme: 1 mark each for the independent variable, the dependent variable, a standardised procedure and why it helps.
  • Expected Result: 2 marks, 1 for the prediction and 1 for the reason.

Practice Question 3: Ethical Issues and How to Deal with Them (4 marks)

This is our own question, in the style of AQA’s. AQA’s real ethics questions sit inside research-design questions about an experiment, not about this study.

Question: Name two ethical issues that a researcher should consider in a study like the naughty teddy study. Suggest one way of dealing with each. [4 marks]

Model answer: One issue is informed consent. The paper does not report parental consent. A way of dealing with this is to explain what the children will take part in and ask their parents if they are happy.

Another issue is debriefing. The paper does not say what the children were told afterwards. A way of dealing with this is to explain afterwards what they took part in and why.

  • Mark scheme: In AQA 2020, each issue and way pair earns 2 marks if clear, accurate and in context, and 1 if limited or muddled.
  • Credited Issues Include: Informed consent, deception, right to withdraw, protection from harm, confidentiality and debrief. Parental consent is credited in AQA 2020.
  • Credited Ways Include: Explain the study and ask if people are happy. Tell people if they were lied to. Explain the study afterwards. Ask parents if under 16.

Later Research

The idea is still being tested.

Repeats in the 1970s and 1980s

Other teams repeated it in the 1970s and 1980s. In one experiment, Light et al. (1979) tested 60 children, with an errant teddy bear making the change. They confirmed the higher success rate after an accident, though overall scores were lower than in the original.

Hargreaves et al. (1982) found the same with a mischievous monkey. Both are known here from abstracts only.

Recent Studies

Recent studies ask a similar question. Baratgin et al. (2020) tested 3-year-olds on a false-belief task, in which the child must predict what a person will do who holds a belief the child knows is false. They did better when a robot asked the question than when a human did.

The logic is similar. Two things changed at once: the questioner and the child’s role. So the abstract cannot show which mattered.

The authors argued that young children may fail because of how the question is asked (Baratgin et al., 2020; abstract only).

Children who carried out the change themselves recognised conservation more often than children who watched an adult (Lozada & Carro, 2016). A study of 460 children aged 4 to 7 found the same, with hands-on action, on four conservation tasks (Jia et al., 2025; abstract only).

How Does It Compare with Other Explanations?

Piaget’s Explanation

The usual reading of failure on the standard task is that the child is swayed by how the rows look. Piaget’s theory calls this centration, the habit of fixing on one eye-catching feature, such as the length of a row, within the preoperational stage.

The teddy study accepts that children fail the standard task. It offers a different cause. So it challenges how the failure is explained, not the failure itself.

Rose and Blank’s Repeated Question

Rose and Blank (1974) tested another idea. They thought that being asked twice, before and after the change, may tell a child to change the first answer. First-grade children made fewer errors when asked only once, and also on a standard task a week later (abstract only).

So Rose and Blank blamed the repeated question. McGarrigle and Donaldson blamed the experimenter’s deliberate action. The two ideas fit together, because both say that children read social cues. A repeated question may act as a demand characteristic, a cue that the first answer was wrong.

More Simply Psychology Answers
Instant answers built on 19+ years of trusted expertise.

References

Baratgin, J., Dubois-Sage, M., Jacquet, B., Stilgenbauer, J.-L., & Jamet, F. (2020). Pragmatics in the false-belief task: Let the robot ask the question! Frontiers in Psychology, 11, Article 593807. https://doi.org/10.3389/fpsyg.2020.593807

Hargreaves, D. J., Molloy, C. G., & Pratt, A. R. (1982). Social factors in conservation. British Journal of Psychology, 73(2), 231–234. https://doi.org/10.1111/j.2044-8295.1982.tb01805.x

Jia, W., Deng, X., Yang, J., Wang, R., Sun, X., & Xiao, E. (2025). Embodied action enhances conservation reasoning: Age and task effects in early childhood. Early Child Development and Care, 195(9–10), 806–824. https://doi.org/10.1080/03004430.2025.2530043

Light, P. H., Buckingham, N., & Robbins, A. H. (1979). The conservation task as an interactional setting. British Journal of Educational Psychology, 49(3), 304–310. https://doi.org/10.1111/j.2044-8279.1979.tb02430.x

Lozada, M., & Carro, N. (2016). Embodied action improves cognition in children: Evidence from a study based on Piagetian conservation tasks. Frontiers in Psychology, 7, Article 393. https://doi.org/10.3389/fpsyg.2016.00393

McGarrigle, J., & Donaldson, M. (1974). Conservation accidents. Cognition, 3(4), 341–350. https://doi.org/10.1016/0010-0277(74)90003-1

Neilson, I., Dockrell, J., & McKechnie, J. (1983). Justifying conservation: A reply to McGarrigle and Donaldson. Cognition, 15(1–3), 277–291. https://doi.org/10.1016/0010-0277(83)90047-1

Rose, S. A., & Blank, M. (1974). The potency of context in children’s cognition: An illustration through conservation. Child Development, 45(2), 499–502. https://doi.org/10.2307/1127977

Olivia Guy-Evans, MSc

BSc (Hons) Psychology, MSc Psychology of Education

Associate Editor for Simply Psychology

Olivia Guy-Evans is a writer and associate editor for Simply Psychology, where she contributes accessible content on psychological topics. She is also an autistic PhD student at the University of Birmingham, researching autistic camouflaging in higher education.


Saul McLeod, PhD

Chartered Psychologist (CPsychol)

BSc (Hons) Psychology, MRes, PhD, University of Manchester

Saul McLeod, PhD, is a qualified psychology teacher with over 18 years of experience in further and higher education. He has been published in peer-reviewed journals, including the Journal of Clinical Psychology.