Is Prior Knowledge Essential? Additional Training Opportunities Restore Sleep-associated Memory Benefits Under Conditions Of Low Prior Knowledge Part 2

3.2.2 | Recall rounds

In all recall rounds, the Dutch words were again acoustically presented in a different order than during learning, followed by a question mark on the screen. Subjects were asked to come up with the correct translation or skip the trial if the word was unknown. There was no time limit to answer. In n = 40 subjects, this was the only recall round and was taken as a pre-sleep performance level. In n = 75 subjects, one additional cued retrieval round including feedback was provided before this last recall, independent of the subject's performance. Feedback was the presentation of the correct translation and appeared on the screen for 2000 ms. In n = 39 subjects, a second cued retrieval plus feedback round was presented before the final recall round (Figure 1). This last recall performance before the retention interval was taken as the pre-retention performance level. We further calculated the relative memory performance across the retention interval. This is the percentage of remembered words after the retention with words remembered before set to 100%.

Sleep is one of the indispensable physiological needs of human beings. We need enough sleep every day to keep our bodies healthy and our brains active. In this process, people's performance level and memory also have a lot to do with it.

First, pre-sleep performance levels are closely related to our memory. If we roll over and over in bed and have difficulty falling asleep, our brains will not be flexible enough and we will be unable to remember things the next day. On the contrary, if we relax before going to bed and ensure good sleep quality, we will be more energetic and have stronger memory the next day.

Secondly, good sleep improves our performance levels. A good sleep state can ensure that our mind is clear and our body is healthy so that we can work and study more focused and confidently, and achieve better performance.

Finally, there are a few things we can do to improve performance and memory before bed. Moderate exercise, avoiding staying up late, maintaining a regular sleep schedule, and relaxing physically and mentally can all help us maintain good sleep quality and improve performance and memory.

In conclusion, there is a strong link between pre-sleep performance levels and memory. A good sleep state can help us face life and work more energetically and improve our performance level and memory. Therefore, it is very important to develop a good sleeping habit. It can be seen that we need to improve memory, and Cistanche deserticola can significantly improve memory, because Cistanche deserticola can also regulate the balance of neurotransmitters, such as increasing the levels of acetylcholine and growth factors. These substances are very important for memory and learning. In addition, Meat can also improve blood flow and promote oxygen delivery, which can ensure that the brain receives sufficient nutrients and energy, thereby improving brain vitality and endurance.

Click know ways to improve brain function

We additionally report the descriptives for the Levenshtein distances, which indicate the phonological similarity between the Dutch word and the translation in German and French, respectively. The closer to zero, the more similar the words' phonology is. For the German translation, the Levenshtein mean is 0.61 ± 0.28, and for the French 0.82 ± 0.27 (mean ± SD). Those are significantly different from each other (t119 = 7.27, p < 0.001). The median in the German list is 0.59, and in the French 1.00. The single values are listed in Table S1 in the Supplementary Materials.

3.3 | Procedure

Subjects were invited to two sessions. They either took place in the morning and evening of the same day (wake group), or across a night of sleep at home in the evening and morning (sleep group; Figure 1). On average, the wake group's morning sessions started at 08:07 hours, and the second session at 7:21 p.m. hours. In the sleep group, the evening session started at 19:03 hours, and the second appointment at 08:32 hours in the morning. At the first appointment, subjects were informed about the study flow and aim. 

After signing the informed consent, we controlled their knowledge of Dutch with the word screening task. Afterward, the word learning task was presented, followed by cued retrieval without feedback or with one or two additional feedback rounds (Figure 1b). Subjects either regularly spent their day, returning to the laboratory for the second session in the evening of the same day, or went home across the night. In the second session, approximately 12 hr later, only the cued recall without feedback was repeated. Subjects were compensated. Six–eight days after the first session, they were contacted via email with an invitation to a surprise third session to prevent the externalization of the learned vocabulary in the meantime. Again, cued recall was tested and an additional reimbursement was provided.

3.4 | Statistical analysis
We first conducted a two-way ANOVA with the between-subject factors language (German versus French) and type of retention interval (sleep versus wake). After adding the French-speaking group with two feedback rounds and the German-speaking group without a feedback trial, we conducted a three-way ANOVA with the between-subjects factors language (French versus German), retention (sleep versus wake), and memory strength (low versus high). In an additional analysis, we added learning performance as a covariate to this model and investigated the correlations between learning performance and retention performance for each group with a Pearson correlation. Significant main effects or interactions were followed up by two-way ANOVAs or independent t-tests. In the case of significant Levene's test, we corrected the degrees of freedom. The level of significance was set to p = 0.05.

