Still just a matter of taste? Sensorial appreciation of seafood is associated with more frequent and diverse consumption

Improving health and sustainability outcomes in WEIRD (Western, Educated, Industrial, Rich, Democratic) nations necessitates a reduction in red meat consumption. Seafood is often overlooked in achieving this goal. However, simply consuming more of familiar fish species places high stress on production of these species. For this reason, diversification of seafood consumption is also critical. Here the motives for seafood consumption (frequency and diversity) are investigated across two studies by adapting the 4Ns survey to the seafood category. This 16-item survey measures four factors underpinning meat consumption: namely that it is ‘Natural ’ , ‘Necessary ’ , ‘Normal ’ and ‘Nice ’ . Swedish consumers ’ hedonic and sensory expectations of two herring concepts (traditional pickled contra novel minced and presented as a burger) are also evaluated in relation to the 4Ns. Study 1 (N =


Introduction
Changing our diets is necessary to mitigate the effects of climate change (Clark et al., 2020).Given the heavy environmental toll of meat and dairy production (Poore & Nemecek, 2018;Willett et al., 2019), guiding consumers to substitute meat (red meat especially) and dairy with more sustainable proteins is increasingly relevant.However, psychological, social, cultural, and sensory barriers to reducing meat consumption persist among many consumers.One pitfall that consumers can fall into is rationalising, or justifying, their current behaviour.Rationalisation is interesting from the perspective of dietary change because it can manifest even in individuals who intend to change (Collier, Normann, Harris, Oberrauter, & Bergman, 2022;Rothgerber, 2020).Regarding meat consumption, this could be because both aligning one's beliefs with one's behaviour and vice versa can relieve the mental discomfort of the 'meat paradox': omnivorous consumers' cognitive dissonance resulting from holding both positive and negative views about meat consumption simultaneously (Loughnan & Davies, 2019;Loughnan, Bratanova, & Puvia, 2012).
To measure the main justifications for meat consumption, Piazza et al. (2015) developed the 4Ns scale.This scale consists of 16 items (statements) and is theoretically comprised of four subscales (or subcategories), each comprised of four related items: 'Natural' -the belief that craving meat is a natural, biologically driven inclination for humans, 'Necessary' -that meat is necessary for good health, 'Normal'that eating meat is a socially acceptable and common practice, and 'Nice' -that eating meat provides sensory enjoyment.The 4Ns scale is grounded in understanding the moral rationalisations for meat consumption, which is a more specific conceptualisation than the more general concept of 'motives for consuming meat'.Nonetheless, the 4Ns scale appears to capture the underlying factors that drive individuals' meat consumption.This is supported by the finding that scores on the Motives to Eat Meat Inventory (MEMI) and the 4Ns are strongly positively correlated, leading to Hopwood et al. (2021) suggesting that the latter can also be interpreted as measuring overall motivation to consume meat.Hopwood et al. (2021) further noted that, despite the conceptual similarities and even reuse of the subscale names between the MEMI and the 4Ns scale, there are still subtle differences between rationalisations and motivations.Rationalisations typically serve to relieve psychological discomfort (i.e., ambivalence) and justify the continuation of some behaviour.On the other hand, motivations describe the reasons individuals provide for engaging in some behaviour and do not necessarily imply internal conflict.Therefore, in the absence of ambivalence (induced or pre-existing) or having a behaviour challenged, rationalisations and motivations seem to be conceptually highly similar.
Seeking to understand the similarities between motives for consuming meat and dairy products, Collier, Harris, Bendtsen, Norman, and Niimi (2023) adapted the 4Ns scale for the dairy category and found that sensory enjoyment ('Nice') was consistently the strongest driver of consumption for both product categories.That 'Nice' scores appear particularly relevant for meat and dairy consumption patterns matters because the sensory aspects of proposed alternatives to meat remain a powerful barrier to increased acceptance (Cordelle, Redl, & Schlich, 2022;Fiorentini, Kinchla, & Nolden, 2020; though see Niimi et al., 2022).Alternatives such as plant-based proteins and legumes are often highlighted as viable substitutes for meat (Graça, Godinho, & Truninger, 2019;Onwezen, Bouwman, Reinders, & Dagevos, 2021).However, seafood could also play a role in this protein shift as it generally outperforms terrestrial animal-sourced proteins in nutritional and environmental terms (Hallström et al., 2019).Intrinsic characteristics, such as aroma, taste, and texture have been identified as main factors influencing both approach and avoid behaviour for fish consumption (Morales & Higuchi, 2018;Saidi, Cavallo, Del Giudice, Vecchio, & Cicia, 2022).However, diversification should also be encouraged, as simply substituting red meat with the most commonly consumed seafood species (e.g., salmon) may result in disproportionate stress on those species (Hornborg, Bergman, & Ziegler, 2021).A recent study segmented Swedish consumers based on fish consumption frequency (i.e., of cod and salmon), with the majority being occasional or infrequent consumers (Armbrecht, Lundberg, & Skallerud, 2023).Thus, examining seafood consumption more broadly, including other seafood groups aside from fish, remains relevant.
