A propagation of behaviours between individuals is known as “social contagion”. This is just a fancy term that includes a range of things from information spread between users of the media to flocking behaviour in birds. However, while measuring the rate and direction of information spread is reasonably straightforward, how various forms of contagious behaviour are initiated and what social interactions act as clues for the spread of particular behaviour in nature is not well understood. Over the years many different hypotheses have been proposed to explain the workings of social contagion in different animals; from suggestion of communication via telepathy in flocking birds to, a more reasonable sounding, changes in water pressure caused by sudden movements in schooling fish. One strategy of studying social contagion in animals is to create models of networks between individuals in the group, where each individual is connected to another one by one or more strings of information that can act as social clues to help to decide what behaviour to adopt.
In the case of schooling fish effective social contagion can be the difference between life and death. Schooling is used as a strategy not only to spot a predator earlier (with so many members in the group, at least one is bound to spot the incoming attack) but also to confuse the predator (commonly schooling fish have shiny outside features that, from the point of view of an attacker, make it difficult to concentrate on a particular target when hundreds of blinking and twisting spots are around it). Golden shiners are a good model to study schooling in fish because they exhibit what is known as ‘fast-start’ response. Fast-start is a rapid escape behaviour in response to various stimuli (such as visual or acoustic signals that can act as a proxy for incoming predator) but in golden shiners it can sometimes also be observed to occur spontaneously in random individuals. Interestingly, the fast-start response is mediated by a reflex and indeed, it can be shown that the shiners cannot distinguish if the escape behaviour in any given individual was caused by a real threat or a spontaneous startle. The latter point perhaps suggests that the time costs required to distinguish between real and spontaneous startles in individuals in real life situations are too high and therefore, the response has evolved as a binary system (if there’s any signal you run, if there’s nothing you don’t).
The spontaneous induction of fast-start response in golden shiners allows to model the transmission of response through the remaining group of fishes without the need for targeted induction and allows to clearly distinguish the inducers and responders in the group. Furthermore, camera-based tracking systems can be used to construct networks between the members of the group, which can then be used to predict the decisions of individual fishes in different situations. Using such fancy tracking systems a group of scientist from Princeton University, Max Planck Institute and University of Konstanz were able to determine the communication networks that underlie the schooling behaviour in golden shiners (movie 1 and figure 1 show how the tracking system is set up).
In the study the group has observed the effects of spontaneous startle events in groups of around 150 golden shiners. The movement of each fish was tracked after the startle event and it has been noted that the variations in the initiator (the fish that started the fast-start response) behaviour cannot explain or predict the transmission of startle signal to the responders (individuals that observe the initiator) or the rest of the group. Consequently, this led to development of a model that takes into consideration the visual fields that are available to any given responder in the group in relation to the initiator. It turns out that the function that can best predict the signal spread within a group depends on the log of the distance between initiator and responder and the angular ranking. The latter essentially refers to an observable area of the initiator for responder and it depends on the angle between observer and the initiator as well as the amount of obstructed visual field caused by any other fishes that are in between responder and initiator. So from the point of view of responder, the closer you are to initiator and the better you see it the more likely you are to respond. In addition, the more the view of the initiator is obstructed by other fishes the stronger the inhibitory (i.e. just ignore it) signal is (fig.2).
The model based on these two parameters allows to build hypothetical networks of response in schooling fish, where each string between the individuals can be have different length, strength and direction depending on individual’s position. The networks based on this model predict that the response is most likely to travel via the shortest string path because it represents the path of strongest influence in the network. In addition, it shows that individuals at the boundaries of the group are the most likely to propagate the signal throughout the group as well as to respond to signals given by others. By contrast, central individuals are the least responsive to signal communication because of multiple inhibitory signals coming from the large number of strings that they have with individuals which have not yet received the contagious signal (fig.2 and mov.2 show predicted network dynamics in schooling fish).
Now, you might wonder what’s the point of studying social contagion in fish? Obviously, no one is going to be thinking about initiators, responders and signal transmission when watching one of those amazing nature documentaries no matter how amazing it all looks. But apart from interest for interest’s sake argument (which is my primary (selfish) argument for most of science), models developed for crowd behaviour in various animals could be adapted to be used in humans as well. By adapting similar models, for example, we can predict the likelihood of a viral video or image spread on the Internet (which depends on the types of social network connections of each individual involved). More importantly, same principles can be adapted to develop models for understanding the spread of diseases. For rapid targeted response and prevention of further outbreak spread for such diseases as Ebola or flu, it is crucial to have experience and the right tools at hand to develop prediction models when need arises.
Rosenthal SB, Twomey CR, Hartnett AT, Wu HS, & Couzin ID (2015). Revealing the hidden networks of interaction in mobile animal groups allows prediction of complex behavioral contagion. Proceedings of the National Academy of Sciences of the United States of America, 112 (15), 4690-5 PMID: 25825752