In that video the chimp beats one human who thought he was smart but appears particularly bad at that test.
They are different tests, though. You don’t need sequential memory to win that test, you just need near photographic recall and the general knowledge of numerical order.
The paper [1] makes a kind of subtle distinction that these aren't sequences, because the information is all presented at once:
> We do not focus on how animals represent single stimuli, or many stimuli that are presented simultaneously. For these reasons, test paradigms that involve simultaneously presented arrays of stimuli are beyond the scope of this study [39, 40], as responding to simultaneous input does not require the recognition of temporal stimulus sequences, even if subjects perform behavior sequences in response to complex input [41]. This also applies to the well-known studies where chimpanzees learned to point to the location of up to nine numerals that were presented simultaneously (see [42, 43] for studies on chimpanzees, and [44–46] for further discussion about these results).
Indeed, the videos of chimps casually acing those tests shows that all the numbers are given at once, and only disappear when the 1 is touched -- the test is getting the positions of 2 through 9 in the right order without being able to see them. The authors of the paper seem to argue that this is different from memorizing information that is presented sequentially.
That said, this does feel like an incorrect finding for other reasons. For example, some gulls seem to rely heavily on scents to navigate, learning a route of thousands of kilometers by a sequence of scent landmarks [2]. This information is presented to them in sequence and seems like a counterexample to my layman's eye.
