兽研1. 引言
预计到2031年，全球猫粮市场规模将超过920亿美元，反映出市场对营养全面且能被猫持续接受的产品需求日益增长[1]。适口性是决定自主采食量的首要因素，因此被广泛认为是宠物食品商业成功的关键因素[2]。对于猫这种具有特定营养需求且感官辨别力强的专性食肉动物而言，适口性至关重要。即使是营养全面的日粮，若适口性不足也可能被拒食，从而降低自主采食量并导致不良健康后果[3,4]。因此，宠物食品行业投入大量资源用于适口性测试和产品优化[5]。
猫适口性的研究可追溯至20世纪中叶[6]，随后在20世纪90年代的研究确立了猫作为具有特化味觉受体的专性食肉动物的地位[7,8]。尽管这些基础概念数十年来一直指导着猫食品的测试与配方设计，其后也出现了若干关于感官驱动因素[3,9]、评估方法[2,10]和食品配方[11]的综合性综述，但关于导致猫适口性研究结果不一致的方法学和概念性因素，仍缺乏系统梳理。这些不一致的核心在于缺乏普遍接受的适口性定义。根据不同研究，适口性被描述为食物的感官特性、动物的自主采食量或与进食相关的享乐体验。这种概念上的分歧伴随着整个文献中更广泛的不一致。一项研究中被视为食物接受证据的行为指标，在另一项研究中被解读为厌恶指标，而关于采食量季节性变化的综述往往相互矛盾。这些不一致的影响超出了学术研究范畴。适口性定义、测量和解读方式的差异使得研究间难以比较，难以识别猫食物偏好的感官和行为驱动因素，也难以选择适当的适口性评估方法。解决这些问题将提高研究的可比性，加强循证产品开发，并有助于将真正的猫偏好与主人感知区分开来。
因此，本综述将探讨猫适口性研究中四个尚未解决的关键矛盾领域：（i）适口性定义的不一致及整合框架的构建。（ii）行为指标的矛盾及标准化行为谱的需求。（iii）方法学的碎片化及评估方法的决策框架。（iv）我们称之为“主人感知悖论”的问题——将宠物适口性与主人购买意图区分开来。针对每个领域，我们识别具体空白、回顾示例并提出未来研究的具体方向。
2. 材料与方法
采用Scopus数据库作为主要检索系统，以识别猫适口性文献中的不一致之处[12]。选择Scopus作为主要文献数据库，是因为它是最大的同行评审文献摘要和引文数据库之一，收录来自7000多家出版商的超过1亿条记录[12,13]。其对动物科学、兽医学、农业和食品科学的广泛覆盖，使其非常适合识别猫适口性这一多学科领域的相关文献。
检索在文章标题、摘要和关键词范围内进行。以“palatability”为术语的初步检索获得12,635条英文记录；然而，由于本综述的主要焦点是猫食品适口性，检索词被细化为“palatability” AND（“cat” OR “feline”），得到215条记录，其中200条以英文发表。对检索到的记录标题和摘要进行相关性筛选，排除以下研究：“CAT”指代无关缩略词、仅简要提及适口性或仅关注药物产品接受性的研究。随后，保留141篇出版物并进行批判性审阅，以识别反复出现的主题、不一致和方法学挑战。此后，通过解释性综合将类似的不一致进行归组，形成构成本综述结构的四个总体主题领域。在批判性审阅的141篇出版物中，最终有52篇在综述中被引用，作为直接支持或说明所讨论主题和问题的文献。
本批判性叙述综述专门考察了猫文献中“palatability”一词的使用和解读；因此，检索策略以该术语为中心。 consequently，使用替代术语（如偏好、接受度或采食动机）的相关研究可能在初步检索中被遗漏。为尽量减少这一局限性，通过定向检索和引文追踪识别了额外出版物。本综述共引用97篇出版物，包括通过初步Scopus检索识别的52篇和通过定向检索及引文追踪识别的45篇额外出版物。
3. 定义问题
“适口性”一词最初用于描述人的行为及进食欲望，随后被引入实验营养学，尤其是动物研究。Hill等[14]将适口性定义为“看到高度偏好的食物会显著增加受试者自评的进食欲望”。后续研究表明，适口性更为复杂，不能仅用欲望来定义。Kissileff[15]提出将“内在适口性”（食物的感官特性）与“习得性适口性”（先前经验和摄入后效应）区分开来。这是一个重要的转变：不再仅将适口性视为食物本身的属性，还考虑了学习情境和摄食个体的特征。大约同一时期，Rogers[16]进一步完善了这一概念，揭示了享受味道与享受进食之间的区别。因此，适口性是一种享乐性（基于愉悦的）体验，而不仅仅是感官体验。尽管有这些改进，Yeomans等[17]仍指出“关于适口性的本质缺乏共识”。这种分歧不仅仅是语言层面的问题。适口性的定义方式直接影响其测量和解释方式。如果将适口性视为一种生理需求，则会强调食物摄入量和动机。相反，如果将其视为一种享乐过程，则偏好或奖赏相关行为更为相关。因此，研究可能使用同一术语却评估根本不同的构念，使得跨学科和跨物种的比较变得困难[17]。
同样的共识缺乏在宠物营养学中也很明显[18]。宠物食品适口性的历史定义侧重于观察到的行为，如摄入量或接受度[2,18,19]。然而，这带来了一个根本性问题；尽管高适口性食物通常被摄入更多，但摄入量也受到与适口性无关的因素影响，包括饥饿和饱腹感[20,21,22]、健康状况[23,24,25]、绝育状态[26]、饲喂环境[27,28,29]、食物可获得性[30,31,32]以及既往膳食经验[33,34]。此后，该定义通过强调接受度还取决于动物对食物的感官感知而得到扩展[35,36,37]。然而，一些作者[11,28,38,39,40]仅关注食物的感官特性来定义适口性。直到2015年，宠物食品适口性的定义主要集中于感官和消费属性。较近期的文献表明向整合性享乐概念的转变，例如从愉悦感知、喜爱和食物吸引力的角度来定义适口性[3,9,10,41]。Watson等[9]特别指出适口性与愉悦而非生理需求相关，这与早期人类适口性研究相呼应。Rosita等[42]进一步支持了这一点，他们仅将适口性定义为“动物喜爱食物的程度”。同样值得澄清的是适口性不是什么：它不是单一的感官属性，也不是仅由风味决定的简单化测量[18]。因此，关于适口性应被视为食物的感官特性、行为结果、享乐反应还是这些因素的组合，仍存在不确定性，如图1所示。
与其将这些观点视为相互竞争的定义，不如将其视为更广泛构念的互补组成部分。在此框架内，感官维度指对食物特征（如味道、香气、质地和外观）的感知；享乐维度指这些感官感知所引发的愉悦、喜爱或吸引力；消费维度指可观察的自愿接受、偏好和摄入行为，适口性正是通过这些行为被推断出来的。重要的是，消费不应被视为适口性的同义词，因为它还受到与食物享乐价值无关的生理[20,21,22,23,24,25,26]、环境[27,28,29]和经验因素[33,34,38]的影响。据此，适口性可被概念化为一个整合过程，即食物的感官特性被动物感知并进行享乐评价，进而影响随后的接受度、偏好和消费。这种方向性关系如图2所示。
4. 矛盾的行为指标
