Limits of Social Learning

What are the limits of social learning in situations that require analytical thinking?
By Iyad Rahwan

Online social networking may have started out as a simple way to share news, but has become much more than that. A whole industry has sprung from the impact social media can have on people’s behavior. Scientists are exploring how to capitalize further on its potential to inspire change in people’s purchasing habits through viral marketing, and to solve difficult problems such as increasing voter participation, and searching large geographies for objects or people.

Social influence is a mechanism that allows people to learn from one another, and for habits, norms, and beliefs to propagate through society. Some argue that we owe our success as a species to our unique ability to learn from each other, thus humans live in the so-called cultural niche. This suggests that social media, as an amplifier of opportunities for social learning, is a good thing.

But in order to be able to utilize social media effectively, we need to understand what it can and cannot do. It is for that reason I collaborated with social psychologists Azim Shariff and Jean-Francois Bonnefon to explore how connectivity affects social learning.

Most of our learning takes place implicitly, by observing other people. This is especially true for social and cultural norms – things that are based on collective approval, rather than empirical facts. If you want to find out which restaurant to eat at, or what kind of joke is socially acceptable, your friends and family are a good source of information.  When you poll your circle about a question, bad quality answers filter out and approved answers rise to the top.

But when it comes to solving complex problems, it seems social networking has its limits, as our new experiment shows. In our experiments, we equally divided 100 volunteers into five social networks. Then we asked them questions that did not have obvious intuitive answers (in fact, the intuitive answer was wrong). For example, if I tell you that a bat and a ball cost $1.10 together, and the bat costs $1 more than the ball, how much does the ball cost? The answer that will jump at you is 10 cents, but the correct answer is 5 cents (think about it).

When the volunteers had to answer the questions on their own, they did badly. But when they had access to their social network, more of them were able to find out the correct answer by copying their peers. The correct answer quickly propagated over the social network over multiple rounds. In fact, the better connected a social network was, the more likely the volunteers within that network were to source the right answer from their peers. This means that the bigger the social network an individual accesses to search for a right answer, the better their chance of finding that right answer. The animated graph below shows the actual propagation of correct answers (blue) among 20 participants connected on a Barabasi-Albert network. In the first round, participants attempted the question on their own (and only 3 got it right). In each of the 4 subsequent rounds, every participant saw the answers entered by his/her neighbors on the network.

However, we were in for a surprise. To see if there was any lasting benefit or learning from the question answering process, we then asked the participants new but similar questions, this time without access to their social network. Without their network the volunteers were not any more able to answer the questions correctly. Their performance was indistinguishable from the group that had no access to a network at all! The figure below shows the evolution of the proportion of correct answers in the 5 rounds of each of the 3 questions.


This means that while social networking can help us find answers to our questions by querying our peers (provided they have the answers already) it does not teach us how to find answers on our own by thinking more critically. Next time we face a problem of a similar nature, we make the same mistakes because we have not learned how to think for ourselves.

Why does any of this matter? Well, when it comes to choosing a good restaurant, it does not matter. But when we’re trying to solve more complex problems (say, climate change), where our intuitions betray us or are simply not helpful, it matters a lot. Whether your friends think global warming is human-caused is not an objective indicator of the facts, since it takes deep study and reflection to answer this type of question. In this case, social media can become an amplifier of noise, rather than a source of helpful advice.

With this research, we now have a better understanding of how social networks can falter. It suggests that we should embrace social media for what it does best – allowing for rapid cascading of news bites, and for individuals to poll their network for answers to simple questions. But when it comes to solving complex problems that require reflection, we should recognize that today’s social networks have limits. If we want to utilize this technology to help society tackle problems like climate change or pollution, then we need to create alternative social media institutions that not only provide answers to the more complex problems, but that also promote critical thinking and learning.

See additional coverage of this work by the Royal Society, Live Science,, and the Daily Mail.