Hi Michaela,

Thanks for addressing my comments. I liked the idea of the confidence ratings, perhaps you could collect the data in an exploratory fashion without constructing a hypothesis about it's behavior ahead of time. It could end up being used as a mediating variable down the road.

Best of Luck,

Sharon Bertsch

Thanks to the authors for addressing many of my concerns. However, there are still a few remaining issues that I explain further in the attached:

1. Fully within-subjects design: The authors did not address why they plan to use a mixed design. Besides affording more power (see next), the use of the Nelson norms as stimuli gives plenty of options so that the cell size could increase, in case that was at all a barrier to a fully within-subjects design. I do not insist on this, but it seems like it would address likely issues of power.

2. Power: Thanks for the clarity here, but since an interaction is predicted (H2), it needs to be powered for, not just the main effect of generate vs. read. If sticking with the mixed design, many more participants than 69 (which I assume is the total) will likely be needed.

3. Inconsistencies/difficulties with the conditions explained in the Method section: I have given advice about how to make certain aspects of the Method easier to read/understand and making sure everything is clearly specified and consistent across conditions.

I would qualify these as relatively minor issues, and if anything a little bit fussy of me on the last point, but for sure powering for an interaction is pretty important. 

The current manuscript proposes an experiment crossing the generation effect (i.e., the finding that generating words yields better memory performance compared to reading words) and the multisensory facilitation effect (i.e., the finding that presenting information in two modalities, such as audiovisual, yields better memory performance compared to unisensory presentation, such as presenting only in auditory or visual format). Thus, the proposed design will follow a 2 (task: generate, read) x 3 (modality: auditory, visual, audiovisual) mixed design, with participants reading or generating words presented with related cues in different modality formats. An overall generation effect is expected, replicating extensive prior literature, as well as an interaction, such that the generation effect is stronger for audiovisual (multisensory) formats compared to audio-only or visual-only (unisensory) formats. 

Overall, the proposed study is sensible and well-justified in the literature presented in the Introduction. I think that there are some things that the authors should consider before proceeding. I have outlined them below, but I also attach the preprint with my comments that will hopefully be easier to see what I mean by these comments.

1. First, I think that the hypotheses could be streamlined/simpler. As I explain further in the comments, I think that they should be presented in reverse order to what is presented now: an overall generation effect is predicted (H1) that will be further qualified by an interaction, such that the generation effect is stronger for multisensory versus unisensory formats (H2). The proposed analysis section should then map these hypotheses onto the relevant analysis and expected result more explicitly.

2. I am not sure why the current methodological approach is presentation via flashcards. It seems not very ideal or consistent with typical cognitive research nowadays that uses computer presentation to reduce any impacts of irrelevant factors. My comments give further advice about what can be done instead, which further allow making the design fully within-subjects.

3. There are various other methodological details missing, such as the precise effect size powered for, who the participants are/their compensation, what constitutes performing the task incorrectly to warrant exclusion, and how the word pair stimuli were created along with their plausible lure options (and if the latter are also counterbalanced/randomized). Regarding the latter point, it would be more reassuring if these stimuli were already established in prior work, and I also recommend simply doing a cued recall test rather than a recognition test given that it seems like the lure options in a 3AFC recognition test could be rejected because they are not sufficiently plausible as alternatives. All of these points are detailed in my comments in the manuscript.

4. The current prediction of an interaction is actually an ordinal one, where the generation effect is expected for each modality condition but is simply stronger in the audiovisual condition. Unfortunately these are ambiguous interactions to interpret (see Loftus, 1978; Wagenmakers et al., 2012). What could be done to circumvent this issue is to transform the data and conduct the analysis again, both on the regular data and the transformed data, to show that the interaction still occurs in both cases, in which case it is unlikely to be attributable to an artifact of scale. There is a recent example of this in a recent paper that is also a registered report (Labaronne et al., 2023). Furthermore, the nature of the interaction is important for the language used in interpretation. If the generation effect is "accounted for by the multisensory facilitation effect" (page 19), then there should only be a generation effect in the multisensory condition and not the other unisensory conditions. However, if the multisensory format simply enhances the generation effect, then the predicted interaction would emerge. I also think that there are various places in the introduction that the link between different concepts (i.e., congruence/redundancy) could be spelled out more explicitly with relation to the nuance of this prediction.

References cited in this review

Labaronne, M., Jarjat, G., & Plancher, G. (2023). Attentional Refreshing in the Absence of Long-Term Memory Content: Role of Short-Term and Long-Term Consolidation. Journal of Cognition, 6(1), 5. https://doi.org/10.5334/joc.246

Loftus, G. R. (1978). On the interpretation of interactions. Memory & Cognition, 6, 312–319. https://doi.org/10.3758/BF03197461

Wagenmakers, E.-J., Krypotos, A.-M., Criss, A. H., & Iverson, G. (2012). On the interpretation of removable interactions: A survey of the field 33 years after Loftus. Memory & Cognition, 40(2), 145–160. https://doi.org/10.3758/s13421-011-0158-0