自1979年以来，研究者一直使用猫的行为指标来评估其反应[43]，近期更将其与摄食数据结合以全面了解适口性[9,10]。文献已识别出一组行为指标（摄食前、摄食中和摄食后），可大致分为与所提供食物适口性相关的正向或负向指标。正向指标包括眼睛睁开程度低于50%的持续时间增加[4]、舔唇增加[4,21,39]、咂嘴更频繁[4]以及伸舌更频繁[4]。负向指标包括耳朵向后甩动或尾巴甩动[39]、伸舌打哈欠更频繁[4]，以及刨地、摇头和肢体挥舞组合行为的非显著趋势[44]。某些行为指标，如姿势（坐或站）[21,22]、背部皮肤抽动[39]和舔爪[21,44]，似乎与日粮适口性无关，而玩耍食物可能表示饱腹感而非适口性[45]。这种传统分类法促进了猫食物摄入量与行为反应之间关系的示意图概述，例如Pekel等人[3]所展示的图（图5）。虽然这些总结提供了有价值的概述，但不可避免地简化了更复杂的文献体系。进一步审视发现，若干行为指标无法始终如一地归类为纯正向或负向，且此类展示中忽略了若干争议和细节。例如，面部理毛被视为摄入适口食物后的正向指标[21]，而身体理毛——常在摄入适口性较差的食物后或应激期后出现——被视为负向指标[39]。因此，将“理毛”归类为单一行为指标可能掩盖这些重要区别，导致过度简化的解读。
除此类简化外，若干行为指标在不同研究间似乎相互矛盾。Hanson等人[4]将更频繁的舔鼻与愉悦的味觉体验相关联；Savolainen等人[39]和van den Bos等人[21]均将其与对负向或适口性较差味觉体验的反应相关联；Bennett等人[43]将其解读为猫的恐惧反应——对同一行为的三种根本不同的解释。嗅闻食物也呈现类似模式。Van den Bos等人[21]和Becques等人[22]将嗅闻增加解读为对适口性较差日粮的采食犹豫；Owens等人[44]发现日粮对总嗅闻时间无影响，接受测试期间的嗅闻与随后的偏好-摄入比呈负相关，但偏好测试期间的嗅闻与摄入比无相关性；Kim等人[46]注意到两种适口性相同的日粮在首次嗅闻回合中存在显著差异。采食潜伏期也表现出不一致的解读。Van den Bos等人[21]和Owens等人[44]发现快速接近食物表明高适口性，而Becquest等人[22]仅注意到猫最终接近食碗时的显著差异，此时摄入量已无显著差异。因此，表1通过区分始终与适口性相关的行为与具有矛盾或情境依赖性解读的行为，对既往总结进行了扩展。
若干因素导致了这些相互矛盾的发现。行为相似性使得面部表情检测的准确编码变得困难。Becques等[22]和Owens等[44]都明确承认了区分舔鼻和舔唇的困难。然而，这两种行为似乎具有相反的适口性解读，舔唇被视为积极指标[4,21,39]，而舔鼻则更普遍地被接受为消极指标[21,39,43]。如果观察者无法可靠地区分二者，分类就会受到影响。若干研究使用了根本不同的测试形式（序贯单碗测试[21]、双碗偏好测试[4]和五碗偏好测试[46]），以及不同的暴露时长（5–30分钟[4,21,44,46]、20小时[22]或直至吃完[39]），随后对同一行为得出了不同的解读。Owens等[44]观察到，一些猫始终更频繁、更长时间地嗅闻右侧的碗而非左侧的碗，提示侧偏好可能影响嗅闻行为。这一问题在多碗设计[4,44,46]中更为复杂，因为嗅闻等行为可能对应任何一种特定日粮。这种方法学上的碎片化也体现在行为的评估方式上，如“舔/嗅食物”[21]、“嗅闻/探究”[47]、“嗅闻时长”[22]、“嗅闻回合”[44]和“首次嗅闻回合”[46]。因此，矛盾不仅仅在于解读；这些研究在不同条件下测量了不同的指标（或指标组合）。同样，进食速度也显示出相互矛盾的解读。Owens等[44]基于采食量和总进食时间客观量化了进食速度（g/min），而Knight和Satchell[47]则依赖主人填写的问卷，其中进食速度未经标准化或客观测量，使结果容易受到解读偏倚的影响。个体变异性进一步使解读复杂化，并可能影响进食模式[48]、嗅闻行为[44]和饥饿反应[21]。因此，样本量较小的研究[44]比样本量较大的研究[21,22,39,46]更容易受到个体效应的影响。除设计局限性外，当前文献还存在若干空白。尚无研究直接比较性别对食物适口性行为指标的影响，尽管有证据表明在其他情境中存在性别特异性的行为模式，例如与公猫个性相关的重复嗅闻行为在母猫中不存在[49]，以及特定环境条件（如气候季节）下食物偏好的性别相关差异[50]。这些发现提示，性别可能影响食物感官特性的感知方式或适口性的行为表达方式。例如，母猫可能对挥发性线索更敏感[50]；然而，这一假设需要进一步研究。因此，未来研究应分别测试公猫和母猫，并在多个行为指标上检验性别与适口性的交互作用。另一个空白是不同日粮类型（如生食、湿粮或干粮）的影响[10,42]。日粮的物理特性可能影响适口性行为指标的表达方式。例如，舔唇与较高的适口性相关，主要见于涉及湿粮或其他潮湿食物的研究[4,21,39]；然而，目前尚不清楚这一反应是仅反映享乐评价，还是也可能受到水分和食物在口腔周围附着等物理特性的影响。因此，未来研究应在控制适口性差异的同时，直接比较对湿粮和干粮的行为反应。同样，大多数研究比较的是预期适口性不同的日粮，但文献尚未探讨当适口性差异增大时行为反应是否会变得更加明显。应当注意，没有哪种行为指标或面部表情是绝对积极或消极的，频率和持续时间起着重要作用[24,37]。例如，耳朵扁平化被认为是消极反应，在89%的消极情境中观察到，但在55%的积极情境中也被观察到[37]。同样，某些行为指标（如嗅闻）无论有无食物都会发生[51]。这意味着仅记录行为的存在或缺失[39]是不够的；频率和持续时间的定量测量至关重要，然而一些研究[22]仅报告其中之一，且尚无研究建立阈值。
这些矛盾有一个直接后果：按照目前文献的呈现方式，行为指标不能作为适口性的独立衡量标准来解读。然而，一些论文利用这些行为指标从调查数据中判定食物适口性。例如，Knight和Satchell[47]利用现有文献[4,21,22,39]将“快速接近食物”和“快速进食”归类为适口性的正向指标，将“嗅闻/探查食物”和“掉落食物”归类为负向指标。基于宠物主人对这些行为的报告，他们得出结论认为纯素饮食和肉基饮食的适口性相似。然而，用于支持这些分类的文献并不一致：Van den Bos等[21]和Becques等[22]关于快速接近食物的研究结果相互矛盾；Hanson等[4]和Becques等[22]关于嗅闻的结论相反；Savolainen等[39]指出，掉落食物不能被视为独立指标，因为它只在猫已经进食或正在进食时才会发生，这意味着其在适口性好的食物条件下频率更高是摄入量更大的结果，而非适口性的独立衡量。相同的行为指标可能是正向的、负向的，或没有有意义的反应，取决于研究背景。因此，行为数据应与摄入量数据结合解读。例如，猫在进食大量食物后舔嘴唇可能表明适口性良好，而猫在拒食后舔嘴唇则不然。此外，舔嘴唇也可能仅仅是清理食用特别 messy 的食物后残留物的结果。脱离背景，仅凭行为本身是模糊的。正如感官词汇表为描述宠物食品的感官特性提供了标准化词汇[52,53]，该领域也将受益于一个标准化的猫进食行为行为谱。它应定义需要编码的行为、这些行为具有信息性的条件、频率和持续时间的定量阈值，以及每种指标已经过验证的食物形式。标准化这些标准将提高研究之间的一致性，减少相互矛盾的解释，并为验证猫食物适口性的行为指标提供更坚实的基础。
5. 方法学碎片化
