Can cognitive biases promote misinformation?
What is cognitive bias?
A bias is, by definition, "an indirect and clever way of solving a problem". In a pejorative sense, we understand a bias as a "subterfuge" or a "deformation", which makes it possible to deal with a problem by compensating for cognitive limitations that can lead to cognitive saturation. It is therefore a question of the mechanisms that every one of us deploys unconsciously. They allow us to accelerate and facilitate our reasoning, especially the simplest ones, to preserve our intellectual resources for more complex tasks. Sébastien Dathané, the author of “Décider dans un monde complexe” (“Making decisions in a complex world”), points out that "our daily lives are fueled by thousands of automatic decisions, reflexes, conditioning, and various influences, especially social ones. The figure of 90% of non-conscious decisions is generally circulated". We can consider biases as rational behavior but unsuited to decision situations since they are part of a heuristic approach. According to the book Judgement and Choice, by psychologist Robin Miles Hogarth, we can consider it as "a relatively empirical approach, establishing tentative hypotheses in which imagination, experience, and personal history have a significant place."
What are the main biases?
Dozens of cognitive biases exist, some of which have been identified as particularly critical for organizations. Buster.Ai has selected eleven of them to facilitate their understanding.
Blind spot bias
A concept developed by Emily Pronin, Daniel Lin, and Lee Ross, social psychologists at Princeton University, the blind spot bias means that an individual thinks he is less biased than others.
He sees himself as more objective and with fewer biases.
We can consider confirmation bias as the cornerstone of all biases. Faced with a multitude of information, it leads to favoring an interpretation that confirms one's beliefs and refusing to consider the elements that refute them.
It can therefore naturally encourage the propagation of fake news. Indeed, an individual will be less likely to verify information if it confirms his initial beliefs.
The attention bias refers to the differences in processing several pieces of information according to the individual’s preoccupations.
If the information is favored by a stimulus, such as the perception of a threat, we speak of a facilitation bias. If the individual’s attention remains focused on another piece of information, we speak of a disengagement bias. Finally, if the attention is directed towards an opposite stimulus, it is an avoidance bias.
The representativeness bias
The representativeness bias consists of basing one’s judgment on an insufficient and unrepresentative base of elements.
The correlation illusion
The correlation illusion consists in making a causal link between two events that are in reality only slightly or not at all correlated.
The illusory truth effect
Repeated exposure to a piece of information will promote its truthfulness for the individual. A piece of information is evaluated by confronting it with its understanding.
We position a new piece of information according to the field of truths previously established. The redundancy of a piece of information makes it easier to process than a new one. The illusory truth effect can be linked to the “hindsight bias”.
This is a tendency in which an individual overestimates the probability of predicting or anticipating an event that has occurred.
With this bias, the individual tends to establish a causal link between two correlated events. The meaning of this correlation may be erroneous, or even totally illusory.
This is a human tendency to overweight a piece of information, which makes it difficult to give it a more relevant place.
This concept is reflected in the tendency to favor empirical hypothesis mechanisms and the most recently assimilated information.
Mainly studied in the context of personal life, attribution bias consists in explaining an event by the behavior of a person without considering external factors.
How do biases impact our decisions and disseminate misinformation?
First, it is important to remember that everyone has a set of beliefs, opinions, or prejudices that are based on his or her experiences and environment. When these are contradicted by one's actions or new information, the individual faces an internal conflict, called cognitive dissonance. To avoid this psychological discomfort, individuals deploy the mechanisms listed above to preserve their beliefs.
Also, a massive flow of information causes cognitive saturation. In this situation, cognitive biases will be called upon to select the information that is considered relevant and to discard information that is not perceived as such. In a static or not very complex environment, these biases can be useful to help make decisions in a relatively efficient way.
However, by creating beliefs, cognitive biases can distort the reality perceived by the individual. This is reflected at each stage of a process, starting with observation. The individual does not have a clear, factual representation of the environment in which he or she evolves. Biases can establish or, on the contrary, mask causal links while altering the analysis of those that are perceived. Thus, by constructing stereotypes, the individual develops a form of vulnerability. For example, repeated exposure to information will legitimize it. This means that false information, which the individual has been confronted with many times, and which is relayed by someone close to the individual, will give this intoxication significant credibility.
For these reasons, biases, which are based on prejudices established in situations that seem similar, can prove to be unjustified, inappropriate, and even harmful in a dynamic universe. They encourage the consideration of partial or erroneous elements.
Moreover, misinformers use techniques that rely on our biases to propagate their fake news. As we mentioned in our article dedicated to financial misinformation, campaigns can target already weakened organizations, as in the case of Metro Bank and Credit Suisse. The authors of the campaigns that targeted these banks relied on the confirmation and belief biases of the different stock market actors since factual elements already suggested the financial fragility of these two banks.
Beyond confirming beliefs, the authors of disinformation will appeal to emotion which is a powerful vector of disinformation. It has been established that surprising information has a higher degree of virality. Secondly, it is possible to generate intoxication by false dichotomies. This is a fallacy in which the authors present only two alternatives that clash, even though others exist. These situations lead to Manichean choices, either of which can be a losing proposition.
According to Professor Erik Hollnagel, when the evaluation of the situation is deficient, the heuristics will be extremely simple, often irrational, and random. In this situation, the use of cognitive biases is therefore important. To this, we can add the time pressure which does not allow for alternative visions, locking the decision-maker even more into his biases.
Finally, the dynamic and evolving nature of decision environments represents a major challenge for fact-checking. Financial wrongdoers play on the time constraint to take advantage of the situation. An actor under a time constraint is not able to perfectly inform his decision-making and will therefore unconsciously resort to his biases to make his choice.
How does Buster.Ai support organizations in the representation of the situation?
A study conducted by McKinsey shows that debiasing credit and insurance underwriting decisions can reduce losses by more than 25%. To evolve in a complex world, in which the influx of information is massive, and misinformation is omnipresent, companies and financial players must be able to rely on decision support systems that factually clarify situations, independently of the biases that may affect decision-makers. The aim is to achieve a situation of cognitive and contextual adequacy, i.e., where the decision-maker's level of understanding is as close as possible to the reality of the facts.
The solutions developed by Buster.Ai can ensure the representation of the situation, by identifying and analyzing the trends of critical information, rapidly and exhaustively.
How does it work? The algorithms developed by Buster.Ai are based on NLP (Natural Language Processing) deep learning artificial intelligence. The machine is thus able to read the documents submitted to it and compare them, to represent them in their symbols with millions of other documents from a wide variety of sources (press articles, academics, databases, reports). Thus, the Buster.Ai API solution can deliver a verdict that supports or refutes the initial query. This verdict is enriched with relevant data from the analyzed sources. It is accompanied by a chronological tool that allows for the analysis of the evolution of the trend.
Buster.Ai accompanies the decision-maker throughout the evaluation process of the situation by providing him with a precise verdict, the factual data on which it is based as well as the evolutionary dynamics of the request in real-time.
Contact us to find out more about implementing our fact-checking solution in your organization and discuss use cases and your needs.
Sources of the article :