Is fearfulness a trait that can be measured with behavioural tests? A validation of four fear tests for Japanese quail

If fearfulness is stable, consistent and trait-like, then valid measures of fearfulness should be stable, consistent and independent of influences unrelated to fear. We assessed the validity of six fear measures using Japanese quail, Coturnix coturnix japonica, a common species in fear research. Measures were made during emergence, novel object, novel food and predator surprise tests. These were considered to have internal validity if they were stable over 18 days, when we controlled for nonexperimental variables including season of testing and cage location. We determined convergent and discriminant validity by factor analysis of fear measures plus measures of sociality, activity level and repetitive behaviour. Fear measures with good convergent validity showed agreement in their factor loadings. Those with good discriminant validity loaded on to different factors from nonfear measures. Most of the fear measures examined were moderately stable over time, but only half had good discriminant validity. Convergence was good among measures from the same test but poor across tests. Measures from each fear test loaded separately. Overall, flight distance and freezing duration in the predator surprise test and amount eaten in the novel food test showed the best internal, convergent and discriminant validity. When we considered only these three measures, convergence remained higher among measures from the same test than from different tests. Fearfulness thus appeared somewhat unstable over time and inconsistent across situations, which, if true across species, greatly limits the utility of fear tests. (c) 2006 The Association for the Study of Animal Behavior. Published by Elsevier Ltd. All rights reserved.