Open-Source Benchmarking of Plant-Based and Animal Meats

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The paper itself is interesting but the analysis is lacking. some major concerns around statistical soundness.

Methods

Were there ethics approval for this? If yes, add.

In data analysis, separate the question and the statistical approach, everything is mixed up here..

There's only 100 participants per product category, how did the author control for the variation of product consumed here? (Fig 1)

The referencing number seems odd and will need to be fixed

Fig 4. Is the ">" and "<" inverted? supposed to be p <0.001 as an example

L150, Fig ?? something seems wrong with the referencing here?

Fig 6. overall liking do not indicate preference, so this assumption is not correct.

Fig 7. rather than having promoters, passives, and detractors - shouldn't a cluster analysis gets carried out here before segmenting this?

Fog 9. Wouldn't running a SEM be more valuable here? or perhaps a classical sensory drivers of liking?

iThenticate shows that this is 82% similar to their arxiv paper https://arxiv.org/html/2603.03370v1 pls check with journal policy on this

Author Response

The paper itself is interesting but the analysis is lacking. some major concerns around statistical soundness.

Thank you for the constructive comments and thoughtful feedback. We have now addressed all comments. All edits are highlighted in the revised manuscript and outlined in detail below.

[1] Were there ethics approval for this? If yes, add.

Thank you for this comment. We have now highlighted the IRB review feedback in the appropriate IRB section at the end of the manuscript:

“Institutional Review Board Statement: This research was determined by the Stanford University Institutional Review Board to not involve human subjects as defined in 45 CFR 46.102(e).”

[2] In data analysis, separate the question and the statistical approach, everything is mixed up here.

Thank you for this comment. We have now restructured the section and clarified that the study evaluated category-level sensory performance rather than precise ranking of individual formulations and therefore aggregated responses across products within categories to reduce sensitivity to product-specific variability.

Data analysis: “We calculated mean scores for each product and compared three benchmark groups […] Participants also completed check-all-that-apply (CATA) questions […] Because the study aimed to benchmark category-level sensory performance rather than precisely rank individual formulations, we aggregated responses across products within each category to reduce sensitivity to product-specific variability.”

[3] There's only 100 participants per product category, how did the author control for the variation of product consumed here? 

Thank you for this remark. We have now clarified that the study focuses on category-level sensory trends and relative benchmarking rather than precise estimation of individual product rankings, which reduces sensitivity to within-category product variability.

Data analysis. “Because the study aimed to benchmark category-level sensory performance rather than precisely rank individual formulations, we aggregated responses across products within each category to reduce sensitivity to product-specific variability.”

[4] The referencing number seems odd and will need to be fixed.

Thank you for this comment. The references were numbered alphabetically.

[5] Fig 4. Is the ">" and "<" inverted? supposed to be p <0.001 as an example

Thank you for catching this error. We have now revised it according to your comment. 

[6] L150, Fig ?? something seems wrong with the referencing here?

We apologize that this Figure was not properly cross-linked and included. We have now fixed this and included the Figure and its reference.

[7] Fig 6. overall liking do not indicate preference, so this assumption is not correct.

We thank the reviewer for this clarification and revised the manuscript to distinguish overall liking ratings from direct within-subject preference comparisons.

Figure 6. “Although the animal product received slightly higher overall liking ratings, within-subject comparisons were broadly split…”

Figure 8. “For each category, we inferred relative preference from paired within-subject overall liking comparisons…”

[8] Fig 7. rather than having promoters, passives, and detractors - shouldn't a cluster analysis gets carried out here before segmenting this?

We agree that consumer segmentation could provide additional insight, and we now acknowledge in the Discussion that the current analysis focuses on aggregate sensory trends rather than latent preference clusters.

Discussion. “Future work could further resolve heterogeneous consumer subgroups through clustering or latent preference modeling to identify category-specific acceptance patterns.”

[9] Fig 9. Wouldn't running a SEM be more valuable here? or perhaps a classical sensory drivers of liking?

Thank you for this suggestion. We agree that structural equation modeling could provide additional insight into multivariate sensory relationships, and we now acknowledge this opportunity in the Discussion.

Discussion. “Future work could further resolve these interactions through structural equation modeling and causal sensory frameworks.”

[10] iThenticate shows that this is 82% similar to their arxiv paper https://arxiv.org/html/2603.03370v1 pls check with journal policy on this.

This is correct. This arXiv paper is the preprint related to this publication. This preprint and its citations will be crosslinked to the paper once published.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

In this manuscript the authors present a very large blinded sensory study comparing 121 plant-based foods with 14 animal foods, spread across 14 categories. In total 2,684 U.S. volunteers participated in more than 11,000 sensory evaluations. The results show that in overall liking plant-based foods are rated lower than their animal counterparts. However, there are several product categories where the liking score of the plant-based foods even equals or exceeds that of the corresponding animal food. Notably, such similarity of liking scores between plant-based and animal foods is strongly correlated with the respective market shares of these plant-based products. The data set underlying the manuscript is provided as open resource.

