Abstract: I investigated the behavioural background of the way horses learn to follow humans in the “round pen technique” suggested by “horse whisperers” as a gentle method for initial horse training. Though the practicability of this technique has been adequately demonstrated in the past, the horses' behaviour during such training has not yet been documented in detail. In a riding arena, horses, that did not follow the trainer immediately, were chased away so that they galloped around the trainer. Galloping horses showed specific behaviour such as turning the ear to the trainer, chewing, licking, and stretching head and throat downwards. In subsequent trials horses needed to be chased for less time and finally followed immediately, even when conditions were changed or the trainer was replaced by another person. This suggests that horses learn to follow in this particular situation and also show some generalisation. However, following did not occur on a pasture even after several successful trials in the riding arena.

Abstract: Factors related to dominance rank and the functions of aggression were studied in a herd of Sorraia horses, Equus caballus, under extensive management. Subjects were 10 adult mares 5-18 years old and a stallion introduced into the group for breeding. Dominance relationships among mares were clear, irrespective of rank difference, and remained stable after introduction of the stallion. The dominance hierarchy was significantly linear and rank was positively correlated with age and total aggressiveness. Higher-ranking mares received lower frequency and intensity of agonistic interactions. Nevertheless, higher-ranking dominants were not more likely to elicit submission from their subordinates than lower-ranking dominants. Neither close-ranking mares nor mares with less clear dominance relationships were more aggressive towards each other. Agonistic interactions seemed to be used more importantly in regulation of space than to obtain access to food or to reassert dominance relationships. Contexts of aggression were related to mare rank. The results suggest that dominance relationships based on age as a conventional criterion were established to reduce aggressiveness in a herd where the costs of aggression are likely to outweigh the benefits.

Abstract: Living in large, stable social groups is often considered to favour the evolution of enhanced cognitive abilities, such as recognizing group members, tracking their social status and inferring relationships among them. An individual's place in the social order can be learned through direct interactions with others, but conflicts can be time-consuming and even injurious. Because the number of possible pairwise interactions increases rapidly with group size, members of large social groups will benefit if they can make judgments about relationships on the basis of indirect evidence. Transitive reasoning should therefore be particularly important for social individuals, allowing assessment of relationships from observations of interactions among others. Although a variety of studies have suggested that transitive inference may be used in social settings, the phenomenon has not been demonstrated under controlled conditions in animals. Here we show that highly social pinyon jays (Gymnorhinus cyanocephalus) draw sophisticated inferences about their own dominance status relative to that of strangers that they have observed interacting with known individuals. These results directly demonstrate that animals use transitive inference in social settings and imply that such cognitive capabilities are widespread among social species.

Abstract: This study examines the role of observation during the formation of triads in female domestic hens. Results indicate that during hierarchy formation, a hen observing agonistic interactions and conflict settlement between its former dominant and a stranger uses this information when in turn confronted by the latter. Under a first condition (E, N = 15 triads), bystanders witnessed their prior dominant being defeated by a stranger before being introduced to them. In a second condition (C1, N = 16 triads), bystanders witnessed the victory of their prior dominant over a stranger. In a third condition (C2, N = 15 triads), bystanders witnessed two strangers establishing a dominance relationship before being introduced to their prior dominant and to a stranger the former had just defeated. The behavioural strategies of bystanders depended on the issue of the conflict they had witnessed. Bystanders of the E condition behaved as having no chance of defeating the stranger. They never initiated an attack against it, and upon being attacked, readily submitted in turn to the stranger. On the contrary, bystanders of the C1 condition behaved as having some chances against the stranger. They initiated attacks in 50% of cases, and won 50% of conflicts against the stranger. Under condition C2, bystanders first initiated contact with the strangers in only 27% of cases, which approximates the average of their chances for defeating the stranger. However, bystanders finally defeated the strangers in 40% of cases. These results suggest that bystanders of conditions E and C1 gained some information on the relationship existing between their prior dominant and the stranger and that they used it coherently, perhaps through transitive inference, thus contributing to the existence of transitive relationships within the triads. Alternate explanations are examined.

Abstract: In many different species it is common for animals to spend large portions of their lives in groups. Such groups need to divide available resources amongst the individuals they contain and this is often achieved by means of a dominance hierarchy. Sometimes hierarchies are stable over a long period of time and new individuals slot into pre-determined positions, but there are many situations where this is not so and a hierarchy is formed out of a group of individuals meeting for the first time. There are several different models both of the formation of such dominance hierarchies and of already existing hierarchies. These models often treat the two phases as entirely separate, whereas in reality, if there is a genuine formation phase to the hierarchy, behaviour in this phase will be governed by the rewards available, which in turn depends upon how the hierarchy operates once it has been formed. This paper describes a method of unifying models of these two distinct phases, assuming that the hierarchy formed is stable. In particular a framework is introduced which allows a variety of different models of each of the two parts to be used in conjunction with each other, thus enabling a wide range of situations to be modelled. Some examples are given to show how this works in practice.