4 | RESULTS

We first aimed at replicating the effect of prior knowledge on sleep-associated memory consolidation. Therefore, we compared French-speaking participants (low prior knowledge) and German-speaking participants (high prior knowledge) who learned Dutch words in an identical learning paradigm, i.e. with one cued recall with feedback. Consistent with previous findings, a robust sleep-associated memory benefit occurred in German-speaking participants: subjects remembered 100.91% ± 1.64% of the word pairs after a period of sleep (with performance at encoding set to 100%), whereas they recalled only 93.87% ± 1.69% after a retention interval filled with daytime wakefulness (t37 =
2.99, p = 0.005). 

In contrast, French-speaking participants remembered 94.17% ± 1.67% after sleep and 93.96% ± 1.67% after wakefulness. This difference was not significant (p > 0.90). In a two-way ANOVA with the factors language (German versus French) and type of retention interval (sleep versus wake), the interaction was very close to the significance threshold (F1,71 = 3.87, p = 0.053, eta2 = 0.053). Given that the direction of the interaction was set a priori in this replication attempt, the interaction is significant when using one-tailed testing (p < 0.03). Thus, our result confirms previous reports that the level of prior knowledge is important for sleep-associated memory benefits.

However, as previously suggested, prior knowledge strongly affected memory strength during the encoding phase before sleep: after two training trials, German-speaking participants were able to recall the translation of 65.90 ± 1.84 Dutch words (averaged oversleep and wake groups), whereas French-speaking participants memorized only 39.44 ± 2.03 translations. The difference was highly significant (t73 = 9.68, p < 0.001). Thus, the sleep-associated memory benefits are calculated for very different levels of memory strength at encoding in the two language groups.

To equalize memory strength during the encoding phase, we added two additional groups to our analysis: (1) French-speaking participants who received an additional feedback round (total of two feedback rounds); and (2) German-speaking participants who performed only the learning trial (no feedback round). The additional group of German-speaking participants performing only the learning trial was able to recall 38.08 ± 1.70 words, which was comparable to the memory strength of French-speaking participants performing two training trials (39.44 ± 2.03 translations, p > 0.60). Therefore, these two groups were categorized as having a “low memory strength” before sleep. French-speaking participants with two feedback rounds increased their recall performance before sleep almost (but not fully) to the level of German-speaking participants with two trainings (French: 60.05 ± 2.16 words; German: 65.90 ± 1.84 words; t76 = 2.06, p = 0.04). Still, we considered these two groups to have “high memory strength” before sleep. In the exploratory ANOVA on the last recall (without feedback) before the retention interval including all four groups using the factors language (French versus German) and memory strength (high versus low), the main effect of language was not significant (p > 0.24), and also the interaction did not reach significance (F1,150 = 3.48, p = 0.064). Only the main effect of memory strength was highly significant (F1,150 = 156.83, p < 0.001). Thus, learning performance before the retention interval was only modulated by memory strengths before the retention interval, but not by the level of prior knowledge (high for German, low for French-speaking participants; Figure 2).

We now conducted an overall analysis including all four groups using the following three factors: type of retention interval (sleep versus wake); language (French versus German); and memory strength (low versus high). If high prior knowledge would be essential for sleep-associated memory consolidation, we would expect a two-way interaction language and type of retention interval. However, this interaction was not significant (p > 0.70). Also, the three-way interaction did not reach significance (p > 0.60). However, independent of the language factor, the level of memory strength significantly interacted with the type of retention interval (F1,146 = 7.09, p = 0.009, eta2 = 0.046): a sleep-associated memory benefit occurred in both language groups when the level of memory strength before sleep was high (sleep: 101.85% ± 1.22% versus wake: 94.80% ± 0.96%; t76 =
4.57, p < 0.001), while sleep versus wakefulness did not affect retention scores when memory strength was low (sleep: 96.84% ± 1.26% versus wake: 95.73% ± 1.18%, p > 0.50; see Figure 2 for a summary of the effects of memory strength and languages group on sleep-associated memory benefits, and Figure S1 for the presentation including individual data points). On the level of main effects, retention scores were generally higher after sleep versus wakefulness (F1,146 = 12.94, p < 0.001, eta2 = 0.08), and tended to be higher for high versus low memory strengths groups (F1,146 = 3.57, p = 0.06). Finally, we observed an interaction between language and level of memory strength (F1,146 = 7.12, p = 0.009, eta2 = 0.046): the difference for retention scores in the low versus high memory strength groups was larger for French-speaking participants than for German-speaking subjects (for the descriptives, see Table 1).

In an additional analysis, we tested if learning performance had an impact on results. We first performed the correlations for each of the groups separately (Table 2). We found that while in the German-speaking sample, the correlations tended to be significant, none of them was in the French-speaking sample.

In a second step, we included learning performance as a covariate in our three-way ANOVA, reported above with the factors type of retention interval (sleep versus wake), language (French versus German), and memory strength (low versus high). This did not change any of the significances, and only turned the previous trend in the main effect of encoding level non-significant with p = 0.19. This makes sense, as in this analysis the covariate is the encoding level.

