Do capuchin monkeys (Cebus apella) show analogical reasoning when using tools?

Evidence for analogical abilities in animals is widely based on studies carried out with chimpanzees. Here we investigated how capuchin monkeys (Cebus apella) solve tool problems requiring the appreciation of the length of the tool(s) in relation to the length of the tube(s) containing a reward. This same experiment is underway with chimpanzees and children.
Eight capuchins were required to retrieve a reward placed at the end of a horizontal tube positioned inside an apparatus. In particular, they had to select the longest tool from different sets of three sticks differing in length (functional feature). To investigate whether seeing the consequences of the tool movements inside the tube (visual feedback) improved performance, four subjects were tested with a Transparent Apparatus (T-group), and four with an Opaque Apparatus (O-group).
Phase 1 included (i) a training in which each stick had a different handle and (ii) a transfer test in which the handles were switched among sticks, so that the functional tool had the same length than in the training, but a different handle. We expected subjects having learned to select the tool on the basis of the functional feature (i.e., length) to pass the transfer and subjects having learned to select the tool on the basis of the non-functional feature (i.e., handle) not to pass the transfer. Phase 2 included (i) a training in which the same sticks of Phase 1 were used with handles switched across them in every trial, and (ii) a transfer test in which all sticks had a same new handle and the baited tube was longer than in Phase 1. Therefore the stick which was the longest in the training became the medium one in the transfer. Consequently, success in the transfer required to use the relational rule of selecting the longest stick and not the stick whose length was previously rewarded. In both phases, training lasted until the acquisition of 12 correct out of 18 trials in two consecutive sessions (criterion). Transfer tests consisted of two 18-trial sessions.
In Phase 1, seven subjects did not pass the transfer and one belonging to the T-group succeeded and directly entered the transfer of Phase 2, in which it failed. In Phase 2, in which length/handle relation was systematically varied across trials, three out of eight subjects passed the transfer (one O-group subject and two T-group subjects). In the training of Phase 2, but not in the training of Phase 1, visual feedback seemed to enhance the appreciation of the functional feature of the tools: the T-group successful subjects needed 6 and 8 sessions whereas the O-group successful subject needed 15 sessions to reach criterion.
These results show that varied experience is fundamental to build knowledge. When capuchins were provided with a fixed association between stick length and stick handle they tended to base their selection on the most evident perceptual cue, i.e., the shape/colour of the handle. Conversely, they can learn to attend to the functional feature irrespective of the distracting feature if provided with variable associations between stick length and stick handle. In this way, three subjects out of eight solved the task by using the relational rule of selecting the longest stick. Finally, having visual access did not improve performance in a notable way. Comparisons with the performance of chimpanzees and human children in the same task will allow to better appreciate the differences in analogical reasoning in primates.