ablation studies: by collecting converging evidence with a variety of pharmacologic, lesion, and genetic manipulations. Because mammalian genome mapping currently focuses on mice (Mus musculus), standardized behavioral testing of mice should be adopted (Brown et al., 2000; Crawley, 2000).

Altered behaviors of knockout mice are often sufficiently obvious or unusual that they catch the attention of animal-care personnel, who then notify the investigators. Dramatic behaviors that include increased aggression, altered maternal care, decreased sexual behaviors, seizures, and impaired motor coordination and sensory abilities are commonly reported for knockout mice (e.g., Barlow et al., 1996; Brown et al., 1996; Brown et al., 2000; Chen et al., 1994; Crawley, 2000; Nelson et al., 1995; Saudou et al., 1994). Presumably, knockout mice may have more subtle behavioral changes that have not yet been discovered, even among mutants with no obvious behavioral phenotypes. Some of the behaviors probably will be revealed only if the animals are housed in conditions that are ecologically relevant with respect to space and social organization (Cabib et al., 2000; Pfaff, 2001; Potts et al., 1991).

Behavioral performance is compared among wild-type (+/+), heterozygous (+/–), and homozygous (–/–) mice in which the gene product is produced normally, produced at reduced levels, or missing, respectively. The comparison of +/+ and –/– littermates of an F₂ recombinant generation is probably the minimal acceptable control in determining the behavioral effects of knocking out a gene or genes (Morris et al., 1996).

In the past, many knockout strains were generated by using stem cells from one strain and blastocysts from another strain (see “Knockout and Knockin Mutants” in Chapter 3 for review). Therefore, behavioral differences shown by knockout mice may reflect strain effects rather than the effects of the absence of the missing gene (Broadbent et al., 2002; Gerlai, 1996; Threadgill et al., 1995). Given the potentially important effect of background genotype on ability to detect effects of targeted mutations (Crabbe et al., 1999a; Lariviere et al., 2001), behavioral neuroscientists should attend to the genetic background of the transgenic animals under study to ascertain that proper controls for strain differences are used. Another limitation of the interpretation of behavioral data from knockout mice is the possibility that compensatory or redundant mechanisms might be activated when a gene is missing. For example, mice lacking the gene for the neuronal isoform of nitric oxide synthase (nNOS–/–) have a 20% increase in the expression of the endothelial isoform of nitric oxide synthase (Burnett et al., 1997). A compensatory mechanism may spare behavioral function and cloud interpretation of the normal contribution of the gene to behavior. Knockout mice are almost always raised by their natural mothers, which are missing one or more genes that may directly or indirectly affect maternal behavior. Thus, any changes in behavior observed in the knockout offspring may reflect the absence of the missing gene or reflect alterations in maternal care. Cross-fostering of matched-size litters can be used to untangle these influences.