Kline provides us with a comprehensive and thought-provoking review of our current understanding of teaching. We particularly welcome the focus on increasing dialogue among researchers in different disciplines studying the topic. The comparative study of teaching is still in relative infancy, with few cases of teaching in nonhuman animals having been identified (Fogarty et al. Reference Fogarty, Strimling and Laland2011; Hoppitt et al. Reference Hoppitt, Brown, Kendal, Rendell, Thornton, Webster and Laland2008; Thornton & Raihani Reference Thornton and Raihani2010). Increased dialogue between fields can only assist the furtherance of research in this area. By proposing a unified framework for the study of teaching, Kline aims to encourage better formalized and more constructive dialogue.

We agree with Kline that there are different mechanisms by which teaching can occur; however, we are concerned over the implementation of the framework as it currently stands. Thornton and Raihani (Reference Thornton and Raihani2010) propose a functional approach to identifying teaching, arguing that researchers need to examine the behavior of learners as well as teachers, focus on the learning outcomes, and examine the costs for teachers. By an increased focus on the behavior of teachers during a teaching event, we fear that operationalizing Kline's framework in nonhumans may depend too much on inference by the researcher about the purpose of the behavior.

Let us take as an example the category teaching by social tolerance. Kline proposes that adults may be tolerant of other individuals in the population, particularly juveniles, specifically in order to facilitate learning in those individuals. If “teachers” are deliberately creating a learning environment for “pupils,” this clearly goes beyond inadvertent social learning and may qualify as a teaching event: “opportunity teaching” in Caro and Hauser's (Reference Caro and Hauser1992) classification or “teaching by local enhancement” in Hoppitt et al. (Reference Hoppitt, Brown, Kendal, Rendell, Thornton, Webster and Laland2008). However, the reason for social tolerance among individuals is an empirical question itself. Tolerance varies between species (de Waal & Luttrell Reference De Waal and Luttrell1989) and within species (Burkart & van Schaik Reference Burkart and van Schaik2010), and may be influenced by the culture of the group (Sapolsky Reference Sapolsky2006). There may also be many different reasons why animals are differentially tolerant towards other individuals, including kinship bonds (Sueur et al. Reference Sueur, Petit, De Marco, Jacobs, Watanabe and Thierry2011) and strategic allegiances (Byrne & Whiten Reference Byrne and Whiten1988), without active teaching occurring. This differential social tolerance may increase the likelihood of inadvertent social learning, as those animals that are more tolerant of one another may produce more opportunities for social learning to occur, a phenomenon known as “directed social learning” (Coussi-Korbel & Fragaszy Reference Coussi-Korbel and Fragaszy1995).

Care should, therefore, be taken to ensure that the behavior that is identified is due to active teaching and not inadvertent social learning, driven by the learner. Such a distinction may be relatively easy to make in human populations, where individuals may be asked why they are acting in a particular manner; however, in nonhuman species, there remains a need to rule out alternatives through empirical methods (Thornton & Raihani Reference Thornton and Raihani2010). Functional approaches to the study of teaching, with a focus on teaching costs, learning opportunities, and learning outcomes should reduce the need for inference and subjective judgment on the part of researchers. For example, without considering these things, would Kline describe the tolerance of nut-cracking tufted capuchins for the presence of scrounging juveniles (Ottoni et al. Reference Ottoni, Dogo de Resende and Izar2005) as teaching? In reality it may be of direct benefit to individuals (and thus not “costly” teaching) to tolerate scrounging if (1) there is a chance the juveniles are their own offspring, and (2) the loss of nut-meat is less than the opportunity cost of fending off juveniles. An alternative reason why this would not qualify as teaching may be seen from the juvenile (or potential pupil's) point of view: there is conflicting evidence regarding the utility of scrounging in learning outcomes (Caldwell & Whiten Reference Caldwell and Whiten2003; Giraldeau & Lefebvre Reference Giraldeau and Lefebvre1987). Hence, even were it to be documented that adults are more tolerant of juveniles during nut-cracking than at other times, this evidence alone is not sufficient to classify it as teaching.

The framework proposed has the advantage that it has few categories, allowing initial comparison among species, and research fields, to be more straightforward. We are concerned, however, that in studies with nonhuman species, categories may not be as intuitive as Kline has proposed. To illustrate, consider an example given by Kline for teaching by stimulus or local enhancement. Otters and sea lions drag their offspring into the water to ensure that they learn to swim. It is far from clear that this corresponds to stimulus enhancement, as described in the social learning literature, in which attention is drawn to an object or part of an object (Whiten & Ham Reference Whiten and Ham1992). While pupils will have their attention drawn to the water, placing an animal in water to ensure that it swims would equally appear to fit within Kline's teaching by opportunity provisioning category. In this category, teachers create the opportunities for pupils to practice, using asocial learning, which would not otherwise exist. By placing a pupil into water, it could be equally argued that the teacher is creating an opportunity that the pupil would not otherwise encounter. We propose that this represents a broader deficiency with the framework, in which categories are not exclusive and independent. The strength of a comprehensive framework is that it can provide researchers with an objective means of categorizing these mechanisms, even if multiple mechanisms are occurring in one teaching event. The framework's utility may be reduced due to the issue of the proposed mechanisms being judged differently by different researchers.

There is much to commend in Kline's attempt to construct a framework that would allow a comparative and collaborative approach to the study of teaching. However, the merit of any interdisciplinary approach must be in how applicable it is to all fields involved. We have concerns that, particularly for comparative researchers, the framework as currently proposed may prove to be overly reliant on subjective judgment and inference, reducing the comparative and collaborative utility of the exercise.