The Role of Recommender Systems in Amplifying Empathic Bias and Perceived Consensus

Recommender systems can personalize user experiences but also risk reinforcing "echo chambers," where like-minded views are amplified, potentially leading to increased cognitive and affective polarization. This study investigates the impact of these algorithms on empathy and cognition, within the context of legal abortion. Participants' (n = 129) attitudes toward legal abortion and their dispositional empathy were assessed. They then viewed abortion-related tweets, with (n = 71) or without (n = 58) being curated by a simulated recommender algorithm. Subsequent assessments measured their situational empathy (prosocial resource allocation, perspective-taking, and empathic concern) to simulated profiles who had clear stances on legal abortion and were in distress including death of daughter, job loss, and major injury. Findings reveal that the recommender algorithm amplified biases in cognitive empathy aspects, but it did not affect affective empathy components such as empathic concern. The algorithm also increased perceived consensus. These results underscore the need for advanced recommender systems to incorporate depolarization measures to help mitigate the negative impacts on social cohesion.