文献中的许多矛盾无法通过重新审视单项研究来解决，因为根本原因是方法学不匹配。虽然现有文献已经对可用的适口性测试方法提供了清晰的描述[2,9,10,28]，但一些研究[25,36]仍然采用与其既定目标不一致的评估方法。其中一种方法学不匹配是互换使用接受性测试和偏好性测试。例如，Ilias等[36]旨在确定处方粮的“适口性和可接受性”以“确保DSH猫持续进食”。这本质上是一个接受性问题，然而却使用双碗偏好性测试来比较治疗性日粮[36]。在另一个案例中，Bernachon等[25]研究了“五种补充剂的比较适口性”，暗示目标是偏好性或比较排名；然而，每只猫只接受了分配给它的产品，这意味着该研究测量的是不同组内的接受性，而非产品之间的直接偏好性[25]。偏好性测试的优势在于区分两种差异细微的产品，而非确定单一产品是否达到最低摄入量阈值，后者更适合通过接受性测试来解决[10,54]。这一区别得到了Hours等[55]的支持，他们报告称自愿采食量测量和双碗偏好性测试结果可能对相同日粮得出不同的结论。鉴于这两种测试方法解决的是根本不同的问题，令人惊讶的是只有少数已发表研究依次进行了两种测试以进行更全面的评估[24,55,56]。同样，喜好性测试[57]——理论上处于接受性测试和偏好性测试之间的理想位置——也未被研究界采用，仅在综述论文中被提及[10,37]。与并行时间表范式[58,59]和并行配对联想偏好[60,61]不同——后者采用有限可归因于实际限制——喜好性测试在文献中的缺失更难解释。它不需要专门设备，使用标准的单碗形式，并且专门设计为可由专家小组实施[57]。其缺失代表了一个独立的验证空白。
文献中部分方法学碎片化是由资源限制驱动的。宠物食品生产商和学术实验室可能缺乏时间、动物、基础设施和/或资金来运行多种测试形式以确保完整的适口性图景[28]。认识到这一现实，图3提出了一个决策框架，以主要研究目标为指导进行方法选择，分支问题代表适用于每种方法的研究问题。这使研究人员能够为其特定问题确定单一最具信息性的测试，而不是默认遵循惯例。
导致适口性研究结果相互矛盾的另一个方法学因素是样本量和试验持续时间的选择。根据Tobie等[10]的研究，专家小组偏好试验至少需要30只猫以确保统计学稳健性，而家庭测试则需要100只。然而，更近期的研究表明，理想样本量取决于统计框架和测试方法，因为猫和食物特征、适应或训练程度以及测试环境等因素可能影响变异性，进而影响所需动物数量[62,63]。Pires等[62]使用功效分析证明，23只猫的偏好试验足以在75%功效下检测摄入比率0.10的差异，而Bos等[63]使用基于精度的方法证明，在双碗偏好试验中，根据所测试干性膨化饲料之间的对比度，17–23只的样本量足以达到10%的误差范围（MoE）。对于接受性试验，Bos等[63]指出变异性更大，需要12–43只猫才能达到相同的MoE。这是两种根本不同的方法。统计功效是拒绝错误零假设的概率[64]；因此，功效分析可用于确定研究是否具有足够功效来检测特定效应，而MoE用于确定平均摄入量估计的精确度[65]。目前，尚无研究为专家小组单碗接受性试验提供经统计学推导的样本量。Bos等[63]指出“较短的研究周期配合较多的猫只数量更优”，因为延长测试时间可能导致脱漏，降低准确性。Pires等[62]和Bos等[63]均表明单个测试日可能已足够；然而，这应与预测试适应或熟悉化的需求相区分，当动物不习惯测试程序时可能需要进行预适应。尽管如此，许多研究[36,66,67,68]仍继续使用延长方案——5–10天，小组规模为10只或更少。此类设计容纳个体间变异性的能力可能有限，而延长测试还可能增加脱漏相关偏倚的风险。相比之下，行业实践和若干学术研究[40,69,70,71]通常使用20只猫的小组测试2天，更接近Bos等[63]的建议。报告样本量依据和统计框架以及研究结果的重要性常被忽视。Pires等[62]强调了这一点，他们发现，在其系统综述纳入的16项研究中，仅3项报告了足够信息以允许对猫粮偏好试验的统计功效进行全面评估。这种透明度的缺乏可能导致了研究之间的矛盾，因此建议根据预期摄入比率差异、预期变异性、显著性水平和期望统计功效，通过功效分析确定适当的样本量[62]，而非采用固定的猫只数量。
除样本量和测试持续时间外，更深层次的方法学碎片化可能源于 claims 在二手来源间的传播而未经原始证据验证。猫食物摄入的季节性变化提供了一个清晰的例子。Watson等[9]指出猫倾向于“冬季比夏季吃得少”，引用Tobie等[10]，后者同样指出猫“冬季吃得较少”。然而，Tobie等[10]引用Serisier等[27]作为 underlying 原始证据，而Serisier等[27]实际报告了相反的模式，冬季食物摄入增加。这一方向也与Alegría-Morán等[50]的发现一致，他们报告在较冷条件下食物摄入增加。因此，此处说明的问题不仅仅是研究之间的分歧，而是错误报告的效应方向从二手来源传播到后续文献中。另一个例子包括Watson等[9]的陈述“猫被高蛋白和高脂肪含量的食物驱动，并避免富含碳水化合物的食物”，引用Hewson-Hughes[72]和Salaun等[73]，但未说明Hall等[74]所确定的高碳水化合物目标是在适口性相等的条件下专门观察到的。这通过创建不一致的证据基础导致了方法学碎片化，使得难以比较研究、准确解释发现和开发稳健的实验设计。
6. 主人感知悖论
猫适口性的核心矛盾在于，选择和购买宠物食品的宠物主人并非最终消费者。宠物食品被视为在宠物日常生活中创造积极体验的便捷机会[4]。宠物主人将宠物愿意进食解读为满意的即时指标[75]，当宠物拒绝食物时，主人会觉得自己没有提供充分的照顾，因而寻求更好的替代品[2,76]。然而，当宠物既没有直接拒绝食物，也没有觉得食物达到最大适口性时，会发生什么？猫会在可接受的范围内下调其摄入量，而主人则假定适口性足够，并不知道猫对偏好的替代品会显著摄入更多。这就是我们所称的“主人感知悖论”；主人的感官偏好、认知偏差和情感依恋凌驾于猫偏好的生物学现实之上。
主人和猫通过不同的感觉通道感知食物。宠物主人通常仅根据外观和香气来评价食物[36,77,78]，而猫则评价完整的适口性体验，包括香气、味道和质地[38,39,79,80]。生理差异是巨大的。猫的嗅觉上皮面积为21 cm²，而人类为3–4 cm²[36]，其嗅觉灵敏度大约是人类十四倍[81]，仅对氧化物的检测就比人类敏锐一百倍[82]。犁鼻器进一步使它们能够检测人类鼻子无法察觉的挥发性化合物[82,83]。猫拥有约470个味蕾，远少于人类（约10,000个），提示味觉敏感性较低[3]。然而，这种印象具有误导性，因为猫对某些苦味化合物具有增强的敏感性[4,29,37]，它们的鲜味受体不被相同的氨基酸激活[84]，并且它们完全缺乏甜味的功能性受体[18,85]，这意味着宠物食品营销中常吸引人类注意的甜味香气对猫的味觉来说是不可见的。因此，主人为猫选择饮食时所使用的感官属性与猫消费食物时所使用的感官属性从根本上不同。
