Mild environmental aversion is detected by a discrete-choice preference testing method but not by a free-access method

Preference tests are a widely used method in animal welfare science but the influence of specific methodologies has not been widely investigated, and methods are often selected arbitrarily. In these experiments we assessed the environmental preferences of 72 individual laying hens, where environments contained either a foraging box (F), an element of mild risk presented by an insect-shaped robot (R) or both (FR). Preferences were assessed for paired sets of environments (F vs FR; FR vs R; F vs R) in a balanced sequential design. Two different methods were used, a discrete-choice procedure using a T-maze where choices were followed by a short confinement period in the chosen environment and a free-access procedure where hens were free to move through a tunnel between the two environments of a pair. The discrete-choice method revealed an overall preference ranking F > FR > R, with a high level of transitivity expressed between sets. The free-access method did not detect the mild aversion to R and revealed an overall preference ranking F = FR > R, with a much weaker level of transitivity expressed between sets. The additional control provided by the free-access method, meaning that hens could leave the R environment whenever they wished, may have significantly reduced the aversiveness of the R stimulus. Such influences need to be considered when selecting a preference testing method in an animal welfare context.