The study has a good scale and design. With around 100 tasters per product (tasted in their restaurants), in a blinded within-subjects design, and a sufficiently balanced panel of omnivores and flexitarians the data should be good.

Major issues

1. Fig. 4 reports 70 Wilcoxon tests; Figs. 7 and 9 add more. With 70+ tests, some "significant" results are expected. The effect of sizes should be reported, not only p-values.
2. "No significant difference = parity" is statistically wrong. It only fails to detect a difference. This appears in Fig. 6 (lines 164–165: "p = 0.314"). 
3. "Leader" is defined as the highest-rated plant-based product per category (lines 113–114). When ~9 products per category are tested and the maximum is selected and compared to one benchmark, the leader's score is upward-biased. This inflates the apparent closeness of leader-vs-benchmark.
4. Market-share data in Fig. 12: Only "plant-based sales divided by total category sales in 2024 retail data" (lines 234–235). No source, no channel coverage (retail? foodservice?), no category definitions. Since this figure supports a main claim, the source and method must be described in Section 2.
5. Figure 4 caption says "Across all 5 × 14 rankings, in 11 categories, there was no significant difference," but 5 × 14 = 70 comparisons, not 70 categories. The text says "11 cases out of 70." The caption and text should be coherent.
6. AI section (lines 300–318): Many self-citations about AI/PINN/generative AI methods, but none used in this paper. Sounds like an advertisement, not a discussion of present results.
7. The study collected sensory and demographic data from 2,684 identifiable consumers, which most reviewers would call human subjects’ research. Saying it "does not involve human subjects" (lines 355–356) without explanation is unusual. Justify.
8. NECTAR/Food System Innovations is an alt-protein advocacy organization, and the funding includes Food System Innovations and Humane America Animal Foundation. The "no conflicts" statement (line 364) does not fit with this. More transparent disclosure is needed.

 

Minor Issues

• Section 1 is titled "Motivation" — In Foods shouldn’t it be "Introduction"?
• Fig. 4 caption: significance footnote uses ">" instead of "<". Appears multiple times.
• Line 150 "Fig. ??"
• Jumps from Figure 4 to Figure 6.
• Line 166: duplicated "that."
• Fig. 7 : says "n = 100 per test" but isn’t it about % and not the number of evaluations?
• Fig. 12: what is the source of the 2024 retail data? Dollar, unit, or volume share? Which channels?
• Line 269: "almost en par with" — should be "on par with."
• Reference 35: missing DOI.

 

Author Response

In this manuscript the authors present a very large blinded sensory study comparing 121 plant-based foods with 14 animal foods, spread across 14 categories. In total 2,684 U.S. volunteers participated in more than 11,000 sensory evaluations. The results show that in overall liking plant-based foods are rated lower than their animal counterparts. However, there are several product categories where the liking score of the plantbased
foods even equals or exceeds that of the corresponding animal food. Notably, such similarity of liking scores between plant-based and animal foods is strongly correlated with the respective market shares of these plant-based products. The data set underlying the manuscript is provided as open resource. The study has a good scale and design. With around 100 tasters per product (tasted in their restaurants), in a blinded within-subjects design, and a sufficiently balanced panel of omnivores and flexitarians the data should be good.

Thank you for this encouraging feedback. We appreciate your detailed comments and suggestions.

Major issues

1. Fig. 4 reports 70 Wilcoxon tests; Figs. 7 and 9 add more. With 70+ tests, some "significant" results are expected. The effect of sizes should be reported, not only p-values.

Thank you for this comment. We agree that multiple category-by-attribute comparisons increase the likelihood of false-positive findings if interpreted solely through individual p-values. Our conclusions therefore emphasize the consistency, direction, and magnitude of the observed effects across categories rather than isolated significance tests. We revised the manuscript to clarify that the analyses are primarily
comparative and exploratory, and we now report effect magnitudes alongside p-values using mean score differences (Δ) across categories, for example as indicated below:

Methods. “[…] Because the study involved multiple predefined category-level comparisons, we interpret statistical significance together with effect magnitude and consistency across categories rather than as isolated hypothesis tests.”

Results. “[…] For attributes that showed significant differences, the effect magnitudes also varied by category and attribute, from unbreaded chicken filet (Δ = 0.1) to bacon ( Δ = 2.0), suggesting that certain animal products may be more successfully replicated than others.”

2. "No significant difference = parity" is statistically wrong. It only fails to detect a difference. This appears in Fig. 6 (lines 164–165: "p = 0.314").

