versus mesencephalic and striatal regions is differentially mediating aggressive versus defensive behavior patterns. Such specificity with regard to brain region and behavior pattern should prompt the development of more selective diagnostic behavioral assessment and pharmacotherapeutic intervention; it also casts doubt on single indices of 5-HT that summarily attempt to represent 5-HT activity in the entire brain.

Impairments of brain 5-HT systems by removing the dietary precursor l-tryptophan, blocking the synthetic enzyme tryptophan hydroxylase, depleting the 5-HT vesicular storage, or cytotoxically or electrolytically destroying serotonin neurons lead mostly to suppression of attack and threat behavior by isolated mice (see Table 3, section B; e.g., Poschlová et al., 1975; Rolinski, 1975; Eichelman, 1981; Payne et al., 1984; Svare and Mann, 1983; Ieni and Thurmond, 1985). However, pharmacologic manipulations with opposite biochemical effects such as tryptophan loads, administration of the precursor 5-HTP (5-hydroxytryptophan) or releasing agents, blockade of enzymatic inactivation of 5-HT with MAO (monoamine oxidase) inhibitors, or uptake inhibition, either acutely or chronically also decreased attack and threat behavior (Table 3, section B; Thurmond et al., 1979; Eichelman, 1981; Chamberlain et al., 1987). Most of these manipulations lack behavioral specificity in that sedation and motor incapacitation accompany the antiaggressive effects.

Defensive-aggressive responses in rats reacting to painful electrical shock pulses may be facilitated through impairing 5-HT by omitting l-tryptophan from the diet, inhibiting 5-HT synthesis, or cytotoxically or electrolytically destroying 5-HT-containing neurons, at least under some experimental conditions (see Table 3, section B; e.g., Ellison and Bresler, 1974; Benkert et al., 1973; Eichelman, 1981; Rolinski and Herbut, 1981; Pucilowski and Valzelli, 1986). A reliable facilitation of defensive reactions, but not attack behavior, is seen after chronic inhibition of 5-HT reuptake or MAO with antidepressants (see below; e.g., Delini-Stula and Vassout, 1981; Mogilnicka and Przewlocka, 1981; Prasad and Sheard, 1983a,b). In cats, but not in monkeys, inhibition of 5-HT synthesis amplified affective defense (e.g., MacDonnell et al., 1971; Redmond et al., 1971a).

By far, the strongest evidence for an inhibitory role of 5-HT in animal aggression has been accrued by neuropharmacologic studies

of 5-HT and killing behavior by laboratory rats, usually directed toward a mouse—''muricide" (see review by Miczek and Donat, 1989; also Table 3, section B). Because of its similarity to killing in the sequence of predatory stalking and hunting by carnivores, muricide by omnivorous rats has been termed "predatory aggression." From an ethological view, it is conceptually inconsistent to combine predation and aggression, since the causative and functional dimensions of these behaviors differ fundamentally. More than 70 studies during the past 25 years demonstrate that laboratory rats that have not killed a mouse previously are more likely to do so after electrolytic or neurotoxic insults to the serotonin-containing raphe nuclei, omission of l-tryptophan from the diet, or inhibition of 5-HT synthesis (see Table 3, section B; DiChiara et al., 1971; Eichelman and Thoa, 1973; Isel and Mandel, 1989; Banerjee, 1974; Breese and Cooper, 1975; Vergnes et al., 1973, 1988; Gibbons et al., 1978). Conversely, tryptophan loads in the diet, precursor administration, and inhibition of enzymatic inactivation with MAO inhibitors or reuptake blockers effectively suppress the killing response (Table 3, section B; e.g., Kulkarni, 1970; Bocknik and Kulkarni, 1974; Gibbons et al., 1978, 1981).

However, the evidence for a close link between 5-HT and killing needs to be qualified: (1) Many severely 5-HT-depleted rats fail to show the killing response, and others without any detectable abnormality in or even increased levels of 5-HT or 5-HIAA engage regularly in this behavior (e.g., Salama and Goldberg, 1973b; Miczek et al., 1975; Broderick et al., 1985). (2) It has not been possible to specify a threshold value of 5-HT impairment or, alternatively, to relate 5-HT depletion to the probability of killing in a systematically graded dose-effect manner. (3) Rats that have been previously exposed to the potential prey, will not develop the killing response after insults to 5-HT activity (e.g., Marks et al., 1977; Vergnes et al., 1977; Vergnes and Kempf, 1981). (4) Once the killing behavior has become part of the animal's repertoire, it persists in the absence of any changes in 5-HT levels, synthesis, or metabolism (e.g., Vergnes and Kempf, 1981). (5) Some carnivores such as cats, ferrets, or grasshopper mice are actually impaired in their killing behavior when 5-HT synthesis is blocked, and the killing response cannot be blocked by antidepressants that are 5-HT reuptake inhibitors (McCarty et al., 1976; Leaf et al., 1978; Schmidt and Meierl, 1980; Schmidt, 1980).

