My main concerns centered around recruitment strategy and exclusions, counterbalancing of the method sequence, and clarification on randomization in reporting and the color wheel. The authors suitably amended their recruitment strategy and exclusions, included an additional discussion of why counterbalancing was not possible for their method, and clarified the randomization procedures. Additionally, I enjoyed the expanded discussion on the a priori power calculation, simulated data set, and power analysis. I look forward to hearing the results!

The authors have addressed the comments from both reviewers and made necessary changes to the manuscript. I look forward to seeing the outcome of their proposed study!

The current study looks to investigate the enhanced processing account and its memory advantage for synesthetes and experts. Specifically, they will investigate both aspects of the enhanced processing account by looking both dorsally, with sequence-space synesthetes and spatial experts, and ventrally, with grapheme-color synesthetes and color experts. They aim to find a double dissociation between the memory advantage dorsally for space and ventrally for color. There will be three versions of the visual task aimed at assessing spatial and featural (color) perception, short-term memory, and long-term memory. This is a well-designed and interesting study, with much to offer to the wider literatures on both memory and synesthesia.

• 1A: Scientific Validity
  • This research question is scientifically justifiable according to the enhanced processing theory. The prior research has only investigated the ventral pathway, while investigating both pathways simultaneously would better address the question.
• 1B: Logic, rationale, and plausibility of hypotheses
  • The logic, rationale, and plausibility of the three hypotheses are well-conceived and plausible. The hypotheses fully address the research question and the underlying theory.
• 1C: Soundness and feasibility of methodology and analysis pipeline
  • Methodology
    • Given the preponderance of drop out rates in a longer-term study such as this, as well as the difficulty in finding suitable participants, I would err on a larger initial sample size (currently only 12 participants can be excluded from each group).
      • Are participants who skip a day in the sequence excluded, or do they pick up again when they return to the experiment?
    • Inter-rater reliability, synesthesia tests, and no overlapping scores are all good controls.
    • Can the sequence of the method (perceptual task, WM, and LTM tasks, as well as loads within WM) be counterbalanced? The spatial and featural reports are already counterbalanced.
    • Will the report of features for the stimuli in the STM 3 and 5 load tasks be randomized? Or will it always go clockwise in stimulus order, for example?
      • Also in the higher VWM load task, will the beginning point on the wheel be randomly generated with equidistant stimulus positions from there, or will the beginning position remain constant?
    • Does the color wheel remain constant or does it rotate? I’m a bit concerned about having the same type of wheel for reporting space and color, as these could confound each other. Imagine a green bucket that is in the spatial position of where the color green is on the color wheel. Regardless of group assignment, one feature (e.g., color for a color expert) may help the participant remember the other feature (e.g., space for a color expert), potentially be recoding the space as where it would be on the color wheel. If the color wheel rotated, I believe this confound would be alleviated.
  • The analysis is appropriate for the hypotheses.
• 1D: Clarity and degree of methodological detail and analysis to prevent flexibility
  • The specificity of the methodology and analysis provide littler room for researcher bias, are fully linked to the hypotheses, and can be replicated.
• 1E: Outcome-neutral conditions
  • Other than the checks and exclusion criteria to ensure that no group overlaps, there are no applicable manipulation checks. The design is very well-controlled.

Review of Stage 1 Registered Report, Sachdeva et al: How perceptual ability shapes memory: An investigation in healthy special populations

This study aims to test the “enhanced processing account” of superior perceptual and memory ability in synesthetes. The authors predict that participants with color expertise – color-grapheme synesthetes and color hobbyists/experts – will perform better on color perception and memory tasks than those without such expertise. Likewise, those with spatial expertise – sequence-spatial synesthetes and spatial hobbyists/experts will perform better on spatial perception and memory tasks than those without such expertise. 

This is an interesting investigation, as there are open questions about what drives superior perception and memory performance in synesthetes compared to control subjects. The paper was well-written and easy to follow, and provided detailed information in terms of design as well planned statistical analyses. I was especially impressed with the links to various aspects of the study, including the ChatGPT output and especially the task itself. I wish all published articles would include a link to the task for readers to fully grasp what the participants are experiencing. While the methods were described clearly, there was no substitute to experiencing the task itself. 