Ilias 等人[36]通过比较人类感官评定小组的气味评分与猫小组在四种处方日粮下的采食结果，证明了这种差异。两个物种都认为低敏日粮的偏好度最低，但此后它们的区分能力有所不同。猫根据采食量对剩余三种日粮给出了清晰的排序（尽管这一排序在统计学上不显著），而人类气味评定小组则将剩余三种日粮评为大致相当[36]。同样，至少自20世纪90年代初以来，宠物主人调查一直是從主人视角评估宠物食品适口性的广泛使用方法[86,87,88,89,90]，但由于偏差[9,47]、判断不一致[86]以及对宠物行为解读的差异[91]，其解释力也有限。这在 Sanderson 等人[86]开展的研究中显而易见：主人在不知情的情况下连续几个阶段被给予相同的日粮喂养其犬，27%的主人给出了不一致的回答。虽然该研究涉及的是犬而非猫，但其基本原理仍然适用。同样，Bos 等人[63]发现，主人对其猫挑食程度的分类与猫的采食量并不对应。被主人归类为“容易”进食的猫，其总体采食量反而低于“困难”或“非常困难”的猫。然而，主人并非对猫的偏好不确定，而是主动将自己的偏好投射到猫身上。Di Donfrancesco 等人[92]发现，主人对产品的喜爱程度与其预测宠物喜爱程度之间的相关系数为0.93。然而，驱动主人喜爱的因素（外观、颜色和香气）并非驱动宠物采食的因素（香气、味道和质地）。这种投射因宠物人性化日益增强的影响而进一步强化[93,94,95]。在对1380名猫主人的调查中，Mace 等人[96]指出，影响购买替代日粮意愿的最重要特征是良好的健康结果，且受限于主人的知识水平。因此，他们的购买行为反映的是自身的价值观（可持续性、有机成分或人类级品质），而非猫的营养和感官需求[93,96,97]。这导致了具体且可证明的错位。主人经常偏好以天然或不含添加剂为卖点的日粮[96,98]，却不知道去除某些成分（如糖类）会损害美拉德反应驱动的风味生成[99,100]。同样，主人偏好蛋白质和鲜肉含量较高的日粮[75]，却不知道块状配肉汁日粮中的肉汁部分才是驱动适口性的关键[101]。Erzurum 和 Kayar[75]的一项调查捕捉到了这一矛盾：宠物主人将“宠物偏好（适口性）”列为最高的选择标准，而将“视觉外观”列为最低。主人相信自己是根据猫的偏好来选择的。然而，正如 Di Donfrancesco 等人[92]所证明的，主人认为其猫偏好的内容与主人个人偏好的内容无法区分。其后果在学术文献中也有所体现。Knight 和 Satchell[47]调查了1135名猫主人，发现日粮类型（常规、生食和纯素）对食物导向行为的差异很小，结论是纯素宠物食品与常规肉类或生肉日粮同样适口。这一发现被后续综述者描述为“极其主观”[9]，并导致了“模糊的结果”[11]。主人报告不仅仅引入噪声；它还可能主动抹平真实的适口性差异。重要的是，主人感知悖论在不同喂养经验或文化背景的主人之间不太可能是一致的。知识更丰富的主人可能对产品属性的权重不同[97]，而积极监测采食量的看护者可能更有能力识别接受度和偏好[24]。文化差异进一步塑造了对宠物食品气味的反应[78]以及对天然性、新鲜度、可持续性、适口性、品牌和兽医建议等属性的重视程度[28,75,96]。然而，人口统计学效应在不同人群中并不一致；例如，在一个土耳其样本中，年龄和教育程度与购买行为无关[75]，而在一个以英国和欧洲为主的调查中，年长的看护者受价格和便利性的影响较小[96]。因此，主人感知不应被视为单一、稳定的偏差来源。一个重要的研究空白是确定在不同文化和喂养背景下，哪些主人特征会增加或减少感知适口性与客观测量的猫适口性之间的一致性。
解决这一悖论的方法并非完全排除主人的解读，因为主人仍然是最终决策者。相反，而是要认识到主人评估和猫采食量测量的是根本不同的构念。主人调查捕捉的是购买意图和主人满意度；采食测试捕捉的是猫的享乐反应。这些是互补的测量，而非可互换的。不应假定主人感知能反映宠物适口性，也不应假定宠物适口性必然带来商业成功。成功的宠物食品产品取决于同时满足这两个群体，而他们的感官通道、认知框架和评价标准并不一致。
7. 局限性
本综述存在若干局限性。首先，叙述性综述涉及对不同研究的证据进行综合，这意味着主题的识别与分组包含作者一定程度的主观解读。其次，主要检索仅限于Scopus数据库和英文出版物，尽管通过针对性检索和引文追踪发现了额外研究。最后，本综述纳入了异质性的证据体，包括实验性、方法学和综述类出版物，这可能限制研究之间的直接比较。
8. 结论与未来研究方向
尽管近几十年来取得了相当大的进展，猫科动物食品适口性研究仍受方法学和概念不一致性的阻碍。本批判性叙述综述表明，概念定义、实验执行和行为指标解读的可变性限制了评估猫科动物食品满意度可靠框架的建立。
因此，未来研究应从描述性方法转向规范性方法，提供方法学和解读指导，而不仅仅是识别或总结研究发现。本综述提出的决策树可能有助于使试验设计与特定研究目标相一致，而常规报告统计功效和/或精确度将提高适口性声明的稳健性和可重复性。同样，将主人印象与猫科动物反应分开评估，并采用标准化的行为指标行为谱，将消除歧义并提高研究之间的可比性。
未来研究优先方向可结构化为三个领域。基础研究应侧重于评估重测信度、评估食品形态（如干粮、湿粮和半湿粮）和动物特征（如年龄、性别、绝育状态和健康状况）的影响，以及比较主人感知与客观测量的猫科动物反应。方法学标准化应优先验证标准化的猫科动物进食行为谱、评估所提出的决策框架是否改善针对不同研究目标的适口性方法学选择，以及制定猫科动物适口性研究的最低报告建议。最后，工业应用应评估这些方法在不同产品、测试设施和商业环境中的适用性。通过协调研究结果的报告和解读方式，行业最终能够将孤立的观察转化为连贯且可操作的证据基础。
查看英文原文▾
1. Introduction
The global cat food market is projected to exceed USD 92 billion in 2031, reflecting the growing demand for products that are both nutritionally complete and consistently accepted by cats [1]. Palatability is a primary determinant of voluntary intake and is, therefore, widely recognized as a critical factor in the commercial success of pet food products [2]. For cats, obligate carnivores with specific dietary requirements and strong sensory discrimination, palatability is fundamental. Even nutritionally complete diets may be rejected if they are insufficiently palatable, reducing voluntary intake and contributing to adverse health consequences [3,4]. As a result, the pet food industry allocates substantial resources to palatability testing and product optimization [5].
