Leaping across the mental canyon: higher-order long-distance analogical retrieval

ABSTRACT

Previous experiments have shown that a comparison of two written narratives highlights their shared relational structure, which in turn facilitates the retrieval of analogous narratives from the past. However, analogical retrieval occurs across domains that appear more conceptually distant than merely different narratives, and the deepest analogies use matches in higher-order relational structure. The present study investigated whether schema abstraction can facilitate analogical retrieval of higher-order relations across written narratives and abstract symbolic problems. Participants read stories which became retrieval targets after a delay, cued by either analogous stories or letter-strings. We replicated prior research that used narrative retrieval cues, and also found clear evidence that a comparison of analogous letter-string problems and explanation of the underlying principle facilitated the retrieval of source stories with analogous higher-order relations. These findings show that the schemas abstracted from comparison of narratives can be transferred to non-semantic symbolic domains.

KEYWORDS:

• analogy
• memory
• analogical retrieval
• letter-string analogies
• cross-domain retrieval
• long-distance analogy

Disclosure statement

No potential conflict of interest was reported by the author(s).

Open practices statement

The relevant data, materials, and program code for all experiments are available at https://osf.io/fmrta/, and none of the experiments were preregistered.

Notes

1 Contingent contracts are agreements based on terms that depend on some future outcome, e.g., a pay rise if a certain performance level is met. They are particularly useful when the two parties in dispute have different predictions about the future. For example, if an author and publisher predict differing numbers of book sales, then a contingent contract would specify different terms based on what the real sales end up being.

2 Some of these studies use the spontaneous application of prior information as the key measure, while others directly ask people what they remember. We group this work together because retrieval seems to be the primary bottleneck of successful application, as often people struggle to draw spontaneous connections to analogous cues that they have been presented with, but have no problem seeing the connections once they are told that some prior information is relevant (e.g. Gick and Holyoak, 1980). Further, research has explicitly shown how retrieval and problem solving are intertwined processes, with the same factors affecting both (e.g. Bernardo, 2001; Mandler & Orlich, 1993; Ross & Kennedy, 1990).

3 We are aware that this is not how this process works at the subatomic level, but Gentner’s original SMT paper uses this explanation, so we decided to use it for simple illustrative purposes.

4 We, furthermore, used the R-packages dplyr (Version 1.1.1; Wickham et al., 2023a), effectsize (Version 0.8.3; Ben-Shachar et al., 2020), glue (Version 1.6.2; Hester & Bryan, 2022), haven (Version 2.5.2; Wickham, et al., 2023b), janitor (Version 2.2.0; Firke, 2023), knitr (Version 1.42; Xie, 2015), papaja (Version 0.1.1; Aust & Barth, 2022), purrr (Version 1.0.1; Wickham & Henry, 2023), readxl (Version 1.4.2; Wickham & Bryan, 2023), report (Version 0.5.7; Makowski et al., 2023), stringr (Version 1.5.0; Wickham, 2022), tidyr (Version 1.3.0; Wickham, Vaughan, et al., 2023), and tinylabels (Version 0.2.3; Barth, 2022).

5 The modified first-order relational schema condition will simply be referred to as the first-order relational schema condition. Only the higher-order relational schema will be differentiated between “original” and “modified” because both of those conditions were included within the same experiment.

6 We of course recognise that a 30-minute delay is not the most impressive retrieval from long-term memory, but it is certainly long enough that participants are not still actively considering the content of the encoding narratives during the delay task.

Additional information

Funding

This work was supported by Australian Research Council [grant number DP150104267].