Why orangutans can start from scratch
Nut-cracking looks like the kind of skill you would expect to be passed down by watching. Yet when twelve orangutans with no nut-cracking experience were given nuts and wooden hammers and shown nothing, at least four worked out how to crack nuts anyway, including both Sumatran and Bornean orangutans [1]. That result is striking because orangutans have never been seen cracking nuts in the wild, so this was not a skill they had picked up from their group.
The authors describe this as individual learning combined with certain kinds of non-copying social learning [1]. In plain terms, the animals did not need a demonstrator performing the full behavior; whatever social input helped them, it was not copying the nut-cracking action itself. This fits the idea that some great ape tool skills can emerge from a species' own problem-solving rather than from transmission of a technique.
When chimpanzees copy, and keep copying
Chimpanzees show the other side of the picture. In one study, two groups of captive chimpanzees were each taught a different way to use the same tool to get food, either lifting or poking with a Pan-pipe apparatus. Each group's technique was copied well enough to become a stable local tradition, and the chimpanzees tended to adopt the method used by the majority in their group, a pattern called conformity bias [5].
That is copying in a fairly strong sense: the animals did not just solve the problem, they solved it the way their group did. A separate study on a juice-extraction task examined how chimpanzees acquire a socially learned sequence of tool actions, comparing groups that had a model to learn from with those that did not [3]. The design speaks directly to the copying-versus-independent-discovery question, though the abstract alone does not settle how much the observing animals relied on imitation.
Copying is not all-or-nothing
Orangutans also appear capable of imitation in tool use, which complicates any simple story that they are the independent inventors and chimpanzees the copiers. A cognitive analysis of orangutan tool use considered what fully acquiring a tool skill by imitation would require, namely copying all the manipulations of the tool, and compared imitative tool use between chimpanzees and orangutans [4]. The framing suggests that imitation is a matter of degree and interpretation rather than a yes-or-no trait.
Chimpanzees and orangutans also actively seek information about what a tool can do. In one study, great apes including both species worked out tool properties in a task that required a two-step action plan before actual tool use [2]. That kind of investigation is not copying a technique from a demonstrator; it is the animal gathering evidence about the object in front of it, which can feed either copied or invented solutions.
What decides which mechanism shows up
The pattern across these studies is that the answer depends heavily on the setup. When orangutans were given tools and no demonstration, individual learning produced nut-cracking in a minority of animals, at least four of twelve [1]. When chimpanzees were given a working model and a group norm, copying produced distinct traditions [5]. Same broad question, opposite-looking answers, because the experiments asked different things of the animals.
Species history matters too. Orangutans have not been observed cracking nuts in the wild, possibly because their arboreal habits offer few chances to do so, which is why they were a reasonable test case for whether the behavior can arise without copying [1]. Chimpanzees, with nut-cracking observed in the wild, have more opportunity for techniques to spread and stabilize socially [1][5]. The evidence therefore does not crown one mechanism as dominant in great apes; it suggests that non-copying learning can generate a complex tool skill from scratch, while copying can lock a technique into a group tradition once it exists.
Sources used in this answer
Naïve orangutans (Pongo abelii and Pongo pygmaeus) individually acquire nut-cracking using hammer tools.
At least four of twelve nut-cracking-naïve orangutans spontaneously cracked nuts with wooden hammers without a demonstration, indicating the behavior can emerge through individual learning and certain non-copying social learning.
Information seeking about tool properties in great apes
Chimpanzees and orangutans sought information about tool properties in a task requiring a two-step action plan before actual tool use.
Acquisition of a socially learned tool use sequence in chimpanzees: implications for cumulative culture
A study compared chimpanzee groups exposed to a model with unexposed groups to examine how a socially learned tool-use sequence is acquired, bearing on copying versus independent discovery.
Orangutans' imitation of tool use: A cognitive interpretation
A cognitive analysis examined orangutan imitation of tool use, considering what fully acquiring a tool skill by imitation would entail and comparing imitative tool use between chimpanzees and orangutans.
Conformity to cultural norms of tool use in chimpanzees
Two groups of captive chimpanzees each copied a different tool-use technique (Lift or Poke) well enough for it to become a tradition, with conformity bias shaping which method they used.