**Experiment 1** Methods *Participants* 1152 participants will be recruited through Amazon Mechanical Turk. Participants will be excluded from the experiment if they do not complete the pre-experiment arithmetic task and they do not contribute words in the main experiment. Participants will be paid 1 dollar for completing the task, plus bonuses for time spent waiting for other participants to arrive in a chatroom to begin the experiment (5 dollars/hour) and bonuses for the number of correct words they recalled and participants who transmitted words to them recalled (.005 dollars/word). *Stimuli* Participants will see 60 words, randomly selected for each group of participants across all conditions, drawn from a list of 200 words from Overschelde, Rawson, and Dunlosky (2004). Words will be presented roughly simultaneously across all participants. *Procedure* Participants will observe wordlists: each word will be presented for two seconds. After seeing the list, participants will complete a 30-second-long arithmetic filler task before advancing to the recall task. Participants will placed in chatrooms alone or with other participants, and will be encouraged to type as many of the words they had seen as possible. Participants will not be told how many other participants are in the chatroom: their responses will appear in blue font, and responses from all others will appear in black. They will see all previous words entered and will not be permitted to submit any word that has already been submitted. This choice--- that any words already present on the group list will not be redisplayed--- was made to encourage participants to read others' submitted words, and because it more closely matches the lab-based version of the collaborative memory paradigm, where verbal recall creates social pressure to not repeat words. Words that are submitted by others in the chatroom, and not by the participant, will also read aloud to the participant, again to increase similarity to the lab-based task. (In the nominal experiments, no words will be read aloud, since no other participants will be present in the chatroom.) There will be no time limit for the recall task. To increase similarity to the methodology of Luhmann & Rajaram (2015), participants in the collaborative experiments will be not permitted to submit words whenever they wish (as will be true in the nominal condition). Rather, participants will take turns submitting words (a common collaborative memory paradigm, as described in Rajaram & Pereira-Pasarin (2010)) in *random order*, following Luhmann & Rajaram (2015). Participants will participate in ''rounds'', in which each participant will have the option to submit one word during their 5-second turn. Turn order will be determined randomly within each round. Participants will be able to either submit a word that had not already been submitted, press a ''Pass'' button, or wait for their 5-second turn to elapse on each turn. When participants exit the study, they will no longer be included in the turn order for the remaining participants. Experiments will contain group sizes of 2, 3, 4, 8, or 16 participants. For each recall method (nominal or collaborative), 48 groups of 2 will be analyzed; 32 groups of 3 will be analyzed; 24 groups of 4 were analyzed, and 12 groups of 8 and 16 will be analyzed. This is a 2$\times$5 design, crossing recall method by group size. In the nominal recall condition, participants will recall words alone, and their recall lists (with redundant words removed) will be added together according to the appropriate group size. In the collaborative recall condition, groups of participants will be placed in chatrooms and recall together. Recalled words that had not been on the original lists will be marked as incorrect and not included. *Analysis* We will run independent 2-sample t-tests, with alpha = .05, for all of the following comparisons: - group size 2 (collaborative vs nominal) - group size 3 (collaborative vs nominal), - group size 4 (collaborative vs nominal), - group size 8 (collaborative vs nominal), - group size 16 (collaborative vs nominal). We will also conduct a between-participants two-way unbalanced ANOVA to check for a main effect of recall method (collaborative and nominal), a main effect of group size (2,3,4,8, and 16) and an interaction effect between recall method and group size. *References* - Luhmann, C. & Rajaram, S. (2015). Memory transmission in small groups and large networks an agent-based model. *Psychological Science, 26,* 1909–1917. - Overschelde, J. V., Rawson, K., & Dunlosky, J. (2004). Category norms: An updated and expanded version of the norms. *Journal of Memory and Language, 50(3)*, 289–335. - Rajaram, S. & Pereira-Pasarin, L. (2010). Collaborative memory: Cognitive research and theory. *Perspectives on psychological science, 5(6)*, 649–663. ------ **Experiment 2** Methods *Participants* 816 participants will be recruited in Experiment 2. Participants will again be excluded from the experiment if they do not complete the pre-experiment arithmetic task and they do not contribute words in the main experiment. Participants will be sorted into one of 12 karate club networks and into 12 different small-world networks. Small-world networks will be randomly generated for each experiment. Participants will be paid as in Experiment 1. *Stimuli* The stimuli will be the same as in Experiment 1. *Procedure* To compare our results with model predictions, we sought to replicate the Luhmann & Rajaram (2015)'s model paradigm as closely as possible with human participants. Each participant will be assigned as a node in a graph with the option to communicate only with individual neighbors, where ''neighbor'' is defined as nodes participants are directly connected to. Every time a participant generates a word, their word will be shared with a randomly chosen neighbor, rather than broadcast to the entire group as in Experiment 1. There will be no turn-taking; participants will be encouraged to generate words at any time. Words shared from neighbors will be read aloud to participants, while any words participants submitted will not be read aloud, as in Experiment 1. Networks will be generated as described in the model, except that 34 agents will be included in the small-world network, to match the karate club numbers. The procedure will be otherwise the same as in Experiment 1. *Analysis* We define ''hops'' as describing how many connections are necessary to link an agent node to another. If agents are connected and can directly communicate with each other, they are separated by a hop distance of 1. If the closest path between agents includes one other node, they are separated by a hop distance of 2. We will run independent 2-sample t-tests for all of the following comparisons: - distance of 1 hop (small-world vs karate club networks) - distance of 2 hops (small-world vs karate club networks) - distance of 3 hops (small-world vs karate club networks) - distance of 4 hops (small-world vs karate club networks) - distance of 5 hops (small-world vs karate club networks) - distance of 6 hops (small-world vs karate club networks) We will also conduct a between-participants two-way unbalanced ANOVA to check for a main effect of network type (small-world vs karate club) across hop distance (1-6 hops), a main effect of hop distance (1,2,3,4,5,6 hops), and an interaction effect between network type and hop distance.