Fritz Breithaupt: Narrative Retellings Preserve Happiness and Sadness

ABSTRACT

Frederic Bartlett pioneered the research on serial reproduction in 1932 and suggested that the stereotypical or schematic form of narratives consists in rationalization, that is, a causal connection within a story. We conducted the largest retelling experiment to date and reach the conclusion that retelling of narratives is focused on the preservation of the story’s degree of happiness and sadness, even when many other aspects related to story coherence and rationalization deteriorate. These findings suggest that the happiness and sadness of a story operate as an anchor of stability for both reception-encoding and for reproduction-retrieval.

INTRODUCTION

Narratives are central to culture and its preservation over time. Retelling occurs naturally in gossip, social media, journalism, education, science, and religion.

Serial retelling studies reveal a variety of changes, including increasing of racial bias, risk, social aspects of information, bizarre story elements, and some strong emotions, while decreasing amount of information and optimizing clustering of events.

We ask: What provides stability in the process of narrative communication? What aspects of a story are transmitted with accuracy and provide the foundation for the construction of the overall story?

Emotions are a good candidate because they play a central role in recall. Narratives are well suited for emotional communication and might be the most effective means of communicating specific emotions.

HAPPINESS AND SADNESS

Our overall hypothesis is that emotions, and happiness and sadness in particular, serve as anchors of stability for retellings. Happiness and sadness are not only personal feelings, but also are commonly connected with the resolution of a narrative arc in the form of a happy or sad ending. We reason that they are decisive for story memory and reconstruction.

METHODS AND STATISTICAL ANALYSIS

In Study 1, participants created short stories that were either happy or sad. Then other participants retold these stories in their own words. They formed 116 chains of 3 retellings each.

In Study 2, we created 97 variations of 15 stories, each involving a different emotion (happiness, sadness, disgust, embarrassment, risk). With 3 retellings per chain we collected a total of 18,738 retellings.

5,234 participants rated the stories for rationalization, coherence (Bartlett), emotion, transportation, and story quality.

We developed a novel Bayesian analysis tool to estimate changes in decline (mean) and compression (variance) across retellings, see Fig. 1. [above] For the full model, see https://osf.io/68y7h/.

RESULTS

The stories shrank considerably in length to only 27.5%-28.8% after three retellings. Still, happiness and sadness both show negligible changes for slope and compression and are highly stable across retellings in both studies, see Fig. 2. 

The factors related to Bartlett’s rationalization (coherence, ability to picture the events, build up to the end) show a notable decline and a small but nonzero expansion across stories. Enjoyment declined and boringness increased as result of retelling (not shown), see Fig. 3.

For disgust the slope is rapidly declining with high compression. For embarrassment, the slope is stable, while compression is considerable. For risk, the slope is notably declining with minimal compression, see Fig. 4.

DISCUSSION

In retelling, most aspects of the original story are not well preserved from iteration to iteration. Length rapidly decreases. Ratings for coherence, story build up and ability to picture the story decline and compress. Boringness increases. In contrast, happiness and sadness both declined at a rate close to zero and showed no notable expansion or compression. So, the preservation of affect is prioritized over rationalization and story coherence.

CONCLUSIONS

How can we explain the robust preservation of happiness and sadness? We suggest that happiness and sadness in narratives function not simply as discrete emotions, but also as verdicts concerning the outcomes. Happiness rewards good behavior or lucky events, while sadness of characters at the end of a story indicates failures, bad luck, or punishment. This dual function assigns a prominent role for happiness and sadness in story reconstruction. We propose a two-phase model with an encode-and-retrieve phase and a choose-to-transmit phase, see Fig. 5. In the initial reception, the narrative sequence of events culminates in an impression of overall emotion and their magnitudes as verdict. In the second phase of reconstruction, the emotion guides the selection, (possible) invention, and assembly of those story elements needed for a reconstructed story with preserved emotion/verdict.

ACKNOWLEDGEMENTS

We thank the members of the Experimental Humanities Lab and Marco Caracciolo, Rob Goldstone, Melanie Green, Rolf Reber, Eliot Smith, and Peter Todd for their input. We are grateful for Funding from IUVPR and IDAH. Finally, we thank Norman Rockwell.