We Might Have No Free Will, But At Least We Are Influencing Half Of The World
    By Catarina Amorim | February 27th 2014 07:57 AM | 3 comments | Print | E-mail | Track Comments
    About Catarina

    After many years as a scientist (immunology) at Oxford University I moved into scientific journalism and public understanding of science. I am...

    View Catarina's Profile
    In times of easy access to the Internet and cheap travel, we consider ourselves part of a global society, but how connected this really makes us will surprise many of us.

    A Portuguese research group has found that social networks are allowing us to influence people everywhere, and not only those that we know, but also people that we never or will ever meet, which is nothing short of extraordinary. 

    Flávio Pinheiro, Jorge Pacheco and colleagues were studying the recently proposed “3-degrees of influence” rule - which suggests that we can influence not only the behavior of our friends, but also of their friends and their friends’ friends (3 degrees of influence) – and were astonished with what they found. They discovered that rather than be an exception, the rule was a “universal behavior of social networks, which is independent of the type of information, on how it spreads, and even of the type of social network that inter-connects individuals” as Pacheco explains. 

    Also surprising was to discover that the exception to the rule were networks with sparser or/and closer connections (meaning with individuals with either fewer contacts, and/or friends that are also friends of each other), in which case peer influence could reach even further (but never much more then 3).

    The study - which will be published next week in Physical Review Letters and comes from the ATP group of the Universidade do Minho, Instituto Superior Técnico and Universidade de Aveiro - have implications, not only for the understanding of behaviors but also for their potential “manipulation”, and can revolutionize fields as diverse as health education, warfare and, of course, marketing. 

    The idea of a society much more connected than we ever believed to be possible was introduced to the larger public by the Kevin Bacon’s “6-degrees of separation” ,  the game that made famous a social theory that claims that anyone in the world is 6 or fewer contacts/people away from everyone else.

    A less known, but as remarkable theory, is the “3-degrees of influence” rule, which says that our behaviour can influence others in the same social network up to 3 degrees of separation (so basically have an impact on friends of friends’ friends, even without us knowing them). 

    This rule/effect was first identified in the medical database of the Framingham Heart Study (which followed thousands of people through decades and several generations to identify factors contributing to cardiovascular diseases), and has now been linked to characteristics as diverse as smoking behavior, alcohol consumption, obesity, cooperation and even levels of loneliness or happiness.

    In the Framingham Study, for example, it was found that an obese friend increases your chances of being obese by a remarkable 40%, while a friend of a friend, or a friend of a friend’s friend increase, respectively, that chance in 20 or 10%. 

    This discovery, already extraordinary, became even more interesting when a study of hunter-gatherers from Tanzania (so a completely different society) found a similar rippling effect of influences, prompting the question of how prevalent was it, and what it meant in practical terms for society.  

    With this in mind, Pinheiro and colleagues developed a computational approach to study peer influence in networked societies, and used it to investigate the spread of cooperation, opinions and diseases. These examples were chosen because together they largely represent how information/behavior spreads in networked populations. And what they found was astounding - independently of the process studied, peer influence in any social network always reaches 2-3 degrees. This suggested that the range of influence was a quality of the social networks per se, and not related to the processes, what would explain why the patterns in real life situations (like the heart study or the research on the hunters’ tribe) were so similar. 

    Also surprising was the fact that not only the rule was valid even if the structure of the network changed, but that the only exceptions seemed to occur when people had less contacts (an unexpected result) and when their contacts were friends among themselves (what makes sense). In these cases peer influence was even more far-reaching (although never significantly bigger than 3 degrees). 

    While it is predictable that we influence our friends’ behavior, these new results tell us that we can actually have an impact on a lot more people everywhere in the world, including complete strangers that will ever know meet, as long as we are somehow linked socially, what is pretty extraordinary.

    As the researchers explain “… in medium to large communities, everybody may influence and be influenced by almost everybody else. This is, to many, an astonishing and unexpected result, given the fact that, in all cases studied, information transmission involves only two individuals at a time. Not only our choices are strongly influenced by a large number of people we actually do not know, our influence on them goes well beyond what one would naively expect.”  

