Figure 3.

The effect of network topology on collective memory alignment. Mnemonic similarity hypotheses matrices corresponding to the clustered topology condition and the non-clustered condition in the experiment described in figure 2 are displayed. Members of the network are identified with a number, e.g. P6. The similarity scores range from 0 (distance to self; dark blue) to 5 (greatest degree of separation; dark red). Behavioural results show that on average, memories aligned more after the conversations in the non-clustered network (b) compared to the clustered network (a). This finding can be explained by the hypothesis matrices, suggesting that alignment depends on the degree of separation. Because the clustered network topology has larger degrees of separation (i.e. longer geodesic distance or network diameter), it leads to lower convergence. The extent of this alignment depends on how many degrees of separation they are from one another in the social network (c,d). (Online version in colour.)