crude physical competition, or attain high social status by attacking other individuals. The serious dangers of simplistic extrapolations from animals to humans have been well explored.

Another complicating feature of dealing with animal aggression is the striking diversity of tests said to measure this attribute in particular species (see Brain, 1981, 1989b). In, for example, the laboratory mouse, aggression is said to be generated by pairing preisolated males (intermale aggression), by exposure of paired males or females to unavoidable foot or tail shock (shock-elicited aggression), by arranging for an unfamiliar intruder to enter the nest area of a lactating female with her offspring (maternal aggression), by placing a lactating female (or an animal marked with her urine) into an established group of females or castrated males, by giving the subject the opportunity to kill a locust or a cricket (predatory aggression), and by confining subjects in a narrow tube where they may bite a target suspended in front of them thus activating a telegraph key (instrumental aggression). Thus even in the "simple" mouse, the tests used to generate aggression are so varied (and the responses generated so qualitatively different), it is highly improbable that all measure the same motivation. Certainly, housing conditions (Brain and Benton, 1983), genes (Jones and Brain, 1987), hormones (Brain et al., 1983), and drugs do not have consistent influences across these different tests. It has been argued (Brain, 1984a) to be highly probable that these diverse harm-directed activities variously tap offensive, defensive, or even predatory motivations. In some cases, mixtures of motivations appear to be involved. Support for this view is provided by the use of video analysis, which reveals that, in some "ritualized" responses, vulnerable areas (i.e., the head and ventral surface) of the opponent's body are rarely bitten (in so-called offensive intermale aggression); in other tests, vulnerable areas are frequently bitten (e.g., "defensive" maternal attack on a potentially cannibalistic male intruder), and a third category involves directed killing strategies (e.g., predatory aggression). Perhaps one should limit the term aggression to offensive displays, and thus clearly separate these utilities of attack and threat from defensive and predatory functions? Having said this, one can make a strong case for the detailed investigation of offense, defense and predation in animals being of great relevance to understanding the possible roles

of biological factors in human behavior. One should note that the terms "offensive" and "defensive" are essentially based on functional explanations of particular events; it is not easy to operationalize them. For example, the action of biting can be used in rodents to carry offspring, to eat, to kill prey, to defend a nest site, or to attack a conspecific. Brain (1984a) has suggested that one can define a number of categories of behavior that all employ fighting and/or threat. These include the following:

• Social conflict: generally intraspecific phenomena involving competition for a substrate (e.g., a mate, territory, social status, or food), the possession of which increases the organism's relative fitness. These are generally ritualized responses in which the potential for serious damage is limited.

• Parental defense: behaviors in both inter- and intraspecific contexts that serve to protect the attacker's young or nest sites from potentially destructive intruders.

• Self-defense: behaviors in both inter- and intraspecific contexts that normally serve to protect the organism per se from potential predators or attacking conspecifics. Such behaviors are generally limited to situations where flight is difficult or precluded, and do not involve ritualization.

• Infanticide: an intraspecific phenomenon involving the killing of young. In males, this may be a method of increasing the individual's reproductive fitness, whereas in females, it is commonly a response to stress or disturbance (recycling of resources?).

• Predation: an inter- or intraspecific response that involves efficient killing and is often followed by feeding activity.

One should note that it is extremely rare (in animal studies) to find purely offensively or purely defensively motivated behavior. The "ethoexperimental" approach to the analysis of animal behavior (Blanchard et al., 1989) seems to offer advantages in studying animal conflict. It basically attempts to fuse the positive features of ethology and experimental psychology by

1. creating laboratory environments that reflect the natural requirements of the feral ancestors of laboratory animals. For example, when dealing with a socially living primate species from an arboreal habitat, it seems appropriate to study mixed-sex groups in complex environments, offering a range of tactile experience. When dealing with animals such as the laboratory rat, it seems appropriate to offer the species the opportunity to construct burrow and nest systems (or to provide an equivalent) and to investigate