The strongest barriers to changing seafood consumption patterns include sensory concerns, perceived difficulty of preparation, food neophobia (FN), and product availability (Carlucci et al., 2015;Govzman et al., 2021;Jaeger, Rasmussen, & Prescott, 2017).Understanding the motivations driving consumption of specific food categories is relevant to support changes in consumption patterns, whether the end-goal is a decrease (in the case of red meat) or increase (in the case of seafood, as a transition strategy).For example, identifying that sensory enjoyment is a strong driver of dairy consumption indicates that the sensory profile of plant-based analogue products should be a primary focus in product development (Collier et al., 2023).Given the previously identified relevance of the 4Ns for predicting consumption frequency and sensory expectations of foods, in this work we assess whether it can be adapted and used to provide deeper insight into the motivations for seafood consumption.In a manner similar to Collier et al. (2023), study 1 compares the original 4Ns scale and its adapted version for seafood within the same consumers.The test-retest reliability of the seafood 4Ns is evaluated, and how 4N endorsement associates with seafood consumption frequency and diversity is also assessed.Study 2 expands on the findings of study 1 by evaluating the relevance of the seafood 4Ns to seafood consumption patterns (frequency and diversity) in a larger sample, where FN and food agency are also measured.This facilitated evaluating how reluctance to try unfamiliar/highly arousing foods and perceived proficiency with food preparation (previously identified barriers to seafood consumption) related to consumption frequency and diversity among Swedish consumers.
Additionally, study 2 investigates in more detail the acceptance of a herring meal, presented in one familiar and one novel format, and how this is affected by endorsement of the seafood 4Ns and other consumer characteristics.Herring is a small pelagic fish that is traditional in the Scandinavian diet and is among the top four most popular seafood species in Sweden (Seafood in Sweden, 2022).The untapped potential of this species makes it an interesting case study: its consumption is currently highly seasonal, typically limited to celebrations and traditions (e.g., Midsummer, Christmas, and Easter), and it is increasingly underutilized in Sweden despite its health and sustainability benefits (Hallström et al., 2019).The overall aim of this work was thus to understand the relative importance of different drivers of seafood consumption for both consumption frequency and diversity, as well as to attempt to identify specific sensory expectations of herring presented in its traditional contra an alternative, novel format.

Study 1: participants and procedure
An online survey was developed and advertised digitally through the authors' social networks and by using a popular research recruitment platform in Sweden (Accindi), as well as dissemination to a consumer database held at RISE Research Institutes of Sweden.This yielded a final sample size of N = 304 (of 401 who initiated the survey) who completed the survey in full, see Table 1.Participants responded to both the original 4Ns questionnaire targeting meat (Piazza et al., 2015) and a version of the 4Ns survey adapted for seafood (within-participants design), which both used 7-pt Likert scales (1 = "strongly disagree" to 7 = "strongly agree").They then provided demographics (gender, diet, age) and consumption frequency of five meat categories (beef, lamb, poultry, pork, and game), and six seafood categories (fish, shellfish, crustaceans, roe, cephalopods, and echinoderms).Example species of all seafood categories were provided.The 4Ns survey was adapted for seafood by replacing the word 'meat' with 'seafood' and adjusting for grammatical correctness where necessary.Consumption was reported on an 8-point ordered category scale (1 = 'never in the past 12 months', 2 = 'once in the past 6 months', 3 = 'once in the past 3 months', 4 = 'once a month', 5 = 'once every 2 weeks', 6 = 'once a week', 7 = '2-4 times a week', 8 = 'almost every day').Participants were also asked to provide their email address if they were willing to complete a follow-up survey related to the study.The follow-up, which was distributed approximately 3 weeks after the initial study and was completed by N = 123 participants (40.1% of the original cohort), was a repeat of the seafood 4Ns questionnaire in isolation, so that test-retest reliability could be assessed.