The study of feline palatability dates back to the mid-20th century [6], with subsequent work in the 1990s establishing the cat’s role as an obligate carnivore with specialized taste receptors [7,8]. While these foundational concepts have guided feline food testing and formulation for decades, and were followed by several comprehensive reviews of sensory drivers [3,9], assessment methods [2,10], and food formulation [11], there remains a lack of synthesis regarding the methodological and conceptual factors that contribute to inconsistent findings across feline palatability research. At the core of these inconsistencies is the absence of a universally accepted definition of palatability. Depending on the study, it is described as the sensory properties of the food, the animal’s voluntary food intake, or the hedonic experience associated with consumption. This lack of conceptual agreement is accompanied by broader inconsistencies throughout the literature. Behavioral indicators regarded as evidence of food acceptance in one study are interpreted as indicators of aversion in another, while reviews of seasonal variation in food intake are often contradictory. These inconsistencies have implications that extend beyond academic research. Differences in how palatability is defined, measured, and interpreted make it difficult to compare studies, identify the sensory and behavioral drivers of feline food preference, and select appropriate methodologies for evaluating palatability. Resolving these issues would improve study comparability, strengthen evidence-based product development, and help distinguish genuine feline preference from owner perception.
Therefore, this review will consider four key areas of unresolved tension in feline palatability research: (i) The inconsistent definition of palatability and the development of an integrative framework. (ii) Contradictions in behavioral indicators and the need for a standardized ethogram. (iii) Methodological fragmentation and a decision framework for assessment methodologies. (iv) What we term the “owner perception paradox”—separating pet palatability from owner purchasing intent. For each area, we identify the specific gap, review examples, and propose concrete directions for future research.
2. Materials and Methods
The Scopus database was used as the principal search system to identify inconsistencies within the feline palatability literature [12]. Scopus was selected as the primary bibliographic database as it is one of the largest abstract and citation databases of the peer-reviewed literature, containing over 100 million records from more than 7000 publishers [12,13]. Its broad coverage of animal science, veterinary medicine, agriculture, and food science makes it well-suited for identifying the relevant literature in the multidisciplinary field of feline palatability.
Searches were conducted within article titles, abstracts, and keywords. An initial search of the term “palatability” yielded 12,635 English-language records; however, as the main focus of this review is feline food palatability, the search terms were refined to “palatability” AND (“cat” OR “feline”), resulting in 215 records, of which 200 were published in English. The titles and abstracts of the retrieved records were screened for relevance, and studies were excluded where the term “CAT” referred to an unrelated acronym, palatability was only briefly mentioned, or the research focused solely on the acceptance of medicinal products. Subsequently, 141 publications were retained and critically reviewed to identify recurring themes, inconsistencies, and methodological challenges. Thereafter, similar inconsistencies were grouped through an interpretive synthesis, resulting in four overarching thematic areas that form the structure of this review. Of the 141 publications critically reviewed, 52 were ultimately cited within the review as directly informing or illustrating the themes and issues discussed.
This critical narrative review specifically examined the use and interpretation of the term “palatability” in the feline literature; therefore, the search strategy was centered on this term. Consequently, relevant studies using alternative terminology (e.g., preference, acceptance, or feeding motivation) were potentially overlooked during the primary search. To minimize this limitation, additional publications were identified through targeted searches and citation tracking. In total, 97 publications were cited in this review, comprising 52 publications identified through the primary Scopus search and 45 additional publications identified through targeted searches and citation tracking.
3. The Definitional Problem
The term palatability was originally used to describe a person’s behavior and desire to eat, and was subsequently incorporated into experimental nutrition, particularly animal studies. Hill et al. [14] defined palatability as “the sight of highly preferred food markedly increased subjects’ rated desire to eat”. Subsequent work revealed that palatability is more complex and cannot be defined by desire alone. Kissileff [15] proposed to separate “intrinsic palatability” (the sensory properties of the food) from “learned palatability” (prior experiences and post-ingestive consequences). This is an important shift: instead of viewing palatability solely as a property of the food itself, it also considers the learning context and the attributes of the individual consuming it. Around the same time, Rogers [16] further refined the concept, showing the difference between enjoying the taste and enjoying eating. Thus, palatability is a hedonic (pleasure-based) experience, not just sensory. Despite these improvements, Yeomans et al. [17] noted a “lack of consensus on the nature of palatability”. This lack of agreement is more than a linguistic issue. The way palatability is defined directly influences how it is measured and interpreted. If palatability is considered a biological need, food intake and motivation will be emphasized. Conversely, if it is viewed as a hedonic process, then preference- or reward-related behaviors are more relevant. As a result, studies may use the same term while evaluating fundamentally different constructs, making comparisons across disciplines and species difficult [17].
The same lack of consensus is evident in pet nutrition [18]. Historical definitions of pet food palatability focused on observed behavior, such as intake or acceptance [2,18,19]. However, this poses a fundamental problem; although highly palatable foods are often consumed in greater quantities, intake is also influenced by factors independent of palatability, including hunger and satiety [20,21,22], health status [23,24,25], neuter status [26], feeding environment [27,28,29], food availability [30,31,32], and prior dietary experiences [33,34]. This definition has since been extended by emphasizing that acceptance is also dependent on the animal’s sensory perception of the food [35,36,37]. Yet, some authors [11,28,38,39,40] solely focus on the sensory properties of the food to define palatability. Until 2015, the definition of pet food palatability was predominantly focused on sensory and consumption attributes. The more recent literature indicates a shift to integrated hedonic concepts, such as defining palatability in terms of pleasure perception, liking, and attractiveness of the food [3,9,10,41]. Watson et al. [9] specifically indicate that palatability is linked to pleasure instead of physiological need, echoing earlier research in human palatability. This is further supported by Rosita et al. [42], who define palatability only as “the degree to which an animal likes food”. It is also worth clarifying what palatability is not: it is not a single sensory attribute, nor a simplistic measure determined by flavor alone [18]. Consequently, there remains uncertainty regarding whether palatability should be regarded as a sensory property of the food, a behavioral outcome, a hedonic response, or a combination of these factors, as seen in Figure 1.
Rather than treating these perspectives as competing definitions, they may be viewed as complementary components of a broader construct. Within this framework, the sensory dimension refers to the perception of food characteristics such as taste, aroma, texture, and appearance; the hedonic dimension refers to the pleasure, liking, or attractiveness elicited by these sensory perceptions; and the consumption dimension refers to observable voluntary acceptance, preference, and intake behaviors through which palatability is inferred. Importantly, consumption should not be regarded as synonymous with palatability, because it is also influenced by physiological [20,21,22,23,24,25,26], environmental [27,28,29], and experiential factors [33,34,38] independent of the hedonic value of the food. Accordingly, palatability can be conceptualized as an integrative process in which the sensory properties of a food are perceived and hedonically evaluated by the animal, influencing subsequent acceptance, preference, and consumption. This directional relationship is illustrated in Figure 2.
