Structured interactions and ecological validity in the study of social cognition in captive great apes

In the current literature on the sociocognitive abilities of the great apes, there are profound differences in the findings reported by researchers using training protocols, on the one hand, and structured interactions, on the other. Structured interactions reveal more sensitivity to the behavioral correlates of intentions and attention than do training protocols. Training protocols go something like this: An operant is trained to a criterion, then that operant is used as a dependent variable during critical transfer to testing conditions in which the intentions, attention, or epistemic status of another social agent is manipulated. Early high performance during the transfer is then taken to demonstrate the discrimination of the mental state in question. Positive findings illustrate effective experiential pathways toward the discrimination of interest, whereas negative findings remain ambiguous. Structured interactions ask, instead, What discriminative competencies do test subjects bring to the experimental situation? The data yielded from structured interactions cannot specify a learning mechanism for the acquisition of any discriminative competencies demonstrated. Training protocols have the potential to specify effective learning pathways, in principle, but in practice are usually far too short in duration to determine whether the training regimen has facilitated or inhibited subsequence performance on transfer tests. Apes in captivity are inescapably embedded in an ecological web and have developed strategies for increasing foraging success in circumstances characterized by an almost complete lack of control over their own feeding schedules. Ecological validity in the context of animals who spend their entire lives in laboratories means administering experimental protocols that are consistent with their pre-experimental life experiences, otherwise they will produce evidence only of a mismatch between the contingencies as conceived by the experimenters and the contingencies as conceived by their test subjects.