Thank you for highlighting this inconsistency. We agree that “no significant difference” could imply equivalence rather than failure to detect a directional preference. We have revised the text and figure caption to clarify that the Wilcoxon signed-rank test did not detect a statistically significant preference for either product, rather than concluding equivalence between products:

Results. “The Wilcoxon signed-rank test did not detect a statistically significant directional preference between the plant-based and animal products (p = 0.314). These results suggest that, for this product category, the leading plant-based product approached the animal benchmark sufficiently closely to support the technical feasibility of taste parity.”

3. "Leader" is defined as the highest-rated plant-based product per category (lines 113–114). When ~9 products per category are tested and the maximum is selected and compared to one benchmark, the leader's score is upward-biased. This inflates the apparent closeness of leader-vs-benchmark.

We agree that defining the “plant-based leader” as the highest-rated product within each category introduces an upward-selection bias that can overestimate apparent closeness to the animal benchmark. We have clarified this point in the Discussion and now explicitly distinguish between category-level averages, which remained substantially below the animal benchmark, and best-in-category leader products, which illustrate the current technical ceiling rather than typical market performance.

Discussion. “[…] Because we defined the plant-based leader as the highest-rated product within each category, these comparisons intentionally represent best-case category performance and therefore likely overestimate the average sensory competitiveness of commercially available plant-based products. The substantially lower scores of the plant-based category averages highlight that sensory parity remains uneven across the broader market despite the emergence of strong individual products.”

4. Market-share data in Fig. 12: Only "plant-based sales divided by total category sales in 2024 retail data" (lines 234–235). No source, no channel coverage (retail? foodservice?), no category definitions. Since this figure supports a main claim, the source and method must be described in Section 2.

This is an excellent point. We have now added this additional section and reference to the Methods in Section 2.

Methods. “To relate sensory performance to commercial adoption, we analyzed 2024 U.S. retail sales data from publicly available industry reports, retail scanner summaries, and NECTAR category analyses [25]. We defined plant-based market share as plant-based category sales divided by total category sales, including both plant-based and animal-based products, and defined sensory performance as the percentage of consumers who rated the plant-based product as the same or better than the animal benchmark in paired overall liking comparisons.”  

5. Figure 4 caption says "Across all 5 × 14 rankings, in 11 categories, there was no significant difference," but 5×14=70 comparisons, not 70 categories. The text says "11 cases out of 70." The caption and text should be coherent.  

Thank you for pointing out this typo. We have now replaced “categories” with “comparisons” in the figure caption and aligned it with the corresponding text.

Results. “[…] Across 70 category-by-attribute comparisons, the animal benchmarks scored significantly higher than the plant-based leaders in 59 cases, while 11 cases showed no significant comparisons (Fig.4).”

Figure 4. “[…] Across all 5 s 14 rankings, in 11 comparisons, there was no significant difference between animal benchmark and plant leader; in 59 comparisons, the animal benchmark scored significantly higher than the plant-based leader.  

6. AI section (lines 300–318): Many self-citations about AI/PINN/generative AI methods, but none used in this paper. Sounds like an advertisement, not a discussion of present results.  

We have now removed the section on AI/PINN/generative AI, including three citations to our work and work of our collaborators.  

7. The study collected sensory and demographic data from 2,684 identifiable consumers, which most reviewers would call human subjects’ research. Saying it "does not involve human subjects" (lines 355– 356) without explanation is unusual. Justify.  

Thank you for letting us clarify this concern. Thank you for raising this concern. This study did not involve direct interaction or intervention with participants by the authors. Palate Insights independently collected the sensory and demographic survey data as part of the Taste of the Industry 2025 report, and all data used in this manuscript were fully de-identified before the authors accessed them. The authors performed a secondary analysis of existing de-identified data and generated no additional human-subject data as part of this study. Stanford University reviewed the protocol and determined that this research does not involve human subjects as defined in 45 CFR 46.102(e). We have included this in the IRB statement:

Institutional Review Board Statement. This research was determined by the Stanford University Institutional Review Board to not involve human subjects as defined in 45 CFR 46.102(e).  

8. NECTAR/Food System Innovations is an alt-protein advocacy organization, and the funding includes Food System Innovations and Humane America Animal Foundation. The "no conflicts" statement (line 364) does not fit with this. More transparent disclosure is needed.  

Caroline Cotto is an employee at NECTAR, as disclosed in the affiliations. We have now emphasized this relation in the funding section. We have also emphasized in the COI section that there is no conflict of interest in relation to any of the tested products.

Funding. This research was supported by Food System Innovations, Humane America Animal Foundation to Caroline Cotto, and by seed funding from the Stanford Bio-X Snack Grant and the Stanford Doerr School of Sustainability Accelerator, and by the NSF CMMI grant 2320933 and the ERC Advanced Grant 101141626 to Ellen Kuhl.

Conflicts of Interest. The authors declare no conflicts of interest in relation to any of the tested products.

Minor Issues

• Section 1 is titled "Motivation" - In Foods shouldn’t it be "Introduction"?  

