all children act like that. Baselessly.
I didn't say all, I said they generally tend towards altruism, or at least not deliberately harming others, which is a hell of a lot better than modern behavior: http://www.pitt.edu/~toddlers/ESDL/Svetlova,etal_ToddlersProsocialBehavior.CDInPress09.pdf
There's much less to that study than you imply. There are various flaws, and what is worse, even some blatant misrepresentations of the data by the researchers in the abstract and in their conclusions.
The three tasks called "altruistic" required lending a hair-clip recently given to them to an adult with messy hair, or their favorite blanket to a "cold" adult, or their favorite toy to a sad adult... but no understanding that the adult would keep it afterward. Altruism is self-sacrifice, but children giving temporarily something that they don't need is dubiously a real sacrifice. One of these tasks, moreover, is merely giving away something that they themselves recently acquired, which is entirely a matter of easy come, easy go.
Finally, here's the part where you need to start paying attention.
The fact that instances of costly helping in the
current study were quite low and often occurred in
response to an adult’s direct and explicit request
suggests that toddlers’ helping responses are unlikely
to be genuinely altruistic. Altruistic motivation
appears to be a later developing phenomenon, which
may build upon the more basic prosocial motivation
emerging and developing in toddlerhood.
The study itself factually concluded one thing, and researchers tacked on an unsupported belief that altruism was "developing" from prosocial behaviours, yet the selfsame study makes clear the 30-month olds already understood the emotional needs of others in the non-costly helping task, and were willing to give items that weren't their own--but not willing to give their own items. Prosocial, yes; altruistic, consciously no. That's the whole of the actual data.
The researchers then overstretch by drawing conclusions outside their own data set. That is unforgiveable. They actually remark on ages for which they did not test. This is a bad study, and only those who superficially read abstracts could be taken in by it. It actually confirms extreme reluctance to engage in activity even remotely altruistic.
Children, by the actual data, are not altruistic. The 30 month-olds engage in at least one altruistic behaviour only in 18% of the time, and often only after directly being asked for the item. How often was that 18% just giving away the hairpin that recently was given to them? Probably a significant amount. The rest is researchers weakly defending what they wanted the data to show. Since it wasn't there, they suppose without supporting data that the magical moment of altruism comes later. They put that unsupported assertion in the abstract, and you incautiously accepted that as being a data-driven conclusion. It wasn't.
What if the real conclusion is true altruism never comes at all? Or is entirely a learned social behaviour based on calculated future reciprocity? What does that mean for human nature? For systems predicated on people naturally acting altruistic?