0. We are gratified to see that Gacsi et al. have been able to show that hand-reared adult wolves in Europe can follow momentary distal points without explicit training, as was previously demonstrated in wolves hand-reared in North America [ref 8 in the paper].

1. However, Gacsi et al.'s characterization of the prior study [8] is not accurate. They state that, "in contrast to the definition for momentary pointing [see 13] the pointing gesture was still visible when the subjects made their choice." This is incorrect. By the time each wolf made its choice, the experimenter's hands had returned to the midline [8].

2. Gacsi et al. further claim that the use of secondary reinforcement (a standard dog-training clicker) in [8], "probably affected the performance." The implication seems to be that the use of the clicker gave the animals an additional cue that made it unnecessary for them to rely on the human's pointing gestures. This cannot be the case. The two choice containers were identical at the time of selection, and the reinforcer was only presented after the canid had made its choice.

3. Gacsi et al., baited the pointed-to container with meat and the container not pointed to was not baited. We found in preliminary tests that at least one wolf could discriminate by smell alone between a baited container and a container around which the same meat had been extensively smeared and a smaller piece of meat trapped in place [8]. Thus prebaiting a target container with meat is an inadequately controlled method in studies in which something else (the human point) is the intended cue. Gacsi et al. reported that their animals showed no preference for the baited container on trials in which no pointing cue was offered. However, the dogs and wolves in Gacsi et al.'s study showed no more than a 20% bias towards the pointed-to container in experimental trials. If this preference were entirely due to odor, then the six control trials offered would be insufficient to pick up such a small preference. A 20% bias would imply around four out of six trials correct. With the variance in these data such a small effect could not be detected. And no controls were reported in Studies 1 and 2.

4. Gacsi et al. do not present data indicating that a single dog or wolf at any age was successful on this task. The only data presented are group means. For groups listed as successful, group mean performance ranges from approximately 65 to 70% correct over 10 to 20 trials. However, for an individual to be successful in Study 1, eight or more correct choices out of 10 are required (binomial test p <= .05). For the 4-month old animals that received 14 trials in Study 2, 78% would have to be correct for an individual to be successful; and at adulthood where the animals received 20 trials (Study 3), 70% correct would be required for statistical significance. Thus if each individual performed at the level of its group's mean, only at the oldest age group - and then only because of the larger number of trials run - would there be any successful individuals on this task at all.

5. Comparison of the different age groups is hindered by a lack of clarity concerning the past experiences of the adult wolves. It appears that the dogs in all three studies were different experimentally-naive animals. The wolves in study 1 were experimentally naive, as were three of the wolves in Study 2. The other four wolves in Study 2 were re-used from Study 1. The past experiences of the adult wolves in Study 3 are not reported. If canids learn to follow human points through experience with human hands indicating reinforcing consequences, the experienced wolves in Study 2 cannot be compared to the experimentally naive wolves in Study 2 but should be analyzed separately. Likewise the past experiences of the adult wolves in Study 3 are crucial to the claim that they were successful on the task "without pretraining."

6. The authors could address points 4 and 5 by making the individual trial data from each subject publicly available along with a note of the past testing and training experiences of each animal.

Monique A. R. Udell
Nicole R. Dorey
Clive D. L. Wynne
University of Florida.