Challenging the "Neutral Summarizer" Assumption: How User Expectations Shape Perception of Bias in Political News Summaries

Burghartswieser & Rothmund (2021) and Scheufele (2000) point to the psychological underpinnings of selective exposure and perception. Yet, most studies implicitly assume that bias is an objective property of the summary, rather than a co-construction between the text and reader expectations. This experiment would present participants with identical political summaries, but randomly assign the stated "source" (e.g., mainstream journalist, AI system, left-leaning outlet, right-leaning outlet, or anonymous). By measuring perceived bias and credibility, the project would directly challenge assumptions about summary objectivity and reveal how "source cues" interact with cognitive biases (see Tang, 2024, on linguistic and expectation biases). Results could disrupt current frameworks for evaluating summarization bias, showing that the same text’s perceived slant can vary dramatically depending on context, thus calling for new multi-perspective evaluation criteria.

References:

1. Role of Language in Emotions and Bias: A Cognitive Linguistic Perspective of Affective Political Media in the US. Bolan Tang (2024). Lecture Notes in Education Psychology and Public Media.
2. Agenda-Setting, Priming, and Framing Revisited: Another Look at Cognitive Effects of Political Communication. Dietram A. Scheufele (2000).
3. Conservative bias, selective political exposure and truly false consensus beliefs in political communication about the ‘refugee crisis’ in Germany. Dominic Burghartswieser, T. Rothmund (2021). PLoS ONE.