Thank you for this comment. We would be happy to change this if needed.  

• Fig. 4 caption: significance footnote uses ">" instead of "<". Appears multiple times.  

That you, we have now changed all “>” to “<”.  

• Line 150 "Fig. ??" / Jumps from Figure 4 to Figure 6.  

We apologize. We have now fixed this.  

• Line 166: duplicated "that."  

Thank you, we have removed the second “that”.  

• Fig.7: says "n=100 per test" but isn't it about % and not the number of evaluations?  

Thank you for this comment, this is correct. We just added the "n=100 per test" to indicate that the percentages are generated from 100 evaluations per category.  

• Fig. 12: what is the source of the 2024 retail data? Dollar, unit, or volume share? Which channels?  

Thank you. We have now included additional information in the Results section and in the figure caption:

Results. “We defined market share as plant-based sales divided by total category sales, plant-based and animal-based, in 2024 retail data [25].”

Figure 12. “[…] Scatterplot shows plant-based market share (percentage of plant-based products of 2024 total retail sales, including both plant-based and animal based) as a function of sensory performance (percentage of consumers rating plant-based products same or better than animal benchmark in overall liking).”

• Line 269: "almost en par with" - should be "on par with."  

Thank you. We have fixed this typo.  

• Reference 35: missing DOI.  

Thank you for catching this. We have now added the doi:10.48550/arXiv.2604.03409.

 

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

This manuscript reported the findings of a large-scale, double-blind sensory evaluation comparing plant-based and animal-derived meat products across 14 distinct categories, involving 2,684 consumers and encompassing more than 11,000 individual product assessments. The objectives are clear and the study has some significance. However, the following problems should be revised.

1. Please ensure that spaces are added before and after the symbols “=”and “>”. The manuscript contains multiple instances of inconsistent symbol formatting. Please check the entire manuscript carefully.
2. “p” in “p > 0.001”, “p > 0.01” and “p > 0.05” should be italicized in the manuscript.
3. The text exhibits numerous inconsistencies in citation placement, formatting, and the alignment between in-text reference numbers and the corresponding entries in the reference list.
4. The manuscript exhibits inconsistencies in its chapter numbering system, and headings across all hierarchical levels lack adequate spacing before and after the main text. Consequently, the overall layout appears visually congested and lacks structural clarity. There are numerous instances throughout the manuscriptwhere headings are immediately followed by body text without any typographic separation, contravening standard academic formatting conventions. To align with the target journal’s formatting guidelines, the author should implement a consistent, hierarchical numbering scheme (e.g., 1, 1.1, 1.1.1) and apply uniform paragraph spacing to ensure a clear visual distinction between headings and adjacent text.
5. Line 47:The term “eliminating” is overly absolute and inconsistent with empirical evidence. While plant-based meat products can substantially mitigate the environmental impacts associated with conventional livestock production, their manufacturing and processing still entail greenhouse gas emissions, land use, and water consumption. This means that they do not eliminate environmental burdens entirely. Therefore, it is advisable to replace “eliminating” with more precise and empirically grounded terms such as “reducing” or “mitigating.”
6. The citation “(Fig. ??)” is incomplete and constitutes an oversight in referencing the figure.
7. Section 2.1 explicitly stipulates that a minimum of five plant-based products per category must be selected as the eligibility criterion for inclusion across the 14 predefined categories. However, the subsequent statement“Among these categories, we prioritized 121 plant-based products for screening” introduces ambiguity. Given the stipulated minimum of five products per category, the theoretical lower bound for the total screened products is 70 (14 × 5). In contrast, the reported figure of 121 products corresponds to an average of approximately 8.6 products per category, suggesting variable sampling intensity across categories. This discrepancy necessitates clarification on two key points: whether all 14 categories indeed met the ≥ 5-product threshold for viable candidate selection; and the operational definition of “prioritized”. Specifically, whether product selection was guided by objective, pre-specified criteria such as market share, commercial availability, formulation complexity, or fidelity to animal-meat sensory and nutritional profiles, rather than exhaustive enumeration of all commercially available options. These methodological details should be explicitly articulated in the manuscript to ensure transparency, reproducibility, and interpretability of the sampling strategy.
8. The text contains redundant instances of the conjunction “that” which impair sentence fluency and undermine academic precision.
9. The fixed expression “neither like nor dislike” occurs repeatedly in the manuscript; however, it is used inconsistently in the current text.
10. Line 269:A typographical error has been identified: the academically standard fixed phrase is “on par”, not “en par”. The author is kindly requested to revise all instances of “en par” to “on par” consistently throughout the manuscript and to conduct a comprehensive review of similar collocational or idiomatic expressions to ensure linguistic accuracy and terminological consistency.

