Curiosity appears in nearly every school prospectus and nearly every early years framework in the country. It is one of those words that everybody uses and nobody has to define, which is a sign that it is doing rhetorical work rather than technical work.
So how does anybody study it? The answer is a small tour through experimental method, and it is worth taking, because understanding how a construct is operationalised is the fastest way to understand what a finding about it actually means.
Looking time, the workhorse of infant research
With pre verbal infants, the standard tool is looking. Show a baby something repeatedly until their attention drops off, then change it. If they look longer at the changed version, they have noticed the change.
This method built a large part of what is known about infant cognition. It has established that very young infants notice violations of physical expectation, distinguish quantities and register unfamiliar speech sounds.
What it measures, though, is attention to novelty, which overlaps with curiosity without being identical to it. A startled look is not a search. And looking time has its own difficulties: the effects are sometimes small, the coding requires judgment, and the field has had to reckon with how easily analytical choices influence results. Large scale collaborative replication efforts in infancy research have been informative and, in places, sobering.
Exploratory choice
With slightly older children, a better proxy is choice. Present two options, one whose properties are already known and one which is ambiguous, and record which the child investigates.
This is closer to the concept, because it captures the preference for information over reward. Studies of this kind have shown that children systematically favour the ambiguous option, and that they explore in ways that would resolve the ambiguity rather than randomly.
Its limitation is that it works over minutes in a room with a researcher. Whether the child who chooses the ambiguous box is also the child who takes a clock apart at home is a question the design cannot answer.
Counting questions
Once children talk, the obvious measure is question asking: how many, of what kind, and how they respond to the answers.
This has produced some of the most interesting work in the area, particularly on how children respond differently to genuine explanations and to non answers, which we discuss in why children ask why.
The problems are practical. Question rate depends heavily on setting, on how comfortable the child is, and on whether questions are welcome. A child in an unfamiliar room with a stranger asks fewer questions than the same child at home, and the difference is about the situation rather than about curiosity. Any measure that varies this much by context is measuring context as well.
Effort and the willingness to pay for an answer
A more recent approach borrows from decision research: make information cost something and see whether the participant will pay.
You can ask a child to wait, or to do something tedious, in order to find out an answer. Willingness to pay is a stronger indicator than willingness to accept information for free, and it distinguishes idle interest from something more like drive.
The trouble is that willingness to wait is also a measure of several other things, including self control, trust that the promised answer will arrive, and how much the child wants to please the adult in the room. Multiple constructs are being measured by one behaviour, which is a chronic condition in this field.
Scales and questionnaires
Finally there are rating scales, where adults or older children rate agreement with statements about interest and exploration.
Scales are cheap, work at large sample sizes, and can be used in longitudinal studies where behavioural measures are impractical. They are also the furthest from the behaviour. A teacher rating a child on curiosity is partly reporting the child's curiosity and partly reporting how the child expresses it, how much they like the child, and what curiosity looks like in that classroom.
When you read that curiosity predicts achievement, it is very often a scale that did the predicting, and scales rated by teachers who also produce achievement judgments are not independent measures.
Why this matters for reading claims
The practical consequence is that a sentence like curiosity predicts learning can be true and nearly empty at the same time, because the word is doing different work in different studies.
When you meet a claim about curiosity, the useful question is which proxy was used. Looking time tells you about noticing. Choice tells you about preference for information. Question counting tells you about verbal information seeking in a setting. Effort tells you about drive mixed with self control. Ratings tell you about the impression a child makes on adults.
Those are five different things. Any of them can be defended. Treating a finding about one as a finding about the whole is where a great deal of educational rhetoric comes from.
A word in defence of the word
None of this means curiosity is not real. The behaviour is unmistakable, in children and in adults, and it is one of the more striking features of the species. What it means is that it is a folk concept that research has had to carve up in order to study, and the carving has not produced a single clean joint.
Which is an ordinary situation in psychology and worth knowing about when you read the confident version. For more on how findings shrink under examination, see replication in plain words, and for the behaviour itself, what play actually does.