5 | DISCUSSION

Which memories are selected for sleep-mediated memory consolidation is still a matter of debate. Here, we investigated whether some effects of prior knowledge on offline retention processes can be explained by differences in memory strength already during encoding before sleep. We therefore made use of the diverging discrepancy between Dutch and German versus French. While it provides German-speaking subjects with a pre-existing schema in which newly learned Dutch words could be integrated, it does less so in French-speaking subjects. If high prior knowledge was the only relevant factor for a beneficial effect of sleep, we hypothesized that this disadvantage cannot be compensated by an increased level of memory strength, operationalized by additional learning opportunities.

Above all, retention was generally improved across sleep compared with wake, which was manifested in a highly significant and robust main effect of sleep. Neither of the other two influencing factors was as decisive for the level of retention across the interval as sleep was. Concerning our hypotheses, our results demonstrated that memory strength but not higher prior knowledge was decisive for whether sleep strengthened memories or not. Strong initial memories benefited from a night of sleep compared with a period of wakefulness, but weak memories did not. The necessary memory strength was achieved after two learning trials in Germany and three in French subjects. Thus, prior knowledge indeed enabled us to achieve the necessary memory strength earlier but this could be compensated by additional learning effort. These results suggest that in our study, it is rather the initial memory strength than the closeness of the native language to the study material that influences whether memories profit from retention during sleep or not. 

Probably, prior knowledge has a moderating role in the relationship between learning intensity and offline retention performance, as memory in German participants somewhat profited from sleep already at low learning intensities. One could assume that if a certain level of pre-existing schema exists, more learning adds more to the total memory strength than when low prior knowledge is available. This is in line with previous literature stating that learning with pre-knowledge is quicker and could be a consequence of more available resources for new learning. Comparing the growth of absolute words recalled from learning round to learning round between both language groups could have shed light on this assumption. It must, however, be assumed that the learning curve is non-linear, and the amount of additionally learned words shrinks with increasing learning rounds. However, we did not measure the growth in French subjects from round one to two, and in Germans, we had no third recall round, thus we cannot disentangle this confound in our data to test the notion.

In our initial analysis when both German-speaking and French-speaking participants had two learning rounds before sleep, only German-speaking participants showed a sleep-mediated memory benefit. Thus, in contrast to some findings (Havas et al., 2018; Payne et al., 2012), we observed a positive effect of higher levels of prior knowledge on sleep-mediated memory consolidation. Thus, although word pairs were phonologically more similar for German-speaking subjects than French-speaking subjects, they profited from an offline retention interval filled with sleep, while French-speaking subjects did not. In contrast to the study of Payne et al. (2012) and Havas et al. (2018), we did not use word pairs from the native language of the participants. It might be possible that learning familiar word pairs in one's native language is so easy that consolidation does not profit from sleep. Here, both vocabularies were in an unfamiliar language, with the only difference being that Dutch is closer to German than the French language. This was also reflected in our measures of Levenshtein distances of the word lists. On average, the phonology of Dutch words was more similar to German than French. As a limitation, instead of prior knowledge, the simple closeness of languages could have made learning easier for the German group. This fits the results of the word screening task, rather pointing towards German speakers having less prior knowledge of Dutch vocabulary than French speakers. It is known from previous literature that difficulty affects sleep-dependent memory consolidation (Cordi & Rasch, 2021). We cannot exclude that memory strength in the German group was higher due to lower difficulty as a result of the closeness of the languages. However, one could also interpret “closeness of languages” as a particular form of prior knowledge where the existing knowledge structures of “closer” languages in the brain are more similar than those of more distant languages. Regardless, there might be an optimal range of prior knowledge/memory strength required for sleep-mediated memory benefits so that too low and too high levels reduce the influence of sleep on consolidation.

It should be noted that our word-pair learning paradigm used a final test before sleep that did not include any feedback on whether the word retrieved was correct or not. This type of trial is also called “retrieval practice trial” in the context of studies on the testing effect (Roediger & Butler, 2011). A previous study has reported that word pairs learned in a retrieval practice trial do not benefit from a retention interval filled with sleep (Bäuml et al., 2014). However, in our study, a robust benefit of a retention interval filled with sleep compared with wakefulness is evident, though word-pair training before sleep ended with a retrieval practice trial. Several other studies have observed sleep benefits on memory on word pairs when retrieval practice trials were used before sleep (Feld et al., 2016; Gais et al., 2006; Schreiner & Rasch, 2015). The exact reason why Bäuml and colleagues were not able to detect a sleep benefit after word pairs learned with retrieval practice trials is not clear but might relate to the specific aspects of the stimulus material (e.g. list lengths, difficulty levels, etc.).