Author Response

This manuscript reported the findings of a large-scale, double-blind sensory evaluation comparing plant-based and animal-derived meat products across 14 distinct categories, involving 2,684 consumers and encompassing more than 11,000 individual product assessments. The objectives are clear and the study has some significance. However, the following problems should be revised.  

Thank you for your encouraging feedback. We have included all comments in the revised manuscript, highlighted them in red, and list them in the response below.

1.Please ensure that spaces are added before and after the symbols “=”and “>”. The manuscript contains multiple instances of inconsistent symbol formatting. Please check the entire manuscript carefully.

Thank you for this comment. We have now consistently added spaces before and after these symbols.  

2. "p"in "p<0.001" , "p<0.01" and "p<0.05" should be italicized in the manuscript.

Thank you. We have now italicized “ "p" ” across the entire manuscript.

3. The text exhibits numerous inconsistencies in citation placement, formatting, and the alignment between in-text reference numbers and the corresponding entries in the reference list.

We have thoroughly addressed these inconsistencies. We had previously listed the references alphabetically and are now listing them in chronological order.

4. The manuscript exhibits inconsistencies in its chapter numbering system, and headings across all hierarchical levels lack adequate spacing before and after the main text. Consequently, the overall layout appears visually congested and lacks structural clarity. There are numerous instances throughout the manuscriptwhere headings are immediately followed by body text without any typographic separation, contravening standard academic formatting conventions. To align with the target journal’s formatting guidelines, the author should implement a consistent, hierarchical numbering scheme (e.g., 1, 1.1, 1.1.1) and apply uniform paragraph spacing to ensure a clear visual distinction between headings and adjacent text.

Thank you for this comment. While uniform hierarchical numbering could improve structural consistency, the manuscript intentionally follows a compact narrative structure with short subsections, topic sentences, and minimal hierarchy. We therefore prefer to preserve the current organization to maintain readability and narrative flow, while carefully revising formatting consistency throughout the manuscript.

5. Line 47:The term “eliminating” is overly absolute and inconsistent with empirical evidence. While plantbased meat products can substantially mitigate the environmental impacts associated with conventional livestock production, their manufacturing and processing still entail greenhouse gas emissions, land use, and water consumption. This means that they do not eliminate environmental burdens entirely. Therefore, it is advisable to replace “eliminating” with more precise and empirically grounded terms such as “reducing” or “mitigating.”

Thank you for catching this. We have now replaced “eliminating” but “reducing” as suggested above.

6. The citation “(Fig. ??)” is incomplete and constitutes an oversight in referencing the figure.

Thank you for this comment. We apologize, this was a technical mistake on our end. We have now repaired the link and included the figure in the revised version.

Figure 5. Overall liking, similarity, flavor, texture, appearance, and purchase intent scores. Performance of animal benchmark (gray) and plant-based leader (dark color) scored by mean overall liking, similarity, flavor, texture, appearance, and purchase intent across all 14 product categories, bacon, bratwurst, breaded chicken filet, breakfast sausage, burger, chicken nuggets, deli ham, deli turkey, hot dog, meatballs, pulled pork, steak, unbreaded chicken filet, and unbreaded chicken strips. Approximately 100 participants evaluated each product ( (meanpertest=99.4) on a seven-point Likert scale from dislike very much (1) to like very much (7). Across all 5 x 14 rankings, in 11 categories, there was no significant difference between animal benchmark and plant leader; in 59 categories, the animal benchmark scored significantly higher than the plant-based leader; *** ∗∗∗p<0.001 ** p<0.01 , ∗p<0.05 Thank you for this comment. We apologize, this was a technical mistake on our end. We have now repaired the link and included the figure in the revised version.

 

7. Section 2.1 explicitly stipulates that a minimum of five plant-based products per category must be selected as the eligibility criterion for inclusion across the 14 predefined categories. However, the subsequent statement “Among these categories, we prioritized 121 plant-based products for screening” introduces ambiguity. Given the stipulated minimum of five products per category, the theoretical lower bound for the total screened products is 70 (14 × 5). In contrast, the reported figure of 121 products corresponds to an average of approximately 8.6 products per category, suggesting variable sampling intensity across categories. This discrepancy necessitates clarification on two key points: whether all 14 categories indeed met the ≥ 5-product threshold for viable candidate selection; and the operational definition of “prioritized”. Specifically, whether product selection was guided by objective, pre-specified criteria such as market share, commercial availability, formulation complexity, or fidelity to animal-meat sensory and nutritional profiles, rather than exhaustive enumeration of all commercially available options. These methodological details should be explicitly articulated in the manuscript to ensure transparency, reproducibility, and interpretability of the sampling strategy.

Thank you for encouraging us to explain the product selection in more detail. We have now revised the product selection section to clarify that the study targeted commercially prominent and broadly representative categories and products rather than an exhaustive list of products that includes all available formulations within each category.  