The neurobiological mechanisms of killing behavior include an important role of 5-HT, particularly in rats that do not exhibit this behavior normally. At the same time, additional mechanisms

override or modulate 5-HT activity in the mediation of killing behavior, especially when this behavior is already part of the organism's repertoire or when the organism is habituated to the provocative stimulus. The predatory killing by carnivores appears not to be critically dependent on 5-HT, which suggests a different interpretation of the mouse-killing response by the laboratory rat, possibly as a model of pathology (e.g., Karli, 1981; Valzelli, 1985).

Investigations of the role of postsynaptic 5-HT receptors in aggressive behavior, particularly killing, in animals have been relatively secondary to the presynaptic events until the important discoveries in the last decade of differentiated receptor subtypes for 5-HT (Peroutka, 1988). The first-generation agonists and antagonists with nonselective affinity for all 5-HT receptors suppressed every type of aggressive behavior in a more or less specific manner (e.g., Table 3, section B; e.g., Malick and Barnett, 1976; Weinstock and Weiss, 1980; Sheard, 1981; Miczek and DeBold, 1983; Svare and Mann, 1983; Ieni and Thurmond, 1985; Winslow and Miczek, 1983; Lindgen and Kantak, 1987).

A new phase of investigating the role of 5-HT receptor sub-types in different aggressive behavior patterns began with the increasing availability of agonists and antagonists that are more selective for each receptor subtype. Agonists at 5-HT_1A receptors, such as 8-OH-DPAT (8-hydroxy-2-(d1-n-propylamino)tetralin), buspirone, and ipsapirone, reduce attack and threat behavior by a male rat confronting an intruder or a female rat defending her litter, but also produce some sedation (Oliver et al., 1989). Newer agents such as eltoprazine, a mixed 5-HT_1A/B agonist, or TFMPP (trifluorometaphenylpiperazine), a more specific 5-HT_1B agonist, appear to have a more behaviorally specific antiaggressive activity in resident male mice and rats, and in lactating female and brain-stimulated male rats (Kruk et al., 1987; Olivier et al., 1985, 1989, 1990; Miczek et al., 1989). At present, no data exist for the effects of 5-HT_1C and 5-HT_1D receptor agonists on aggressive behavior. The 5-HT₂ receptor antagonist ketanserin reduces attack and threat behavior by isolated resident mice and rats effectively, although with limited behavioral specificity (Haney and Miczek, 1989; Olivier et al., 1989). Initial data on the effects of experimental 5-HT₃ receptor antagonists show no specific effects on aggressive behavior in isolated mice or lactating female rats (Mos et al., 1990).

An evolutionary approach to the question of serotonin and aggression seeks to determine a functional constancy or divergence

across different phyla and order. The activity of serotonin-containing neurons has been studied in invertebrates, fish, birds, and a range of mammals, including nonhuman primates before, during, or after performance of aggressive behavior (e.g., Miczek and Donat, 1989). For example, injection of 5-HT into lobsters triggers an "aggressive" looking stance, although cytotoxic destruction of 5-HT neurons is without behavioral effect (Livingstone et al., 1980). Similarly, administration of 5-HTP increased the proportion of ants (Formica rufa) fighting with each other (Kostowski and Tarchalska, 1972); however, intraventricular injections of 5-HT or 5-HT reuptake blockers into South American electric fish (Gymnotidae) decrease their aggressive signaling (Maler and Ellis, 1987). Intense defensive reactions by the prosimian tree-shrews (Tupaia belangeri) are accompanied by strong elevation of the firing of 5-HT-containing raphe cells (Walletschek and Raab, 1982). These observations, together with the earlier data from mice, hamsters, and rats, do not provide evidence for a uniform functional role of 5-HT activity in the postulated inhibitory mechanisms of aggressive behavior that may be generalized across animal species.

5-HT and nonhuman primate aggression have been studied in vervet monkeys (Cercopithecus aethiops sabaeus), talapoin monkeys (Miopithecus talapoin), rhesus macaques (Macaca mulatta), and squirrel monkeys (Saimiri sciureus) (Table 3, section B). A series of studies in vervet monkeys found consistently elevated levels of 5-HT in whole blood or in blood platelets of dominant group members as defined by success in aggressive interactions (Raleigh et al., 1980, 1983). However, at present, it is unclear how measurements of whole blood 5-HT relate to the complexities of the various 5-HT cell bodies, neuronal pathways, and receptors in brain. Preliminary data from vervet monkeys also suggested higher 5-HIAA in CSF of dominant group members than in subordinates (Raleigh et al., 1983). Inconsistent and unreliable correlations between CSF 5-HIAA and aggressive behavior were found in several extensive series of studies in talapoin, rhesus macaque, and squirrel monkeys (Yodyingyuad et al., 1985; Kraemer, 1985; Green et al., personal communication). A particularly instructive example involves studies of talapoin monkeys in which the day-to-day variation in number of attacks or number of threats did not correlate with CSF 5-HIAA and resulted in highly variable scattergrams (Yodyingyuad et al., 1985). When squirrel monkeys of higher or lower social rank are examined, only MHPG, but not 5-HIAA, was elevated in subordinate males, and active conflict led to further elevations of MHPG (Green et al., personal communication,