Unlike in Experiment 1, the out-of-reach tools were all of different lengths (10, 15, 20 and 25 cm).
Our data were not consistent with the hypothesis that subjects 'discovered' the out-of-reach tools in this way, nor that they then relied on a sequence of random probes for tools, followed by probes for food.
(iii) For the one subject (Uék) who received the No-Food condition, we also looked at whether she probed the (empty) food tube and/or extracted out-of-reach tools.
In Tertiary trials, the longest tool was only reachable with the second longest out-of-reach tool, instead of all other out-of-reach tools as in Experiment 1.
Bird and Emery also presented their rooks with a choice of two out-of-reach tools; in this case both options were stones that had been previously rewarded, so the correct choice shown by their subjects was more revealing [24].
The potential for sequential tool use in New Caledonian crows is likely to relate to general problem-solving abilities, rather than to an evolutionary adaptation for acquiring out-of-reach tools (which seems an unlikely scenario in their native forested habitat, where sticks are readily available).
The tool lengths were changed to increase the difference between short and long tools, and the cross-pieces were added to make it easier for subjects to extract out-of-reach tools, following observations that one subject (Pierre) had difficulty with the original design.
On the Secondary-Long condition (where only one of the four out-of-reach tools was suitable), the subjects showed weak evidence for selectivity, using the correct, longest tool to probe first for the food on 5 of 10 trials (pooled across subjects; binomial test; z = 1.83, p = 0.08).
At the same time, it is presented with a readily available tool that is too short to reach the food, but sufficiently long to obtain another out-of-reach tool; only the second tool can reach the food.
Taylor et al. [8] presented a stone as well as an out-of-reach tool, and since their subjects rarely extracted the stone this demonstrates that tool extraction was not entirely random.