Tested products. “We selected fourteen product categories based on two criteria: high sales volume and sufficient market development with at least five plant-based products available within the category (Fig. 1). The study targeted commercially prominent and broadly representative categories rather than exhaustive coverage of all available plant-based products. Within these categories, we selected 121 plant-based products, 5-10 per category, mean 8.6, based on significant market presence, marketready status, design as animal analog, original flavor comparable to the animal product, and distinct ingredients or production technologies relative to other products tested within the category. For each category, we selected a conventional animal benchmark based on highest retail sales volume. The fourteen categories include bacon, bratwurst, breaded chicken filet, breakfast sausage, burger, chicken nuggets, deli ham, deli turkey, hot dogs, meatballs, pulled pork, steak, unbreaded chicken filet, and unbreaded chicken strips.”

8. The text contains redundant instances of the conjunction “that” which impair sentence fluency and undermine academic precision.

Thank you. We have now fixed this and removed the conjunction “that”.

9. The fixed expression “neither like nor dislike” occurs repeatedly in the manuscript; however, it is used inconsistently in the current text.

Thank you for catching this inconsistency. We are now consistently using the terminology “neither like nor dislike” throughout the entire manuscript.

10. Line 269:A typographical error has been identified: the academically standard fixed phrase is “on par”, not “en par”. The author is kindly requested to revise all instances of “en par” to “on par” consistently throughout the manuscript and to conduct a comprehensive review of similar collocational or idiomatic expressions to ensure linguistic accuracy and terminological consistency.

We have now addressed this comment, corrected “en par” to “on par” and checked the manuscript for terminological consistency.

Author Response File: Author Response.pdf

Reviewer 4 Report

Comments and Suggestions for Authors

Overall, I find the manuscript interesting, current, and potentially valuable for researchers, product developers, and stakeholders in the alternative protein industry. The open-access approach to sharing the sensory dataset is also commendable, as it promotes transparency and reproducibility in a field where much of the available data remains proprietary.

However, although the manuscript has clear scientific merit, I recommend revisions prior to acceptance, as several methodological and editorial aspects should be clarified to strengthen the robustness and credibility of the study.

First, the authors should provide a more detailed description of the experimental design, particularly regarding randomization procedures, sample presentation order, control of sensory fatigue effects, and the actual number of products evaluated per participant. Given the large number of categories and products included in the study, these details are essential for readers to fully assess the validity of the comparisons performed.

I also encourage the authors to further clarify the criteria used for selecting both plant-based products and animal benchmarks. Although general selection criteria are described, the manuscript would benefit from a more explicit explanation of how “market-leading” products were defined and whether regional availability, brand representation, or commercial positioning may have influenced product selection.

Another important point concerns the statistical analysis. The use of Wilcoxon signed-rank tests appears appropriate for ordinal sensory data; however, the manuscript includes a large number of comparisons across categories, attributes, and products. The authors should therefore clarify whether any correction for multiple comparisons was applied or justify why such corrections were not considered necessary. Additionally, reporting effect sizes alongside p-values would improve the practical interpretation of the observed differences.

The discussion section is generally well developed and effectively contextualizes the findings within the broader literature. Nevertheless, some statements appear overly assertive, particularly when linking sensory performance to market share. Since these analyses are observational, I recommend adopting more cautious language and acknowledging potential confounding factors such as price, accessibility, branding, consumer familiarity, and marketing strategies.

From an editorial perspective, there are also some minor issues that should be corrected. For example, there is an unresolved cross-reference indicated as “Fig. ??” in the Results section. Additionally, some sentences could be revised to improve clarity and fluency, and there are minor inconsistencies in the reporting of p-values and significance symbols.

Despite these limitations, the manuscript presents an impressive dataset and addresses a topic of high scientific, technological, and societal relevance. With the revisions suggested above, I believe the article will provide a strong contribution to the literature on sensory acceptance, plant-based product development, and sustainable food systems.

Author Response

Overall, I find the manuscript interesting, current, and potentially valuable for researchers, product developers, and stakeholders in the alternative protein industry. The open-access approach to sharing the sensory dataset is also commendable, as it promotes transparency and reproducibility in a field where much of the available data remains proprietary. However, although the manuscript has clear scientific merit, I recommend revisions prior to acceptance, as several methodological and editorial aspects should be clarified to strengthen the robustness and credibility of the study.

Thank you for your supportive comments and thoughtful feedback. We have included all remarks in the revised version of the manuscript, highlighted all edits in red, and address them in detail in the comments below.

1. First, the authors should provide a more detailed description of the experimental design, particularly regarding randomization procedures, sample presentation order, control of sensory fatigue effects, and the actual number of products evaluated per participant. Given the large number of categories and products included in the study, these details are essential for readers to fully assess the validity of the comparisons performed.

