Fig. 2. Task overview.

Participants in our main experiment watched segments of two storylines from the television series “Why Women Kill”. They made free-form text responses to either retrodict what had happened in the previous segment, predict what would happen in the next segment, or recall what happened in the just-watched segment. Across four counterbalanced sequences, we systematically varied whether participants watched the segments in forward or reverse chronological order, whether (or not) responses were cued using the main characters in the target segment, and which other segments participants had watched prior to making a response. For each segment, we collected several retrodiction, prediction, and/or recall responses across different experimental conditions. Experiment time is denoted along the vertical axis, storyline segment orders are indicated along the horizontal axis, and the colors denote experimental tasks (conditions). For an analogous depiction of our replication experiment’s design, see Supplementary Fig. S1.