Comfort objects

Why does a filthy worn blanket soothe a toddler when an identical clean one will not?

A parent who has lost the blanket once will buy a spare. It is the same product, the same size, the same colour, and it is cleaner. It does not work, and the child's refusal is not sulking — it is immediate and unmistakable, the way you would react to being handed a stranger's coat.

The puzzle is that the substitution ought to be perfect. If what soothes is softness, warmth, and a familiar texture, the replacement supplies all three and supplies them better. So the comfort cannot be coming from the properties we assumed. What is the child actually holding on to?

Try the properties account seriously first, because it deserves one honest test. Suppose the blanket works because of how it feels. Then the strongest predictor of comfort would be sensory match, and a fresh identical blanket would be at least as good. Suppose instead it works because of how it smells — worn, unwashed, saturated with the household. That is a real candidate, and familiar odour genuinely is calming; it also has the merit of explaining why washing the thing is a household crisis.

But test it. Wash the original blanket, and a child who is distressed will usually still take it back afterwards, sometimes grudgingly, and settle. Now hand that same child the unwashed spare, worn in and smelling of the same house. Parents report the two are not interchangeable. If smell were doing the work, the ranking should be the other way round.

So neither the feel nor the smell is the thing being tracked. What survives washing, and what a duplicate cannot copy?

The answer is history — the fact that this is the same object that has been there all along. Which means the child's representation of the blanket is not a description at all. A description picks out anything that matches it; two blankets satisfying the same description are two instances of one kind, and either will do. What the child holds is a representation of a particular individual, identified by continuity through time rather than by properties. Philosophers call this the difference between a type and a token, and it is the whole of the puzzle: we assumed the child wanted a type and they wanted a token.

Is there evidence beyond parental anecdote? Yes, and it is neat. Hood and Bloom, in 2008, showed preschoolers a machine they were told could copy objects — a box that appeared to produce a perfect duplicate. Children were generally content to take a copy of an ordinary favourite toy. Children who had an attachment object were substantially more likely to want their original back rather than the copy, even though the copy was, by construction, indistinguishable. The refusal tracked the object's identity, exactly where a properties account predicts indifference.

Now ask why a mind would represent an object that way, since it is the more expensive option. Consider what the object is for. Winnicott, writing in 1953, called these transitional objects: things a child uses to hold on to the security of a caregiver while learning to be apart from them. Notice that a caregiver is itself a particular individual — a mother who was swapped for a perfect duplicate would be no comfort either. An object standing in for a person has to inherit the person's logic. A stand-in that could be swapped for any qualitative match would not do the job, because the job is precisely to be the one continuous thing that is still there when the person is not.

Then the grime falls into place as a symptom rather than a cause. Wear and dirt are the accumulated evidence of that history — the visible record that this is the object that was there. They are not what makes it work. They are what makes it recognisably the one that works.

Three clarifications keep this from over-reaching.

First, this is not a claim that toddlers are little metaphysicians. The representation is not a belief they could state; it is simply how the object is filed, and the same filing shows up all over adult life — in relics, heirlooms, autographs, and the original manuscript that is worth more than a perfect scan. Researchers describe this as a kind of essentialism, treating an object as having an identity that its outward properties do not exhaust.

Second, attachment objects are common but not universal, and their absence is not a warning sign. Rates vary substantially across cultures and caregiving arrangements — unsurprisingly, since a portable stand-in matters more where infants sleep apart from their caregivers than where they are carried and co-sleeping. Reading the object as a developmental requirement inverts the relationship.

Third, an honest limit on what is settled. That identity rather than properties is being tracked is well supported. Exactly how the object acquires that status in the first place — whether it is bound up with a specific window of infancy, how much familiar smell contributes alongside identity, and why one particular blanket rather than another gets chosen — is much less nailed down, and I would not want to state a mechanism there that the evidence does not carry.

• How and when children stop needing the object physically present, and what replaces it.
• Whether the same identity tracking explains children's reactions to a repaired versus a replaced toy.
• Why the objects chosen are so often soft and portable, and whether that is about the object or about how it is used.

Every term the collection defines is gathered in the glossary.