Thank you for this important suggestion. We have revised the methods and data analysis sections to provide additional detail on the randomized within-subject design, product presentation order, sample size per participant, and palate-cleansing procedure between products.

Methods. “We conducted a large-scale, blinded, in-person sensory study to compare plant-based meat products with conventional animal benchmarks across 14 product categories between November 2024 and January 2025. To reflect real-world consumption contexts we performed all tests at restaurant partner locations in San Francisco, CA and New York City, NY. We prepared all products according to manufacturer instructions and allowed participants to add condiments when appropriate. We implemented a blinded, randomized, within-subject design, where participants evaluated one product at a time. Each participant evaluated six products of the same category under blinded conditions, five plant based and one animal based. We randomized product presentation order within each session to minimize order and sensory fatigue effects. After tasting each product, participants completed a standardized survey that captured sensory evaluations, similarity to conventional products, purchase intent, and open-ended feedback questions. Before tasting the next product, participants rinsed with water to reduce carryover effects. The full study included 2,684 participants, more than 11,000 plant-based product evaluations, and more than 800,000 data points.”

Data analysis. “[…] All comparisons reflect paired within-subject sensory assessments because participants evaluated products within the same category under blinded conditions.”

2. I also encourage the authors to further clarify the criteria used for selecting both plant-based products and animal benchmarks. Although general selection criteria are described, the manuscript would benefit from a more explicit explanation of how “market-leading” products were defined and whether regional availability, brand representation, or commercial positioning may have influenced product selection.

Thank you for this suggestion. We have now revised the product selection section to clarify that product selection reflected commercially available products with broad market presence and sensory comparability within each category:

Tested products. “[…] The study targeted commercially prominent and broadly representative categories rather than exhaustive coverage of all available plant-based products. Within these categories, we selected 121 plant-based products, 5-10 per category, mean 8.6, based on significant market presence, market-ready status, design as animal analog, original flavor comparable to the animal product, and distinct ingredients or production technologies relative to other products tested within the category. For each category, we selected a conventional animal benchmark based on highest retail sales volume. Selection emphasized representative products that reflected the commercially available market landscape at the time of testing. […]”

3. Another important point concerns the statistical analysis. The use of Wilcoxon signed-rank tests appears appropriate for ordinal sensory data; however, the manuscript includes a large number of comparisons across categories, attributes, and products. The authors should therefore clarify whether any correction for multiple comparisons was applied or justify why such corrections were not considered necessary. Additionally, reporting effect sizes alongside p-values would improve the practical interpretation of the observed differences.

This is an excellent point, and we apologize for not being clear about it in the first place. Here, because participants evaluated products only within the same category under a paired within-subject design, the analyses focused on predefined category-level comparisons rather than unrestricted global hypothesis testing. We now clarify this rationale in the Methods section and acknowledge that the reported p-values should be interpreted within the context of exploratory sensory benchmarking:

Data analysis. “All comparisons reflect paired within-subject sensory assessments because participants evaluated products within the same category under blinded conditions. The study aimed to benchmark category-level sensory performance rather than precisely rank individual formulations; accordingly, we aggregated responses across products within each category to reduce sensitivity to product-specific variability. The statistical analyses therefore focused on predefined category-level benchmark comparisons rather than unrestricted global hypothesis testing, and the reported p-values emphasize comparative sensory interpretation rather than strict confirmatory inference across all categories simultaneously.”

4. The discussion section is generally well developed and effectively contextualizes the findings within the broader literature. Nevertheless, some statements appear overly assertive, particularly when linking sensory performance to market share. Since these analyses are observational, I recommend adopting more cautious language and acknowledging potential confounding factors such as price, accessibility, branding, consumer familiarity, and marketing strategies.

Thank you for this thoughtful comment. We have now revised the Discussion to adopt a more balanced perspective on consumer adoption. Specifically, at three locations, we now clarify that sensory equivalence alone may not ensure widespread dietary transition and acknowledge additional factors including price accessibility, branding, familiarity, and culinary integration.

Discussion. “[…] Plant-based burgers continue to attract substantial scientific and commercial attention, and our results show their overall liking is almost on par that of the animal counterpart. […] These results show substantial sensory improvement remains technically achievable, although broader adoption will also depend on factors such as familiarity, availability, and socio-cultural norms. […] Adoption of sustainable proteins will depend in part on achieving robust sensory equivalence, as prior behavioral studies consistently demonstrate. However, increased awareness alone does not ensure dietary change, and long-term adoption will likely also depend on economic, cultural, and behavioral factors beyond sensory performance. Achieving this goal requires mechanistic insight rather than incremental iteration.[…]”

5. From an editorial perspective, there are also some minor issues that should be corrected. For example, there is an unresolved cross-reference indicated as “Fig. ??” in the Results section. Additionally, some sentences could be revised to improve clarity and fluency, and there are minor inconsistencies in the reporting of p-values and significance symbols.

