mobile icon


WorkgroupPerception and Action Lab
Duration07/2020 - 06/2023
FundingSondertatbestand Data Science
Project description

When navigating the Internet, people are confronted both with attitudinally congenial and uncongenial information. Typically, congenial information is read and processed more superficially. The long-term goal of this project is to develop a software agent that uses reading time information to infer whether someone is reading congenial or uncongenial information. The agent can then adapt the presentation of content in a way that pro and con arguments are read with the same depth.

Before such a software agent can be developed, a number of basic scientific questions have to be clarified within this project. First, prior research results have to be replicated to find out whether uncongenial information is actually read more slowly and carefully than congenial information. On this basis, it can be assessed whether a software agent could correctly predict congeniality of content on the basis of reading times. Second, several methods derived from research on event cognition will be experimentally tested which involve the introduction of so-called event boundaries into the reading flow (e.g., topical shifts, presentation of videos). Inserting event boundaries should increase processing depth. Third, the project will analyze whether inserting event boundaries have an impact on reading speed, memory performance for the content, and on the attitudes of readers.

  • Prof. Dr. Hendrik Lensch, Computer Science, University of Tübingen