Traditionally, to test whether the hypothesised organisation of behaviour generates the behavioural markers selected for measurement has required experiments or comparisons across species, sex and age. In the last couple of decades, important strides have been made in developing ways to create virtual animals, either on a computer screen or as freely moving robots, that can be programmed to produce the behaviour of interest. If the programmed rules are sufficient to produce the behaviour of real animals, then that adds independent evidence for the proposed organisation. Novel testing methods is one direction for the future. Another is to identify additional organisational principles. For example, some level of randomness seems essential for the production of effective functional behaviour. A challenge for the future is to understand how random processes are integrated with the causal processes described in the preceding chapters.

When engaging in some behaviour, some actions by an animal are more likely to resonate with us as observers than others and those impressions often form the basis for the behavioural markers that we choose for measurement. However, as much as possible, we should view the context from the animal’s perspective as it is what is important to them that guides their behaviour. Of what the animal may be able to sense, some sensations rank as perceptions that are relevant to the animal in that context. Moreover, in dynamic situations, it is often those perceptions that the animal seeks to stabilise. This means the behaviour controls those perceptions, so many actions can be explained as being compensatory. Without knowing what an action is compensating for, actions may be mistakenly abstracted as markers to measure. So, the first principle is to identify the perceptions that are relevant to the animal.

Play fighting in rats is used to show how the four principles can be used to characterise the organisation of the behaviour and then select behavioural markers that can be scored numerically. The partners compete to access the nape of each other’s neck and the behaviour patterns used during these encounters are derived from adult sexual encounters. Body size and agility can affect which tactics are used as can the location in the enclosure in which an encounter takes place. Taking these factors into account reveals that some actions cannot be explained as being compensatory to either gaining or avoiding nape contact. This, in turn, reveals novel aspects of organisation of play fighting and leads to identifying novel behavioural markers to measure those aspects of organisation.

If you have made it to here, we hope that you have enjoyed the journey. The test of the value of the journey is whether, as you gained some new insights, you looked at your pet cat or dog or at animals at a zoo in a different way. Most importantly, if you are contemplating a scientific study, hopefully you have acquired some new insights into the process by which to decide what may be the most profitable aspects of the behaviour to measure. What we have offered is a glimpse into the factors that contribute to how behaviour is organised and have, hopefully, shown how abstracting behavioural markers can be rendered into a more objective process. Even if derived from a formal application of the principles that underlie the organisation of behaviour, once multiple researchers apply them, some behavioural markers will weather the test of time but others will not. In the latter case, the proposed markers will be found to be poor reflections of the organisation they are meant to represent. That is what science is all about; some hypotheses stand up to scrutiny and some do not.

Some actions are intrinsic motor units that can be concatenated as needed to solve a problem. In contrast, many actions are correlated with one another when engaging in particular types of behaviour, such as predation, mating or aggression. Moreover, how those actions are associated depend on intrinsic rules of organisation. Consequently, selecting markers to measure as if they are independent may be misleading. For example, scoring success in grasping a piece of food may fail to reveal that different combinations of limb movements may be capable of comparable rates of success. The lack of independence among actions and the potentially misleading conclusions that can be drawn from end-point measures point us to the second principle – understanding the intrinsic organisation of behaviour. Knowing something about the intrinsic organisation of a behavioural sequence can be critical in identifying markers that reflect that organisation.

Traditionally, to test whether the hypothesised organisation of behaviour generates the behavioural markers selected for measurement has required experiments or comparisons across species, sex and age. In the last couple of decades, important strides have been made in developing ways to create virtual animals, either on a computer screen or as freely moving robots, that can be programmed to produce the behaviour of interest. If the programmed rules are sufficient to produce the behaviour of real animals, then that adds independent evidence for the proposed organisation. Novel testing methods is one direction for the future. Another is to identify additional organisational principles. For example, some level of randomness seems essential for the production of effective functional behaviour. A challenge for the future is to understand how random processes are integrated with the causal processes described in the preceding chapters.