Free-recall paradigms have greatly influenced our understanding of memory. The majority of this research involves laboratory-based events (e.g., word lists) that are studied and tested within minutes. This literature shows that adults recall events in a temporally organized way, with successive responses often coming from neighboring list positions (i.e., temporal clustering) and with enhanced memorability of items from the end of a list (i.e., recency). Temporal clustering effects are so robust that temporal organization is described as a fundamental memory property. Yet relatively little is known about the development of this temporal structure across childhood, and even less about children’s memory search for real-world events occurring over an extended period. In the present work, children (N = 144; 3 age groups: 4–5-year-olds, 6–7-year-olds, 8–10-year-olds) took part in a 5-day summer camp at a local zoo. Pathman, T., Deker, L., Parmar, P.K. et al. Children’s memory “in the wild”: examining the temporal organization of free recall from a week-long camp at a local zoo. Cogn. Research 8, 6 (2023). https://doi.org/10.1186/s41235-022-00452-z
Additional file 1: Figure S1-1 for a frequency distribution of total animals recalled for each age group.)
id: dd986c80de820baf68a44f4ed56f0c67 - page: 6
Page 6 of 12 Temporal clustering: same-context score All three age groups showed evidence of temporal clustering as measured by the same-context score, and this measure increased in magnitude as age increased (4to 5-year-olds: M = 0.80, SD = 1.27; 6to 7-yearolds: M = 1.34, SD = 1.94; 8to 10-year-olds: M = 4.21, SD = 3.98). These mean values will be called the observed same-context score for each age group.
id: e76ee698c8203e6d9ce3287af32b6b85 - page: 6
A permutation analysis was conducted to determine the statistical significance of the observed same-context scores. This analysis involved permuting the order of each participants responses and recalculating the samecontext scores on the permuted response sequences. As such, this analysis naturally accounts for differences in the number of responses across participants and groups. Thus, permutation analysis ensures any observed effects are not confounded with other aspects of behavioral performance. A permutation distribution of same-context scores was created for each age group, using a technique similar to past studies (e.g., Miller et al., 2013a, 2013b). To create a permutation distribution, we randomly scrambled the order of recall responses for each participant, calculated the same-context score using this permuted response sequence, and then averaged the scores across participants in a particular age group. This process was repeated 1000 times for each age group, yielding a
id: ef1078ae23120aa13dd6de516330f243 - page: 6
The observed same-context score was then compared to the permutation distribution; the proportion of permuted same-context scores exceeding the observed same-context score is interpretable as a p value. Figure 1 shows the permutation distribution for each age group as a histogram (1000 samples per age group), with the observed same-context score for that age group indicated with a dashed vertical line. For each age group, the observed same-context score was higher than all 1000 permutation values, indicating reliably above-chance same-context clustering (all ps < 0.001). This analysis revealed substantial temporal clustering in all age groups. But are there age group differences in the amount of clustering? Visually inspecting Fig. 1, we see that the distance between the permutation distribution and the observed same-context score for the youngest age group is smaller than that for the
id: 531fe3723b20ba536b7307acb6ef6b97 - page: 6