An empirical investigation of two assumptions of motivation testing in captive starlings ( Sturnus vulgaris): Do animals have an energy budget to ‘spend’? and does cost reduce demand?

Abstract

The use of demand curves to derive estimates of motivational strength is a popular method for measuring animals’ preferences for a range of different resources in applied animal behaviour research. In a typical experiment, an animal pays a gradually increasing cost (e.g. by pushing through a weighted door) in order to access a resource it wants or needs. The resulting demand curves are used to calculate several measures of the strength of the animal's motivation to access the resource. We tested two assumptions that underlie the majority of applications of this approach: first, that animals have a fixed energy budget to spend on access to resources; and second, that the effect of price on demand is not greatly influenced by the order or magnitude of the price changes. Sixteen European starlings ( Sturnus vulgaris) were trained to push through weighted doors to gain access to one of two resources, either a tray of turf, or protective foliage cover. In the first stage of the experiment reservation prices were established for each bird by daily increasing the force necessary to open the door until the bird no longer accessed the resource. In the next stage, five different forces, chosen to evenly cover the range between free entry and each individual bird's reservation price, were presented in a random order under two levels of food availability. Overall, price was the most important determinant of demand. However, birds’ demand for resources was increased by food rationing, suggesting that the cost of pushing the weighted doors might not have been energy. Birds’ willingness to pay for a resource was also dependent upon the order in which the forces were presented, and specifically the contrast from the force presented the previous day. The results support the continued use of presenting costs on an ascending order to measure the demand in captive animals, but suggest that random order presentations can be used to check for order effects. We also suggest that the concept of a finite energy budget that animals have to spend on resources may not be useful in the measurement of captive animals’ preferences and that a different approach might be needed.