Abstract: Associative learning theories presume the existence of a general purpose learning process, the structure of which does not mirror the demands of any particular learning problem. In contrast, learning scientists working within an Evolutionary Biology tradition believe that learning processes have been shaped by ecological demands. One potential means of exploring how ecology may have modified properties of acquisition is to use associative learning theory as a framework within which to analyse a particular learning phenomenon. Recent work has used this approach to examine whether socially transmitted predator avoidance can be conceptualised as a classical conditioning process in which a novel predator stimulus acts as a conditioned stimulus (CS) and acquires control over an avoidance response after it has become associated with alarm signals of social companions, the unconditioned stimulus (US). I review here a series of studies examining the effect of CS/US presentation timing on the likelihood of acquisition. Results suggest that socially acquired predator avoidance may be less sensitive to forward relationships than traditional classical conditioning paradigms. I make the case that socially acquired predator avoidance is an exciting novel one-trial learning paradigm that could be studied along side fear conditioning. Comparisons between social and non-social learning of danger at both the behavioural and neural level may yield a better understanding of how ecology might shape properties and mechanisms of learning.

Abstract: Horses are quite unique. In most mammals, sexes segregate and maintain bonds only during the breeding season (Clutton-Brock, 1989). Some canids, a few rodents and primate species such as gorillas, hamadryas baboons and red howler monkeys are the exception, where the same males stay with the same females all year round and over many breeding seasons. Typically, both sexes disperse at puberty in these species. In horses, it was clearly shown that the causes for female dispersal were incest avoidance and not intra-specific competition (Monard, 1996). As a rule, this is confirmed for mammal species where tenure length by males exceeds the age at first reproduction in females (Clutton-Brock, 1989). When horses are allowed to choose their mating partner freely, the inbreeding coefficient of the offspring is lower than expected should they mate randomly (Duncan et al, 1984).

Abstract: (Accepted: 23 June 2005)
Summary
Much research in animal communication is aimed at understanding the functional design
features of animal vocal signals. Our detailed analyses of the vocalizations and behavioural
responses elicited in white-faced capuchins by predators and other disturbances point to two
call variants that differ modestly in their acoustic structure and that are accompanied by
functionally distinct behavioural responses. The first variant is given exclusively to avian
predators and is almost invariably accompanied by the monkeys immediate descent from
the treetops where it is most vulnerable; therefore, we label this call variant the aerial
predator alarm?. The second variant, that differs only slightly but noticeably from the first,
is given to a wide range of snakes and mammals, including a range of species that represent
no predatory threat to the monkeys. This second call is also associated with more variable
responses from calling monkeys, from delayed retreat from the source of disturbance, to
active approach, inspection, and sometimes mobbing of the animal involved. We therefore
label this variant more generally as an “alerting call”. Although some other primate species
show a more diverse system of anti-predator calls, and the capuchins themselves may yet
be found to produce a greater variety of calls, a system of two call variants with varying
degrees of predator specificity and behavioural response is not uncommon among primates
and appears functionally appropriate for capuchins. The basic structure of the alerting call
allows conspecific listeners to localize the caller and the source of disturbance readily, thereby
allowing listeners to approach and assist in mobbing in cases where the disturbance warrants
it, or to avoid the area in cases where the disturbance is identified as a predatory threat.
Conversely, the aerial predator alarm is inherently less localizable and therefore conveys the

Abstract: We performed experiments with cichlid fish to test whether several basic aspects of dominance were the same in isolated pairs as in pairs within a social group of three or four. We found that the social context, whether a pair was isolated or within a group, strongly affected the basic properties of dominance relationships. In particular, the stability of relationships over time, the replication of relationships in successive meetings, and the extent of the loser effect were all significantly less in socially embedded pairs than in isolated pairs. We found no significant winner effect in either isolated or socially embedded pairs. These findings call into question many current approaches to dominance that do not consider social context as an important factor in dominance behavior. These findings also cast serious doubt on the validity of empirical and theoretical approaches based on dyadic interactions. Among these approaches are game theoretic models for the evolution of aggressive behavior, experimental designs evaluating how asymmetries in attributes influence the outcome of dominance