The procedure of Experiment 3 will be the same as in Experiment 1. However, this time participants will either read a vignette in which Mark punched John first and then John punched Mark (no manipulation of aspect) or a vignette in which John punched Mark after Mark punched John. The chronological order of the actions is the same for both vignettes, the order in which the actions are described is not. Another difference between Experiment 1 and 3 is that we will also manipulate whether the blame questions will be about Mark first or about John first (see [https://osf.io/32jus/][1]). So this experiment will have a 2 (order of actions) x 2 (order of questions) between subjects design. Participants will be randomly assigned to one of four conditions. We will record the time people spend reading the vignette. We expect this experiment to take about 5 minutes to complete. Participants will receive $0.5 for their participation. Participants will be recruited trough Amazon Mechanical Turk (http://www.mturk.com). The experiment will be presented online in the Qualtrics survey research suite (http://www.qualtrics.com). We conducted a power analysis to determine our sample size for this study (see [https://osf.io/4tdky/][2]). Based on the results of Experiment 2, we expect to find a large effect of order of mentioning (f = .80). This means that we will need at least 4 valid participants per condition to reach a power of .80 (total sample size = 15). However, because we expect to find small effect sizes we conducted another power analysis based on a Cohen's d of 0.25 (see [https://osf.io/sgztq/][3]). According to this power analysis we need at least 127 valid participants per condition to reach a power of .80 (total sample size = 505). Therefore, we will recruit 600 participants who will be randomly assigned to one of the four conditions. The data will be analyzed by a frequentist analysis (ANOVA). [1]: https://osf.io/32jus/ [2]: https://osf.io/4tdky/ [3]: https://osf.io/sgztq/