Cognitive Bias in the Global Information Subway
We “search”. We “search’” a lot. What internet critic Geert Lovink called “the society of the query” (a review), has become a reality, possibly beyond his “worst” dreams. But how does it affect us? Search is helpful but I would say that it does make it easier for us to rely on our cognitive biases, which distorts our sense of reality. To explain this effect I would like to compare the information experience of search to the cultural experience of “doing” Paris by subway. Here we go.
If you ever went to a world city like Paris or New York for a weekend trip you will know how it is like: ‘doing’ a city by subway. Quickly and conveniently you move from hot spot to hot spot. The Eifel tower is separated from the Louvre by a short efficient trip with the metro; the Louvre, in turn, just one trip away from the Notre Dame. This distorts your view of a town in an intriguing way. “My” Paris is a collection city bubbles above metro stations. It is diverse (it has everything) and culturally saturated (every spot is a hot spot). But to its inhabitants Paris can’t be the same. For me the Eifel tower is a hot spot but for locals it must be much more mundane. Possibly its glory is bleached by the familiarity of an everyday sight and knowing the city context of the tower – a context which is just not interesting enough for tourists. Well, anything appears to be saturated, when you shut out all but the brightest colors.
It is easy to draw a parallel of the information architecture of the internet to that of a big city like Paris. You can compare sites with places and hyperlinks with the streets allowing you to go from one place to the other. Search engines are the internet’s metro system. It brings you quickly to any information hot spot. This is good: without search the internet would be a barren place, but it comes at a cost. For instance, it appears as if the internet is bursting of valuable information while isn’t that ‘dense’ in reality. Also, as search is a dominant form of ‘travel’ on the internet, there is a strong concentration of interest. Popular places get a metro station – a top 10 search results ranking, and become even more popular. But the most important effect of search is that search brings us to places without telling us anything about the information context of that place.
To show why this is important, let me try to be more precise about this term ‘information context’. I would like to make a distinction between two kinds of information context that are important in search. First there is the context of production of information. If I write a blog post the information production context of this post includes: why I am writing the post, what my resources are and in what social context I am writing it. An investigative newspaper article has a different production context than a response to a question in a forum. The second type of information context is the retrieval context. With the retrieval context I refer to the underlying need for information that lies behind your query (and the quality standards that go with this need). You need different types of information when you need a quick fact to decide an argument with a friend, or when are you trying to unravel different positions in a debate for a school paper. I would argue that humans use both production context and retrieval context to give meaning to information.
However, in search queries, retrieval context becomes much more important than production context. Search gives you quick information, but it is not so selective on the production context of that information. If you take the Google subway on a medical query, its ‘trains’ could get you to Wikipedia, as well as to a forum, as well as to a news page from a pharmaceutical company; all are ‘hot spots’. While people may recognize the production context of what they find, it is not what they will attend to the most. They will be much more concerned with the question whether the information satisfies their needs. I would say that, in query, people care about retrieval context and hardly about production context.
This is problematic, because the type of information people seek out to gather and the meaning they assign to this information is tends to be biased. In cognitive and social psychology as well as in economic psychology it is well-known that people seek out for certain types of information more than other (equally important) types of information. First, people have a confirmation bias, they seek out information which is congruent with their views, rather than opposing, and they value this congruent information more. So, people do not search for information that could disprove their views and if they get confronted with such information they tend to neglect it – or value it less than congruous information. I guess that Google helps them.
Second, people have a representativeness bias. To explain this: there are two reasons you judge to belong to a certain group. First a “base rate” may be important: 30% of the Dutch population is smoker, so considering no other evidence you have a 70% per cent chance to be a non-smoker. This the “base rase”. Second, ”circumstancional evidence” may be important. Many intellectuals are smokers so if you come across as intellectual you may also believe he is a smoker. Research shows that, given information about a person (such as him being an intellectual), this information takes precedence in your judgment about me over base rate. So if I am an intellectual and you know this, in your perceptions, I am much more likely to smoke, than what the “base rate” would predict. In cognition, “specific information” takes precedence over “general information”. Again: Google, specialized in getting you specific information related to a question or query, helps the bias.
In an information experience that resembles the Paris’ subway. Where everything we meet seems to be important, authoritative and meeting our information needs. We forget that instant information with maximum relevance to our query has its limits. We learn what we want to know, biased to what we’d like to know. This distorts our information experience, in the same way the cultural experience of Paris is distorted by the subway. If much more of what we know, gets ruled by what we would like to know, our cognitive biases have free play. We search to confirm our views and we overgeneralize the results of our queries. The cultural implications of this are still to be seen, but these biases need to be compensated at some point in some way. Our world, as made up by the information about it, is not Paris. It is a normal city.
I wrote about a type of information context: conversational context in my post ‘Does Twitter have a Tempo’. I wrote about the difficulties in seeing the effects of new media in ‘Reading Marshal McLuhans’s Understanding Media’. I wrote about Googlization and Geert Lovink’s views about that, in “Thoughts About Googlization”.
A good resource on cognitive biases is ‘Social Cognition‘ by Ziva Kunda. Also about this topic, but possibly less scientific and more accessible is ‘Sway – The Irresistible Pull of Irrational Behavior ‘ by Ori and Rom Brafman.
Emile Yoffe compares the ‘Google Experience’ to addiction.
Filed under: (re)thinking media, probe | 5 Comments
Tags: Base Rate, Cognitive Bias, Confirmation Bias, Google, Information Bias, Information Experience, Kunda, Paris, Representativeness Bias, Search, Subway