Thank you for pointing out these inconsistencies. This was an oversight on our end, we have now correctly linked the missing figure. We have also revised sentences for clarity and fluency, and unified the notation of the p-values and significance symbols.

Despite these limitations, the manuscript presents an impressive dataset and addresses a topic of high scientific, technological, and societal relevance. With the revisions suggested above, I believe the article will provide a strong contribution to the literature on sensory acceptance, plant-based product development, and sustainable food systems.

Thank you for your encouragement. We appreciate your constructive feedback and helpful suggestions.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

I'd like to thank the authors for revising their manuscript

Author Response

Comments and Suggestions for Authors. I'd like to thank the authors for revising their manuscript  

Response.  Thank you for your positive feedback. 

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

Accept in present form

Author Response

Comments and Suggestions for Authors. Accept in present form.   Response. Thank you for your encouragement and support. 

Author Response File: Author Response.pdf

Reviewer 4 Report

Comments and Suggestions for Authors

The authors have adequately addressed most of the comments and substantially improved the manuscript. The revisions strengthen the methodological clarity and overall interpretation of the results.

However, a few minor issues still remain before final acceptance. In particular, the manuscript would benefit from:
(i) clarification regarding whether any multiple-comparison correction was formally applied;
(ii) inclusion of effect size metrics to improve interpretation of the practical relevance of the observed differences; and
(iii) a slightly more explicit description of the data source or criteria used to define “market presence” and product selection.

Therefore, I recommend minor revisions.

Author Response

Comments and Suggestions for Authors. The authors have adequately addressed most of the comments and substantially improved the manuscript. The revisions strengthen the methodological clarity and overall interpretation of the results.

However, a few minor issues still remain before final acceptance. In particular, the manuscript would benefit from:

(i) clarification regarding whether any multiple-comparison correction was formally applied;

Response. Thank you for this comment. The objective of this study was category-level sensory benchmarking rather than confirmatory hypothesis testing across all product categories simultaneously. Accordingly, we did not apply a formal multiple-comparison correction such as Bonferroni adjustment. Instead, we interpreted p-values together with effect magnitude, consistency across categories, and practical sensory relevance. To clarify this point, we added a statement to the Methods section. We further added a new table reporting Wilcoxon effect sizes for all 14 category-level comparisons.

Data analysis. “All comparisons reflect paired within-subject sensory assessments because participants evaluated products within the same category under blinded conditions. The study aimed to benchmark category-level sensory performance rather than precisely rank individual formulations; accordingly, we aggregated responses across products within each category to reduce sensitivity to product-specific variability. The statistical analyses focused on predefined category-level benchmark comparisons rather than testing a single global hypothesis across all categories. Because the objective was comparative benchmarking, not confirmatory inference, we did not apply a formal multiple-comparison correction. Accordingly, p-values should be interpreted in conjunction with effect magnitude, consistency across categories, and practical sensory relevance rather than as standalone measures of significance.”

(ii) inclusion of effect size metrics to improve interpretation of the practical relevance of the observed differences; and

Response. Thank you for this suggestion. We agree that effect sizes provide important context beyond statistical significance. We therefore calculated Wilcoxon effect sizes r = Z/√n for all 14 category-level comparisons and added a new table reporting sample size, mean differences, p-values, and effect sizes in the new Table 1.

Results.“[…] Effect size analysis across approximately 100 paired comparisons per category (n = 98.6 +/- 8.6, total n = 1,381) revealed negligible to small effects for unbreaded chicken filet (r = 0.05), chicken nuggets (r = 0.07), and burgers (r = 0.19), but large effects for steak (r = 0.62) and bacon (r = 0.76). These effect sizes support the conclusion that some categories are approaching sensory parity while others are not (Table 1).”

(iii) a slightly more explicit description of the data source or criteria used to define “market presence” and product selection.

Response. Thank you for this suggestion. We have expanded the product selection description in Section 2.1 to clarify how market presence was defined. Specifically, we now state that product selection was informed by publicly available retail sales rankings, distribution across major U.S. retailers and food-service channels, and NECTAR industry category analyses, with the goal of capturing the most commercially relevant products available at the time of testing.

Tested products. “[…]  The study targeted commercially prominent and broadly representative categories rather than exhaustive coverage of all available plant-based products. Within these categories, we selected 121 plant-based products, 5-10 per category (mean 8.6), based on significant market presence, market-ready status, design as an animal analog, original flavor comparable to the animal product, and distinct ingredients or production technologies relative to other products tested within the category. We determined market presence using publicly available retail sales rankings, distribution across major U.S. retailers and food-service channels, and industry category analyses from NECTAR. For each category, we selected a conventional animal benchmark based on highest retail sales volume. These criteria ensured that both plant-based and animal products reflected the commercially relevant market landscape at the time of testing. […]”

 

Author Response File: Author Response.pdf