Reward expectation drives dogs' choices in a prosocial test.

Recent research has tried to determine whether and under which conditions domestic dogs (Canis familiaris) engage in prosocial and potentially altruistic behaviour. A common paradigm used to test this is the so-called prosocial choice test. This is a task in which the animal can choose to operate an apparatus based on the outcome. For example, this could involve delivering food to themselves, to another individual, to both, or to neither. The aim of the current study was to investigate dogs' collaborative tendencies with humans, with the use of a novel prosocial choice test paradigm. This paradigm was designed to take into account the fact that predators may have a more natural tendency to pull vertically rather than horizontally. Our paradigm is based on pulleys that are operated by the dogs, an action which dogs mastered without extensive training. All dogs (N = 16) participated in three experimental conditions (selfish-, prosocial- and altruistic condition) and two control conditions (no-reward- and partner-absent control). We further tested to what extent the between subjects' familiarity with the human partner (owner vs. stranger) mattered to the dogs. The dogs were more likely to operate the apparatus when they received a reward (selfish- and prosocial condition) than when they did not (altruistic- and no-reward condition). In addition, the absence of a partner, owner or stranger only marginally affected the dogs' tendency to pull. Based on our findings, we cannot conclude that dogs make altruistic or prosocial choices; rather, they appear motivated by selfish interests. We discuss the possibility that even small procedural changes can have large implications on the behaviour of the dogs during the tests.