Relational matching in baboons (Papio papio) with reduced grouping requirements

Analogical reasoning is a corner stone of human cognition, but the phylogenetic origins of this skill are still unknown. Recent animal studies have suggested that only apes can solve the 2- by 2-item relational matching (RMTS) analogy problem, with potential benefits of language- (Premack, 1983) or token-training procedures (Thompson, Oden, & Boysen, 1997). In this study, 6 baboons were initially trained in an RMTS task in which the same and different relations were exemplified by compound stimuli made of 2 adjacent patches of colors. Learning occurred in this task with a first set of colors and transferred to probe trials with new colors (Experiment 1). Manipulation of the size of the sample or comparison stimuli (Experiment 2) showed that the performance was not merely controlled by the surface of the color patches, suggesting cognitive flexibility. Performance collapsed to chance level when a gap was introduced between the 2 elemental features composing the same or different displays (Experiment 3). Nevertheless, this effect of gap size was abolished by training (Experiment 4). It is suggested that monkeys share the ability to judge relations between relations with humans and apes, even in the absence of language or token training. However, this ability has been previously masked by a local mode of processing that hinders the processing of the stimuli as pairs rather than as independent objects.