What are we measuring when we test strain differences in anxiety in mice?

We examined measures of locomotor and anxiety-like behavior in male and female mice of 15 inbred strains on the elevated-plus maze, light/dark transition box and open field. Strain differences were found on all measures of locomotor activity and anxiety. Strain means for measures of locomotor activity on the three apparatus were significantly correlated, but strain means for commonly used measures of anxiety were not correlated. Principal component analysis revealed a common locomotor activity factor, which accounted for 28.6 % of the variance, but no common anxiety factor. Species-typical behaviors (defecations, stretch-attend postures, grooming) accounted for smaller proportions (<11 %) of the variance. These results plus comparisons with previously published data suggest that the elevated-plus maze, light/dark box and open field measure different facets of anxiety, and that the reliability of genetic differences on anxiety is highly dependent on apparatus, procedural variables and laboratory factors. Locomotor activity, however, is a stable trait that differs across strains and is reliably measured in different apparatus and laboratories. We conclude that anxiety traits of inbred mouse strains are best reflected by species-typical behaviors in each apparatus. These results suggest that new ways of measuring trait anxiety are required in order to determine the neural and genetic correlates of anxiety-like behaviour in mice.