“…Second, to reduce the variability among opponent mice used for aggression testing, we used males of a relatively nonaggressive inbred strain, the 129T2/SvEmsJ strain, as standard opponents (Roubertoux and Carlier, 1987;Maxson, 1992) in aggression tests, and we measured aggressive behavior using a modification of the dangler paradigm (Kessler et al, 1977), in which the opponent mouse is restricted to one part of the cage and thus restricted in its ability to provoke an attack. Aggression was analyzed as a threshold trait (Lynch and Walsh, 1998), i.e., a binary variable that was defined by the presence or absence of reproducible attack behavior; we chose not to measure aggression as a score based on a composite of various behaviors, because other behaviors, such as tail rattling, are only weakly correlated with attack behavior and may have a different underlying genetics (Roubertoux et al, 1999). Only unequivocal bites and/or lunges of at least moderate intensity were considered to be attacks, whereas small nips that occurred during grooming or lower-intensity pushes were not considered to be attacks.…”