    Together, this apparently universal characteristic of networks and Bacon’s 6-degrees of separation, can completely change the way we see and approach the world, from public-health policies and crime control to the understanding of emotions, the spread of innovation and even warfare.

    But the results raise as well many interesting, even philosophical, questions – where is our free will in the middle of this, if we are being influenced by factors that we do not know and can not control? And if we are connected to everyone by 6 steps and can influence 3 degrees of contacts, does that mean that we can influence half of the world?  

    Citation:  Flávio L. Pinheiro, Marta D. Santos, Francisco C. Santos, and Jorge M. Pacheco, Origin of peer influence in social networks, Phys. Rev. Lett. 2014


    Michael Martinez
    It doesn't always work that way, however.  In my Viral Propaganda Theory model I account for conflicts in the transmission of information through peer networks.  When two opposing points of view spread through a population, they ultimately collide somewhere resulting in three states:
    1. State 1: Resolution. Resolution occurs when either leader concedes to the other. The entire group then adopts the successful leader’s point of view.
    2. State 2: Conflict. Conflict occurs when both leaders move to the center of the group in an effort to win over the other leader’s followers. A conflict may result in a division of the group.
    3. State 3: Stalemate. Stalemate occurs when neither leader moves to influence the other leader’s followers. A stalemate always results in the division of the group.

    When you bring, say, an overweight person and a fit person together it does not necessarily follow that either will influence the other.  But the influence can go either way; so when studies reveal that groups of overweight people are more likely to be social peers they are not looking at the timeline.  Some groups of overweight people work together to lose and control their weight; others do not.

    We are social like the insects and herd animals but we are more likely to be influenced by new behaviors around us when we are not already engaged in conflicting behaviors.  We pass information to each other through our actions as much as through our words.

    A lot of this has already been taken into account in the fields you mention, including marketing and warfare.  The influence of contagion is used by viral marketers to manipulate the popularity and visibility of new products and services; and the influence of contagion is used by NATO to draw small communities into greater cooperation.

    The Influence of Contagion can be best described as using a credible reference to persuade someone else to change their point of view about a given topic.  The credible reference may only be someone who does something that others can see.  That might be a fish fleeing in a new direction from a predator or it might be a celebrity who comes out and says, "I had a double mastectomy".

    The natural social cohesion takes over and members of the group begin to follow the innovator.

    There will always be exceptions , this idea is that this is supposed to be the general rule and  the discovery is interesting exactly because of that - to be i an inherent characteristic of social networks.  In the same way, although social influence  is already taken into account for example in marketing, its exact reach and ubiquity was not known
    Michael Martinez
    The inherent power law in the data that studies like this characterize as "exceptions" shows that there is no "exact reach and ubiquity".  We could call that the Richard Simmons Effect, where a few of the workers in the colony rise to the status of queens or sub-queens, capable of influencing more than their usual share.
    In human social experience Dunbar's Number has found support even on the Internet.  But "there are always exceptions", people who either cannot reach that threshold (about 150 connections) or people who exceed them (on the order of millions).  These are not exceptions but simply different strata in the spectrum.

    In a viral propagation model some people have more facets/edges than others and they can propagate their point of view faster and farther than others.  That is why your chances of being found in an obese group are only increased by 40% and not by 100%.  The more-faceted Richard Simmons is actually in a position to counteract the immensely powerful forces of social peerism.

    I have been looking for a model that would show (or disprove) the idea that a social network can achieve perfect percolation -- complete and total saturation of a population with a new idea.  So far "there are always exceptions", but is there an exception to the rule of exceptions?  It should be that there is -- in which case perfect percolation is possible; but if there is no exception, if every social group has exceptions, then perfect percolation is not possible.

    That would be important to know in a variety of fields, including medicine, war, marketing, and even administration of traffic laws.  We would have to develop contingency models to spread information to the exceptions.  Only, in a world without perfect percolation we're stuck with Xeno's Paradox in which we increase the efficiency of transmission of ideas by adding increasingly inefficient secondary models.  Sooner or later we won't get to that last person in the group.