4. Contradicting Behavioral Indicators
Investigating feline behavioral indicators to assess responses has been used since 1979 [43] and, more recently, to obtain a complete picture of palatability alongside consumption data [9,10]. The literature has identified a set of behavioral indicators (before, during, and after consumption) that can be broadly categorized as positive or negative with respect to the palatability of the food provided. Positive indicators include an increase in the duration of having the eyes less than 50% open [4], increased lip licking [4,21,39], more frequent mouth smacks [4], and more frequent tongue protrusions [4]. Negative indicators include flicking of the ears backwards or flicking of the tail [39], more frequent tongue-protrusion gapes [4], and a non-significant trend towards the combined behavior of pawing the floor, shaking the head, and limb flailing [44]. Some behavioral indicators, such as posture (sitting or standing) [21,22], twitching of the back skin [39], and licking of the paws [21,44] do not seem to correlate with the palatability of the diet, while playing with the food may indicate satiety rather than palatability [45]. This conventional categorization has facilitated schematic overviews of the relationships between food intake and behavioral responses in cats, such as those presented by Pekel et al. [3] (Figure 5). While these summaries provide a valuable overview, they inevitably simplify a more complex body of literature. A closer examination reveals that several behavioral indicators cannot be consistently classified as purely positive or negative, and that several controversies and details are overlooked in such displays. For example, face grooming is seen as a positive indicator after consuming palatable food [21], whereas body grooming—often noted after consuming less palatable foods or after a period of stress—is seen as a negative indicator [39]. Consequently, classifying “grooming” as a single behavioral indicator may obscure these important distinctions and lead to oversimplified interpretations.
Beyond such simplifications, several behavioral indicators seem to contradict each other across studies. Hanson et al. [4] associated more frequent nose licking with a pleasant taste experience; both Savolainen et al. [39] and van den Bos et al. [21] associated it with a response to a negative or less palatable taste experience; and Bennett et al. [43] interpreted it as a fear response in cats—three fundamentally different interpretations of the same behavior. Sniffing of food shows a similar pattern. Van den Bos et al. [21] and Becques et al. [22] interpreted increased sniffing as hesitation to consume a less palatable diet; Owens et al. [44] found no effect of diet on total sniffing time, a negative correlation between sniffing during the acceptance test and the subsequent preference-to-intake ratio, but no correlation between sniffing during the preference test and the intake ratio; and Kim et al. [46] noted a significant difference in the first sniffing bout between two diets with equal palatability. Latency to eat also showed inconsistent interpretations. Van den Bos et al. [21] and Owens et al. [44] found a rapid approach to food to be indicative of high palatability, whereas Becquest et al. [22] only noted a significant difference during the cats’ final approach to the bowl, when there was no longer a significant difference in intake. Accordingly, Table 1 expands on previous summaries by distinguishing behaviors that are consistently associated with palatability from those with conflicting or context-dependent interpretations.
Several factors contribute to these contradictory findings. Behavioral similarity makes accurate coding of facial-expression detection difficult. Becques et al. [22] and Owens et al. [44] both explicitly acknowledged the difficulty in distinguishing between nose licking and lip licking. However, these two behaviors appear to have opposite palatability interpretations, with lip licking seen as a positive indicator [4,21,39], while nose licking is more generally accepted to be a negative indicator [21,39,43]. If observers cannot reliably distinguish between them, the categorization is compromised. Several studies use fundamentally different test formats (sequential single-bowl tests [21], two-bowl preference tests [4], and five-bowl preference tests [46]), as well as different exposure durations (5–30 min [4,21,44,46], 20 h [22], or until eaten [39]), and subsequently conclude different interpretations of the same behavior. Owens et al. [44] observed that some cats consistently sniff the bowl on the right side more frequently and for longer durations than the bowl on the left, suggesting that side preference may influence sniffing behavior. This problem is compounded in multi-bowl designs [4,44,46], where a behavior such as sniffing could correspond to any specific diet. This methodological fragmentation is also seen in the manner in which behaviors are evaluated, such as “licking/sniffing food” [21], “sniffing/investigating” [47], “length of sniffing” [22], “bouts of sniffing” [44], and “first sniffing bout” [46]. Thus, the contradiction is not just about interpretation; the studies measure different indicators (or combinations of indicators) under different conditions. Similarly, eating speed also shows contradictory interpretations. Owens et al. [44] objectively quantified speed of consumption (g/min) based on food intake and total eating time, whereas Knight and Satchell [47] relied on an owner-completed survey in which consumption speed was not standardized or objectively measured, leaving the results vulnerable to interpretation bias. Individual variability further complicates interpretation and can impact eating patterns [48], sniffing behavior [44], and the hunger response [21]. Consequently, studies with smaller cohorts [44] are more susceptible to individual effects than those with larger cohorts [21,22,39,46]. Beyond design limitations, several gaps exist in the current body of literature. No study has directly compared the impact of sex on behavioral indicators of food palatability, despite evidence that sex-specific behavioral patterns exist in other contexts, such as the repetitive sniffing behavior associated with personality in male cats, but not females [49], and sex-related differences in food preference under specific environmental conditions, such as climate season [50]. These findings suggest that sex may influence how the sensory properties of food are perceived or how palatability is behaviorally expressed. Female cats may, for example, be more sensitive to volatile cues [50]; however, this hypothesis requires further investigation. Future studies should, therefore, test males and females separately and examine sex-by-palatability interactions across multiple behavioral indicators. Another gap is the impact of different diet types, such as raw, wet, or dry foods [10,42]. Physical characteristics of the diet may influence how behavioral indicators of palatability are expressed. For example, lip licking has been associated with greater palatability, predominantly in studies involving wet or other moist foods [4,21,39]; however, it remains unclear whether this response reflects hedonic evaluation alone or may also be influenced by physical characteristics such as moisture and food adherence around the mouth. Future studies should, therefore, directly compare behavioral responses to wet and dry diets while controlling for differences in palatability. Similarly, most studies compare diets that are expected to differ in palatability, yet the literature has not addressed whether behavioral responses become more pronounced as the difference in palatability increases. It should be noted that no behavioral indicator or facial expression is exclusively positive or negative, and that frequency and duration play an important role [24,37]. For example, flattening of the ears is considered a negative reaction and is observed in 89% of negative conditions, but it is also observed in 55% of positive conditions [37]. Similarly, some behavioral indicators, such as sniffing, occur regardless of the presence of food [51]. This means that simply recording the presence or absence of a behavior [39] is insufficient; quantitative measures of frequency and duration are essential, yet some studies [22] only report one or the other, and no studies have established threshold values.
These contradictions have a direct consequence: as the literature is presented currently, behavioral indicators cannot be interpreted as standalone measures of palatability. However, some papers have used these behavioral indicators to determine food palatability from survey data. For example, Knight and Satchell [47] used the existing literature [4,21,22,39] to classify “rapid approach to food” and “eating quickly” as positive indicators of palatability, and “sniffing/investigating food” and “dropping food” as negative indicators. Based on pet owners’ reports of these behaviors, they concluded that vegan and meat-based diets were similarly palatable. However, the literature used to support these classifications is not consistent: Van den Bos et al. [21] and Becques et al. [22] report conflicting findings regarding a rapid approach to food; Hanson et al. [4] and Becques et al. [22] report opposing conclusions regarding sniffing; and Savolainen et al. [39] noted that food dropping cannot be considered an independent indicator because it can only occur when the cat has eaten or is eating, meaning that its higher frequency with palatable food is a consequence of greater consumption rather than an independent measure of palatability. The same behavioral indicators may be positive, negative, or have no meaningful response, depending on the study context. Thus, behavioral data should be interpreted alongside consumption data. For example, a cat licking its lips after eating a large portion of food may indicate good palatability, whereas a cat licking its lips after food refusal is not. Furthermore, lip licking may also simply be the result of cleaning food residue after consuming a particularly messy diet. Without the context, the behavior alone is ambiguous. Just as a sensory lexicon provides a standardized vocabulary for describing the sensory properties of pet food [52,53], this field would also benefit from a standardized ethogram for feline feeding behavior. It should define which behaviors to code, conditions under which they are informative, the quantitative thresholds for frequency and duration, and the food formats for which each indicator has been validated. Standardizing these criteria would improve consistency between studies, reduce contradictory interpretations, and provide a more robust foundation for validating behavioral indicators of feline food palatability.