It was sometimes argued that pre-existing schema accelerates the speed of consolidation (Groch et al., 2017; Tse et al., 2007). This led some researchers to conclude that if no sleep effect appeared in their data, particularly those conditions with incongruent schema, more time would have been required for consolidation effects to manifest (Durrant et al., 2015). In our design, this can be excluded as a reason as all our experimental groups were exposed to the very same retention interval. Critical to the time frame of consolidation was, however, the congruency of new items with prior knowledge. Durrant et al. (2015) investigated the consolidation of schema-conformant versus nonconformant auditory memories. The conformant melodies fitting well to the pre-existing schema were preferentially and faster consolidated. This is a bit more complex with language as a semantical similarity would probably have to be distinguished from a phonological/ visual one. Still, one could question how congruent the Dutch words were to the prior knowledge of our German sample. On the one hand, the Dutch words were similar to German, on the other hand, a part of the word was not. This might have fostered more interference than “helpful” prior knowledge in terms of how congruent information would have been done. Nevertheless, if German was considered incongruent, the prediction for the pattern of the results would still be that this is superior to low prior knowledge, which French should not be able to compensate for by learning.

On the contrary, assuming that we used congruent learning material, the amount of overlap between the mother tongue and the single items did not differ systematically between groups as we used the same set of word pairs in each group. This overlap of shared elements was considered the relevant aspect of how prior knowledge helps memory consolidation in the iOtA (information overlap to abstract) concept. This model was mainly constructed to explain processes of abstraction and integration during memory consolidation. It thereby rather explains findings of improved gist extraction, statistical learning, insight, or implicit learning, while the authors concede that other forms of memory consolidation exist for which this model does not hold (Lewis & Durrant, 2011). Also, other researchers report the importance of pre-existing schema in the context of item integration, and generation of new knowledge beyond the concrete items that were learned such as in gist extraction, hidden rule learning, implicit learning of other regularities such as artificial grammar (Stickgold & Walker, 2013). The kind of learning task we used here did not require the generation of new knowledge in an integrating sense, but rather mere association learning. While in tasks like the ones mentioned above, a certain level of prior knowledge might be a prerequisite, it might be only one of several factors influencing memory strength and thereby the size of the sleep benefit in others.

AUTHOR CONTRIBUTIONS

Design of the study: TS, BR; data acquisition: TS; statistical analysis and interpretation of the data: MC, BR; manuscript draft: MC, BR; read and approved the manuscript: all.

ACKNOWLEDGEMENTS

The work was performed at the University of Zurich, Institute of Psychology, Department of Biopsychology. It was funded by a grant from the Swiss National Foundation (SNSF No. 100014_162388) and the European Research Council (ERC) under the European Union's Horizon 2020 research and innovation program (grant agreement no. 677875). Open access funding is provided by Universite de Fribourg.

REFERENCES

1. Bartlett, F. C. (1932). Remembering: A Study in Experimental and Social Psychology (Vol. 27, p. 331). Cambridge University Press. https://doi.org/ 10.1086/399084

2. Bartlett, F. C. (1932). Remembering: A Study in Experimental and Social Psychology (Vol. 27, p. 331). Cambridge University Press. https://doi.org/ 10.1086/399084

3.Born, J., & Wilhelm, I. (2012). System consolidation of memory during sleep. Psychological Research, 76(2), 192–203.

4.Brodt, S., Gais, S., Beck, J., Erb, M., Scheffler, K., & Schönauer, M. (2018). Fast track to the neocortex: A memory engram in the posterior parietal cortex. Science, 362(6418), 1045–1048.

5.Cordi, M. J., & Rasch, B. (2021). How robust are sleep-mediated memory benefits? Current Opinion in Neurobiology, 67, 1–7.

6.Diekelmann, S., Born, J., & Wagner, U. (2010). Sleep enhances false memories depending on general memory performance. Behavioural Brain Research, 208(2), 425–429.

7.Diekelmann, S., Wilhelm, I., & Born, J. (2009). The whats and whens of sleep-dependent memory consolidation. In sleep medicine reviews (Vol. 13, pp. 309–321).

8.Drosopoulos, S., Schulze, C., Fischer, S., & Born, J. (2007). Sleep's function in the spontaneous recovery and consolidation of memories. Journal of Experimental Psychology. General, 136(2), 169–183.

9. Durrant, S. J., Cairney, S. A., McDermott, C., & Lewis, P. A. (2015). Schema-conformant memories are preferentially consolidated during REM sleep. Neurobiology of Learning and Memory, 122, 41–50

10.Feld, G. B., Weis, P. P., & Born, J. (2016). The limited capacity of sleep-dependent memory consolidation. Frontiers in Psychology, 7(SEP), 1368.

For more information:1950477648nn@gmail.com