Why do two people who read the same article draw entirely different conclusions from it? A team from the Jagiellonian University's Behaviour in Crisis Lab – Dr Anna Knorr and Prof. Małgorzata Kossowska – show in their published analysis, “Democracy in the Age of Algorithms,” that the answer lies not only in how social media function, but above all in how the human brain works. The authors combine insights from cognitive neuroscience with research on the architecture of digital platforms to explain why facts so rarely change our minds – and what this means for the future of democracy.
More Data, Less Agreement
Never before has access to information been so easy and widespread, and yet societies find it increasingly difficult to reach shared conclusions on public matters. The authors point out that the popular “information deficit model” – the assumption that mistaken beliefs stem from a lack of knowledge, and that providing people with facts is enough to make them change their minds – does not hold up against the evidence. Fact-checking rarely produces lasting changes in opinion, and exposure to well-documented information can even reinforce a recipient's prior beliefs.
The Brain as a Prediction Machine
The key to understanding this phenomenon lies in predictive coding theory. The brain does not receive the world passively; instead, it continuously generates predictions based on prior knowledge and compares them against incoming stimuli. When new information matches expectations, beliefs are reinforced. When it contradicts them, the brain must choose: revise its model of the world, or reinterpret – or even reject – the inconvenient fact. The stronger a prior belief, the harder it is to change – and that strength is shaped largely by the information we have previously encountered.
Algorithms Are Learning Our Weaknesses
Here, as the authors emphasise, the architecture of the internet plays a critical role. Digital platforms' recommendation systems operate on the logic of the attention economy – their goal is to keep users on screen for as long as possible, since screen time translates directly into advertising revenue. Content that provokes strong emotions – outrage, fear, anger – spreads faster than calm, balanced explanations. As a result, users increasingly encounter messages that confirm their existing views, and less often information that might challenge them. Under the concept of active inference, we are not passive recipients of this process – the human mind actively seeks out environments in which its predictions are confirmed, and algorithms are remarkably effective at learning to exploit this tendency.
The Stakes for Democracy
The effective functioning of democracy requires not only free elections but also trust in institutions and a shared understanding of basic facts. When citizens inhabit divergent information worlds, meaningful debate and decisions on matters of shared concern become harder to reach. The authors note, however, that today's information landscape is not an inevitable consequence of technological progress, but the result of specific design, business, and regulatory decisions – and can therefore be designed differently. They point to EU regulations such as the Digital Services Act and the AI Act as a sign that how recommendation systems function is becoming a matter of political and ethical debate, not merely a technological one.
Prof. Małgorzata Kossowska, Dr Anna Knorr, Behaviour in Crisis Lab, Jagiellonian University.
Democracy Does Not Need Unanimity
As the authors conclude, democracy does not need a society that agrees on everything, but one capable of disagreeing on the basis of shared, accepted facts. In a world increasingly shaped by algorithms, this capacity – they write – may prove to be one of the most valuable assets a democratic state possesses. The challenge, therefore, is not simply to combat disinformation, but to build an information environment that encourages the confrontation of different perspectives and the critical evaluation of information.
The Crisis of Informed Decision-Making in Modern Democracy | News - Behaviour in Crisis Lab - Strategic Program Excellence Initiative at the Jagiellonian University