I look forward to learning the results of this investigation. I believe this is a strong experiment, and I only have a few minor questions to be addressed in the next version of the paper. 

My primary question is with regard to the pilot data and power analyses to determine the appropriate sample size. The authors are clearly quite skilled in their statistical analyses; I trust that they are capable of performing adequate power analyses to determine an appropriate sample size for the current investigation. However, I was confused as to how the pilot study was conducted. The authors describe “simulated” data as well as “pilot” data. It is unclear how “simulated” participants were generated and why they could hold value in determining outcomes the same way true participants could. I had expected power analyses to be performed based on prior literature and/or new pilot participants, so the “simulated” participants threw me. More detail is needed here. 

The rest of my concerns are minor. Also in terms of participant recruitment, I would expect there to be considerable overlap among groups. It is common for people with synesthesia to have multiple forms of it, with both grapheme-color and sequence-spatial to both be quite common. As for the non-synesthetes, the color and spatial experts are also likely to have considerable overlap; indeed, the ChatGPT output listed many of the same professions and hobbies for both color and spatial expertise. Finally, it would be unsurprising to find that color synesthetes are likely to share in the color expertise hobbies/professions. That said, the authors include a detailed plan to recruit participants with a screening questionnaire and will continue to seek participants in batches of 1000 until they have the appropriate pool. I did, however, have one question about the target sample sizes. The authors indicated that they targeted 40 participants per group, and would recruit 52 per group to aim for that number, in order to account for eliminations. Would these eliminations be due to performance on the experimental tasks, or would they also include issues such as “overlap” in groups? If the former, I think it is an appropriate plan. If the latter, I would recommend handling overlap cases (since there may be many) in the screening procedure prior to the main experiment.

In terms of the results, the authors make a prediction of an interaction between domain and task in that color experts will perform better on the color task than the spatial task, and spatial experts will perform better on the spatial task than the color task. I wonder if this prediction should be softened to merely predict an interaction including “relative” superiority rather than a more specific outcome. In other words, if one task happens to be easier than another, perhaps both groups will perform better on the color task than the spatial task, even if the color experts perform more better. Either way, the result will be impactful. 

Also in the predictions, the authors state that they predict on the short term memory task for domain, task, and “load” to interact because at load of five there will be floor effects in memory performance, so the domain/task interaction may change compared to loads one and three. The authors then state that for the long term memory task, the critical interaction will be among domain, task, and “day,” but there is no prediction about how “day” will affect the primary interaction. I think it would be useful to include such a prediction, even if it is that “day” will not interact with the variables of interest. 

I finally have a few questions about the synesthesia screen as well as how synesthetes might behave in the primary task. The authors mention that in order to investigate sequence-spatial synesthesia, participants will indicate locations of sequences on the computer. However, many sequence-spatial synesthetes experience their spatial percepts in three dimensions. Will there be a way to account / screen for this experience? If not, how will this dimension limitation affect the screening process for this kind of synesthete? 

During the task itself, do you anticipate an issue with synesthetes (particularly color-based ones) experiencing a synesthetic reaction to a stimulus? For example, the example piano stimulus jumped out at me as one that might elicit a color reaction to it. For those with chromesthesia, the synesthete might associate the timbre of a piano with a particular color that may or may not map onto the color the object is given in the task. It could undermine the results if pre-existing synesthetic mappings for particular objects might be incongruent with the assigned colors in the experiment. Perhaps a questionnaire asking for feedback or reactions could take care of this concern, and any unusual observations could be reported in the discussion. 

Is there a reason that there are very slightly different numbers of simuli / trials in the different loads of the memory task? (This doesn’t need to be changed because they are close enough, but I am just curious.) 

This might have been a coincidence, but I noticed in the demo of the task that in loads 3 and 5, the objects appeared in spectrum order in terms of color (e.g. clockwise: red, golden, green, blue, purple). I wonder if this affects strategy, because in that case it might be necessary only to remember spatial location and then fill in the colors based on where they might appear in order. If the colors are in fact truly randomized, ignore this question! I did appreciate the methodological explanation that described how the colors were selected on each trial to ensure proper color distance; I’m just curious about how they were arranged spatially within a single trial.