5. Methodological Fragmentation
Many contradictions in the literature cannot be resolved by re-examining individual studies, as the root cause is methodological mismatch. While the existing literature already provides a clear description of the available palatability testing methods [2,9,10,28], several studies [25,36] still apply an evaluation method that does not align with their stated objective. One such methodological mismatch is using the acceptance and preference tests interchangeably. For example, Ilias et al. [36] aimed to determine the “palatability and acceptability” of prescribed diets to “ensure that DSH cats continuously eat”. This is fundamentally an acceptance question, yet a two-bowl preference test was used to compare the therapeutic diets [36]. In another case, Bernachon et al. [25] investigated the “comparative palatability of five supplements”, implying an objective of preference or comparative ranking; however, each cat received only its allocated product, meaning the study measured acceptance within separate groups rather than a direct preference between products [25]. The preference test’s strength lies in discriminating between two products with subtle differences, not in determining whether a single product meets minimum intake thresholds, which is better addressed through acceptance testing [10,54]. This distinction is supported by Hours et al. [55], who reported that voluntary food intake measurements and two-bowl preference test results can yield divergent conclusions about the same diets. Given that these two testing methods address fundamentally different questions, it is surprising that only a few published studies have performed both tests in sequence for a more comprehensive assessment [24,55,56]. Similarly, the liking test [57], a theoretically ideal position between acceptance and preference testing, has also not been adopted by the research community and is only mentioned in review papers [10,37]. Unlike the Concurrent Schedule Paradigm [58,59] and the Concurrent Paired-Associative Preference [60,61], whose limited adoption can be attributed to practical constraints, the liking test’s absence from the literature is harder to explain. It requires no specialized equipment, uses a standard one-bowl format, and was specifically designed to be feasible with expert panels [57]. Its absence represents an independent validation gap.
Part of the methodological fragmentation in the literature is driven by resource constraints. Pet food producers and academic laboratories may lack the time, animals, infrastructure, and/or funding to run multiple test formats to ensure a complete palatability picture [28]. Recognizing this reality, Figure 3 proposes a decision framework to guide method selection based on the primary research objective, with the branching questions representing the research questions applicable to each methodology. This enables researchers to identify the single most informative test for their specific question, rather than defaulting to convention.
Another methodological factor contributing to contradictory palatability findings is the choice of sample size and test duration. According to Tobie et al. [10], a preference test on an expert panel requires at least 30 cats to ensure statistical robustness, and in-home testing requires 100. However, more recent work indicates that the ideal sample size depends on the statistical framework and test methodology, as factors such as cat and food characteristics, degree of adaptation or training, and testing environment may influence variability and, consequently, the number of animals required [62,63]. Pires et al. [62] used power analysis to demonstrate that a preference test with 23 cats is sufficient to detect a difference of 0.10 in the intake ratio with 75% power, whereas Bos et al. [63] used a precision-based approach to justify a sample size of 17–23 to reach a 10% margin of error (MoE) in two-bowl preference tests, depending on the contrast between the dry extruded foods tested. For acceptance tests, Bos et al. [63] noted greater variability, with 12–43 cats required to achieve the same MoE. These are two fundamentally different approaches. Statistical power is the probability of rejecting a false null hypothesis [64]; thus, power analysis can be used to determine whether a study is adequately powered to detect a specified effect, whereas MoE is used to determine how precisely the mean intake can be estimated [65]. Currently, no study has provided a statistically derived sample size for one-bowl acceptance tests on expert panels. Bos et al. [63] noted that “a shorter study length with a greater number of cats is superior”, as increasing the test duration may result in dropout, reducing accuracy. Both Pires et al. [62] and Bos et al. [63] indicate that a single test day may be sufficient; however, this should be distinguished from the need for pre-test adaptation or familiarization, which may be required when animals are not accustomed to the testing procedure. Despite this, many studies [36,66,67,68] continue to use extended protocols—5–10 days with panels of 10 or fewer. Such designs may provide limited ability to accommodate inter-individual variability, while extended testing may additionally increase the risk of dropout-related bias. In contrast, industry practice and several academic studies [40,69,70,71] typically use a panel of 20 cats tested over 2 days, resembling the recommendation of Bos et al. [63] more closely. The importance of reporting sample size rationale and statistical framework alongside the study results is often overlooked. This was highlighted by Pires et al. [62], who found that, of the 16 studies included in their systematic review, only three reported sufficient information to allow a full assessment of the statistical power of cat food preference tests. This lack of transparency likely contributes to the contradictions between studies, and it is therefore recommended that an appropriate sample size be determined by using power analysis based on the expected difference in intake ratio, anticipated variability, significance level, and desired statistical power [62], rather than adopting a fixed number of cats.
Beyond sample size and test duration, a deeper form of methodological fragmentation may arise through the propagation of claims across secondary sources without verification against the original evidence. Seasonal variation in feline food intake provides a clear example. Watson et al. [9] stated that cats are prone to “eating less in the winter than in the summer”, citing Tobie et al. [10], who similarly stated that cats eat “less during winter”. However, Tobie et al. [10] cited Serisier et al. [27] as the underlying primary evidence, whereas Serisier et al. [27] actually reported the opposite pattern, with food intake increasing during winter. This direction is also consistent with the findings of Alegría-Morán et al. [50], who reported increased food intake under colder conditions. Thus, the issue illustrated here is not simply disagreement among studies, but the propagation of an incorrectly reported direction of effect from a secondary source into the subsequent literature. Another example includes the statement of Watson et al. [9] that “cats are driven to foods with a high protein and fat content and avoid carbohydrate-rich foods” by citing Hewson-Hughes [72] and Salaun et al. [73] without acknowledging that the high-carbohydrate target identified by Hall et al. [74] was specifically observed under conditions of equal palatability. This contributes to methodological fragmentation by creating an inconsistent evidence base, making it difficult to compare studies, accurately interpret findings, and develop robust experimental designs.
6. Owner Perception Paradox
The central contradiction of feline palatability is that the pet owner who chooses and purchases the pet food is not the final consumer. Pet food is seen as an easy opportunity to create a positive experience in the everyday life of a pet [4]. Pet owners interpret their pet’s willingness to eat as an immediate indicator of satisfaction [75], and when pets reject their food, owners feel they are not providing adequate care and consequently seek a better alternative [2,76]. However, what happens when the pet does not outright refuse the food but also does not find it maximally palatable? The cat adjusts its intake downward within an acceptable range, and the owner assumes that palatability is adequate, unaware that the cat would consume significantly more of a preferred alternative. This is what we term the “owner perception paradox”; the sensory preferences, cognitive biases, and emotional attachments of the owner override the biological reality of the cat’s preferences.
