Abstract: Categorization learning was investigated in 2 horses (Equus caballus). Both horses learned to select a 2-dimensional black stimulus with an open center instead of a filled stimulus in a 2-choice discrimination task. After a criterion of 10 out of 10 correct responses in a random series for 2 consecutive sessions was reached, 15 additional pairs of open-center versus filled stimuli were tested. Each was run to criterion and then incorporated into sessions of randomly mixed problems. Both horses solved the 1st problem by simple pattern discrimination and showed evidence of categorical processing for subsequent problems. New pairs were learned with few or no errors, and correct responses on novel trials were significantly above chance. These results suggest that the horses were making their selections on the basis of shared characteristics with the training stimuli and were using categorization skills in problem solving.

Abstract: In most studies of social learning in animals, no attempt has been made to examine the nature of the strategy adopted by animals when they copy others. Researchers have expended considerable effort in exploring the psychological processes that underlie social learning and amassed extensive data banks recording purported social learning in the field, but the contexts under which animals copy others remain unexplored. Yet, theoretical models used to investigate the adaptive advantages of social learning lead to the conclusion that social learning cannot be indiscriminate and that individuals should adopt strategies that dictate the circumstances under which they copy others and from whom they learn. In this article, I discuss a number of possible strategies that are predicted by theoretical analyses, including copy when uncertain, copy the majority, and copy if better, and consider the empirical evidence in support of each, drawing from both the animal and human social learning literature. Reliance on social learning strategies may be organized hierarchically, their being employed by animals when unlearned and asocially learned strategies prove ineffective but before animals take recourse in innovation.

Abstract: Abstract: The relationship between degree of social bonding (extent of association among individuals) and level of aggression in ruminants is unclear. I examined social bonding and aggression in three groups of female Roosevelt elk (Cervus elaphus roosevelti) over 2 years. I hypothesized that when animals are socially bonded, bouts of aggression will be won by the individual initiating the aggression, occur quickly, and involve little physical contact, and the level of aggression does not correlate with group size. The degree of social bonding was high among individuals in all groups. Dyads of known individuals were together >80% of the time. A permutation analysis indicated that groups with the observed sizes had <0.001 chance of random association, except on one occasion when the probability was 0.72 for one group. Using focal-animal sampling, aggressive interactions were won 72% of the time by the initiator, occurred quickly (<5 s), and involved little physical contact, and the level of aggression was not correlated with group size. The level of aggression was, however, significantly lower in one of the groups. This group may have had access to more abundant food resources than the other groups. Socially bonded elk conducted aggressive interactions in a fashion that did not disrupt social stability. Résumé : La relation entre le degré de liaison sociale (importance des associations entre individus) et l`agressivité n`est pas claire chez les ruminants. J`ai étudié les liaisons sociales et l`agressivité chez trois groupes de femelles du Cerf de Roosevelt (Cervus elaphus roosevelti) pendant 2 ans. J`ai posé en hypothèse que, chez les animaux liés socialement, la victoire devrait être emportée par l`individu qui entreprend l`agression, l`agression devrait être de courte durée, se faire avec peu de contacts physiques et la fréquence des agressions ne devrait pas être liée à la taille du groupe. Des paires d`individus passaient plus de 80% de leur temps ensemble. Une analyse des permutations a démontré que, chez les groupes des tailles observées, la probabilité d`une association aléatoire était de moins de 0,001, sauf en un cas où cette probabilité a été évaluée à 0,72 chez un groupe. Par échantillonnage directionnel, j`ai observé que les interactions agressives étaient gagnées par l`individu attaquant 72% du temps, étaient de courte durée (<5 s), se faisaient avec peu de contacts physiques et leur fréquence n`était pas reliée à la taille du groupe. Il y avait cependant moins d`agressivité chez l`un des groupes. Il se peut que ce groupe ait eu accès à plus de ressources alimentaires que les autres. Chez les cerfs liés par des liens sociaux, l`agressivité ne se manifeste pas de façon à déséquilibrer la stabilité sociale.

Abstract: The evolution of the horse began some 65 million years ago. The horse"s survival has depended on adapative behaviour patterns that enabled it to exploit a diverse range of habitats, to successfully rear its young and to avoid predation. Domestication took place relatively recently in evolutionary time and the adaptability of equine behaviour has allowed it to exploit a variety of domestic environments. Though there are benefits associated with the domestic environment, including provision of food, shelter and protection from predators, there are also costs. These include restriction of movement, social interaction, reproductive success and maternal behaviour. Many aspects of domestication conflict with the adaptive behaviour of the horse and may affect its welfare through the frustration of highly motivated behaviour patterns. Horse behaviour appears little changed by domestication, as evidenced by the reproductive success of feral horse populations around the world.