Although we welcome new thought aimed at stimulating research on animal innovation, we find problems with the definitional approach Ramsey et al. have taken, especially with defining innovation at the level of the individual. We are unconvinced by the restriction on cases involving what the authors term “environmental induction” (proposing instead a distinction between “passive” and “active” innovation), and on cases with significant positive fitness impacts. Finally, we don't believe the distinction between behaviour and repertoire flexibility is useful.

First, we do not believe that defining an innovation as something new to the individual, rather than the population, makes sense. No doubt it is individuals that innovate, and it is possible that more than one individual in a population might produce the same innovation independently. However, if innovation is simply another term for asocial (or individual) learning, we have unnecessary redundancy. The concept of innovation at the population level is useful, because it pertains to the process by which new behaviours enter into populations, with potential evolutionary impacts. At the individual level, the concept becomes entangled with processes such as trial-and-error learning that we might describe under the broad term “asocial learning.” Furthermore, some forms of learning require both social and asocial processes. For example, individuals may be attracted to a new food source through stimulus enhancement, but then use trial-and-error learning to acquire the extraction method. Therefore, perversely, “innovating” as defined by Ramsey et al. would be subsumed within a form of social learning. The commitment to innovation at the individual level is also at odds with our societal sense of what an innovation is – we have a patent system precisely to establish the primacy of whoever first came up with something new. Can all individual learning be innovation? We think not.

One possible response to this question is to restrict innovation to those cases not resulting from environmental induction. This separation is unconvincing, not least because most examples of innovation are linked to novel opportunities in the environment. The adaptive value of innovation, indeed behavioural flexibility in general, is surely in dealing with environmental variation. Thus, periods of environmental change are precisely when one might expect higher levels of innovation. Perhaps a landslide caused a pile of stones to appear next to a tree suddenly – does this mean their subsequent use cannot, by definition, be an innovation? Milk-bottle opening by blue tits in the United Kingdom is widely touted as an innovation, but without the environmental induction of soft-topped milk bottles, would never have occurred.

Nonetheless, understanding the role of environmental change in innovation is potentially worthwhile. However, we would draw a rather different distinction to that made by Ramsey et al. Their hypothetical example of chimpanzees switching to stone tools after a forest fire illustrates the point. The switch to stone tools could only occur through a phase of operant learning by the chimpanzees: the use of stones can only be reinforced once an individual has experimented, and succeeded in using one to crack open a nut. Thus, an environmental change may act to provide either the opportunity or necessity for operant learning, which, in the case of the first individual to succeed, we would term “innovation.” We feel such cases are fundamentally different to cases in which an environmental event directly alters the behaviour of an individual. An individual may learn through being passively exposed to the relationship between two stimuli in its environment. For instance, after being attacked, an individual may learn to associate a particular scent with the arrival of a predator. In future, it may respond to that scent with fear or escape responses. If the individual is the first in the population to form this association, it might be said to be the innovator. We would term that individual a “passive innovator.” In contrast, an individual who has learned to use a stone hammer, through operant learning, we would term an “active innovator.” We feel this distinction is likely to prove useful as active innovation is more likely to reflect the cognitive abilities of the innovator than passive innovation, which is more likely to rely on chance events.

These issues aside, the target article gives the impression that it would be easy to distinguish non-innovation based on environmental induction, but this is far from the case. For example, the claim that environmentally induced novel behaviours are expected to be adopted rapidly by most of the population following a change in the environment does not hold water if the rapid spread was a result of (1) asocial learning, if, for example, the task is easy to solve (i.e., “innovation” sensu Ramsey et al.) or (2) horizontal social transmission of the innovation, which can also be rapid.

The assertion that behaviours with large positive fitness impacts should not be considered innovations because they are likely to have been “Baldwinized” to become at least partly innate, is also unconvincing. We think a stance that discards putative cases of innovation a priori because of the possibility of this having occurred is overly restrictive. The relative contribution of genetic and learned sources of information in the development of behaviour is an empirical question, and there are several lines of possible evidence available to researchers to address it. When this is taken alongside the “environmental induction” restriction, we arrive at a slightly bizarre situation in which we are asked to not consider putative cases of innovation that result in large fitness payoffs in the face of environmental change, when it is precisely these cases that are likely to favour innovation.

Finally, we were confused by the distinction made between behaviour and repertoire flexibility. For example, the authors state that stone hammering is a repertoire addition, but we could argue that actually it is just flexibility in a general “hammering” program: hammer with whatever seems to be a good tool at the time. Similarly, it is unclear why elements of behavioural flexibility cannot be learned, for example, if stone use were subsequently reinforced and resulted in operant learning, and therefore innovations? Deciding whether a particular case pertains to the behavioural or repertoire level will not be an easy task; and in the absence of evidence that the brains of a particular species actually function in this way, it seems to us rather a futile one.

In summary then, we do not believe the case for redefining innovation to the individual level is made, nor is it made for the restrictions on environmental induction or high fitness impacts this redefinition necessitates.