Owners and cats perceive food through different sensory channels. Often, pet owners evaluate the food based on appearance and aroma only [36,77,78], while cats evaluate the complete palatability experience, including aroma, taste, and texture [38,39,79,80]. The physiological differences are substantial. Cats exhibit 21 cm2 of olfactory epithelia, compared to the 3–4 cm2 in humans [36], and their sense of smell is roughly fourteen times more sensitive [81], with oxidation detection alone being a hundred times more acute than in humans [82]. The vomeronasal organ further allows them to detect volatile compounds that are imperceptible to human noses [82,83]. Cats possess ≈ 470 taste buds, considerably fewer than humans (≈10,000), suggesting lower taste sensitivity [3]. However, this impression is misleading, as cats have a heightened sensitivity for certain bitter compounds [4,29,37], their umami receptors are not activated by the same amino acids [84], and they lack the functional receptor for sweet taste entirely [18,85], meaning that the sugary aromas that often attract human attention in pet food marketing are invisible to the feline palate. Thus, the sensory attributes used by the owner to select a diet for their cat are fundamentally different from the sensory attributes used by the cat to consume the food.
Ilias et al. [36] demonstrated this divergence by comparing the odor scores of a human panel to the consumption outcome of a cat panel, across four prescribed diets. Both species agreed that the hypersensitivity diet was least preferred, but their discriminatory capacity differed thereafter. Cats produced a clear rank ordering across the remaining three diets based on intake (although this was not statistically significant), whereas the human odor panel rated the three remaining diets as broadly equivalent [36]. Similarly, owner surveys have been a widely used approach to assess pet food palatability from the owner’s perspective since at least the early 1990s [86,87,88,89,90], but they also provide limited interpretation due to bias [9,47], inconsistent judgment [86], and variable pet behavior interpretations [91]. This is evident in the study conducted by Sanderson et al. [86], where owners were unknowingly given the same diet to feed their dog in consecutive periods, and 27% gave inconsistent responses. While this study involved dogs rather than cats, the underlying principle still applies. Similarly, Bos et al. [63] found that the owners’ categorization of their cat’s pickiness did not correspond with their food intake. Cats categorized as “easy” eaters by their owners had an overall lower food intake compared to “difficult” or “very difficult” eaters. Yet, instead of owners being uncertain about their cat’s preferences, they actively project their own preferences onto their cats. Di Donfrancesco et al. [92] found a correlation of 0.93 between how much owners liked a product and how much they predicted their pet would like it. However, the drivers of owner liking (appearance, color, and aroma) are not the drivers of pet intake (aroma, taste, and texture). This projection is reinforced by the growing influence of pet humanization [93,94,95]. In a survey of 1380 cat owners, Mace et al. [96] noted that the most important feature influencing willingness to purchase alternative diets was good health outcomes, limited to the owner’s knowledge level. Thus, their purchases reflect their own values (sustainability, organic ingredients, or human-grade quality) rather than the cat’s nutritional and sensory requirements [93,96,97]. This leads to specific, demonstrable misalignments. Owners frequently prefer diets marketed as natural or free from additives [96,98], unaware that removing certain ingredients, such as sugars, compromises Maillard-driven flavor generation [99,100]. Similarly, owners prefer diets with higher protein and fresh meat content [75], unaware that the gravy portion of a chunks-and-gravy diet drives palatability [101]. A survey by Erzurum and Kayar [75] captures this contradiction: pet owners rated “pet preference (palatability)” as their highest selection criteria, and “visual appearance” as the lowest. Owners believe they are choosing based on their cat’s preference. Yet, as Di Donfrancesco et al. [92] demonstrated, what the owner believes their cat prefers is indistinguishable from what the owner personally prefers. The consequences are also seen in the academic literature. Knight and Satchell [47] surveyed 1135 cat owners and found that diet (conventional, raw, and vegan) made little difference to food-oriented behavior, concluding that vegan pet foods are as palatable as conventional meat or raw meat diets. This finding was characterized by subsequent reviewers as “extremely subjective” [9] and contributes to “ambiguous results” [11]. Owner reporting does not merely introduce noise; it can actively flatten real palatability differences. Importantly, the owner perception paradox is unlikely to be uniform across owners with different feeding experiences or cultural backgrounds. Owners with greater knowledge may weigh product attributes differently [97], while caregivers who actively monitor food intake may be better positioned to recognize acceptance and preference [24]. Cultural differences further shape responses to pet food odors [78] and the importance assigned to attributes such as naturalness, freshness, sustainability, palatability, brand, and veterinary advice [28,75,96]. However, demographic effects are not consistent across populations; for example, age and education were unrelated to purchasing behavior in one Turkish sample [75], whereas older guardians were less influenced by price and convenience in a predominantly UK and European survey [96]. Thus, owner perception should not be treated as a single, stable source of bias. An important research gap is to determine which owner characteristics increase or reduce alignment between perceived and objectively measured feline palatability across different cultural and feeding contexts.
The resolution to this paradox is not to exclude owner interpretation entirely, as owners remain the ultimate decision-makers. Rather, it is to recognize that owner evaluation and feline consumption measure fundamentally different constructs. Owner surveys capture purchasing intent and owner satisfaction; consumption testing captures the cat’s hedonic response. These are complementary measures, not interchangeable. Owner perception should not be assumed to reflect pet palatability, nor should pet palatability be assumed to result in commercial success. Successful pet food products depend on satisfying both audiences, whose sensory channels, cognitive frameworks, and evaluative criteria are misaligned.
7. Limitations
This review has several limitations. First, a narrative review involves synthesizing evidence across different studies, meaning that the identification and grouping of themes involve some degree of subjective interpretation by the authors. Second, the primary search was limited to the Scopus database and English-language publications, although additional studies were identified through targeted searches and citation tracking. Finally, the review incorporates a heterogeneous body of evidence, including experimental, methodological, and review publications, which may limit direct comparison between studies.
8. Conclusions and Future Research Directions
Despite considerable advances over recent decades, feline food palatability research continues to be hindered by methodological and conceptual inconsistencies. This critical narrative review demonstrates that variable conceptual definitions, experimental execution, and interpretation of behavioral indicators limit the development of a reliable framework for evaluating feline food satisfaction.
Thus, future research should shift from a descriptive to a prescriptive approach, providing methodological and interpretation guidance instead of only identifying or summarizing findings. The decision tree proposed in this review may help align trial designs with their specific research objectives, while routine reporting of statistical power and/or precision would improve the robustness and reproducibility of palatability claims. Similarly, separating owner impressions from feline responses and adopting a standardized ethogram for behavioral indicators will resolve ambiguity and improve comparability between studies.
Future research priorities can be structured into three areas. Basic research should focus on evaluating test–retest reliability, assessing the influence of food format (e.g., dry, wet, and semi-moist) and animal characteristics (e.g., age, sex, neuter status, and health status), and comparing owner perceptions with objectively measured feline responses. Methodological standardization should prioritize validating a standardized feline feeding ethogram, assessing whether the proposed decision framework improves the selection of palatability methodologies for different research objectives, and developing minimum reporting recommendations for feline palatability studies. Finally, industrial application should evaluate the applicability of these approaches across different products, testing facilities, and commercial settings. By harmonizing how findings are reported and interpreted, the industry can finally transform isolated observations into a coherent and actionable evidence base.