Our field experiment shows that the innovation we studied
in captivity can spread in a wild population, in line with the
results obtained by Webster and Lefebvre (2001). However,
the success rate of pigeons (35 of 42 individuals in captivity
and eight of 12 flocks in the field) was much higher than that
obtained by even the most innovative species tested by Webster
and Lefebvre, the Carib grackle Quiscalus lugubris (six
of 15 in captivity and two in the field). Moreover, when we
look at results from the two species that are taxonomically
closest to our pigeons, the zenaida dove Zenaida aurita and
common ground dove Columbina passerina, success rates
are even lower; only one of 30 individuals (3%) solved the
problem in captivity and no dove opened the box in the
field (Webster and Lefebvre 2001). In a study of individual
learning and neophobia, Seferta et al. (2001) also noticed a
marked difference in performance between feral pigeons and
Zenaida aurita. When data on reversal learning in eight avian
species (taken from Gossette 1968) were examined, pigeons
again are outliers and perform much better than expected
from their innovation rate and brain size (Timmermans et al.
2000; Lefebvre and Bolhuis 2003). This apparent superiority
of pigeons is probably a result of artificial selection. All
pigeons in the New World descend from captive individuals
and are thus presumably bred for tolerance of captive conditions
and human proximity (Johnston and Janiga 1995). It
is not a coincidence that feral pigeons are often the species
chosen in captive experiments; they tolerate experimental
conditions much better than many other species, and consequently
perform well on a variety of tasks. It would be
important to conduct a similar study on a species that has
not gone through artificial selection to make sure the positive
correlation between innovation and social learning found
here is a general one.