Study 2: participants and procedure
Participants in Study 2 were recruited through an international recruitment service (Norstat AB).An attention check was also included where participants were instructed to select "strongly agree" on the presented scale.After screening for diet (excluding veg*ns and individuals with seafood allergies), and passing the attention check (excluding participants giving any response other than 'agree' or 'strongly agree'), 514 respondents (see Table 1) had fully completed the same seafood 4Ns survey as in study 1, the short version of the Cooking and Food Provisioning Action Scale (CAFPAS-short; Karlsson et al., 2023), the alternative Food Neophobia Scale (FNS-A; De Kock et al., 2022), and provided demographic (gender and age) and consumption information.
The consumption questions were the same as in study 1 except that 'small pelagic fish' was included as a separate category.This is because participants also provided their sensory expectations of two different preparations of herring (shown to participants visually as photographs, see Fig. S1, Supplementals).This is also the reason veg*n respondents were excluded in study 2 but not study 1.One photograph showed a traditional preparation of pickled herring pieces and the other showed an alternative preparation of herring-mince in the shape of a wallenbergare, a type of burger commonly served in Sweden.Participants first reported how much they expected to like each herring concept using a 9 pt hedonic scale (1 = "dislike extremely" to 9 = "like extremely").Their sensory expectations were then captured using check-all-that-apply (CATA) where 22 attributes were included (Table S2 in the supplementals, see also Fig. 5).The attributes included in the CATA task were based on qualitative data collected in an as-yet unpublished study conducted by the authors (EC, JN, and ESC) targeting consumers' sensory expectations of different seafood species, including herring, as well as on input from an industry partner involved in the parent project of this study.Both sample (traditional vs novel) and CATA attribute presentation order was randomised across participants (except 'none of the above' which always appeared last).

Data analysis
4N scores for meat (study 1 only) and seafood were calculated as the sum of responses across all items (total score) as well as within each subscale for each study (theoretical range of 4-28 pts for each subscale, and grand score of 112 pts for the full 4Ns scale).The structure of the 4Ns questionnaire was analysed using exploratory factor analysis (EFA) on both the meat and seafood consumption versions using XLStat ver.2022.4.1.Missing data from either the meat or seafood 4Ns were estimated with mean or mode (a separate analysis was run after removing participants with missing values, however the change in EFA model was negligible and it was decided to retain N = 304 for consistency with the other data analyses related to study 1).The similarity of EFA scores between the meat and seafood 4Ns from four factors were analysed with RV coefficient analysis using XLStat ver.2022.4.1.The test-retest reliability was assessed by contrasting scores for the full seafood 4Ns and its subscales across test time using multilevel regression (rethinking::ulam in R, version 4.1.1)to account for the repeated measures design.In study 2, CAFPAS-short and FNS-A scores were calculated as described in Karlsson et al. (2023) andDe Kock et al. (2022) respectively.For all survey measures, items were reversed where specified by the original developers and internal reliability was assessed using Cronbach's α.
To describe the diversity of respondents' seafood consumption, a dichotomous variable named 'diverse consumer' was defined and participants classified as either: a diverse seafood consumer who reported consuming at least three categories of seafood other than fish (and small pelagic fish in study two) at least once a month (1) or a non-diverse seafood consumer who did not fulfil the above criteria (0).This variable was defined before conducting statistical analysis.
For both studies, conditional associations between 4N subscale scores and total seafood consumption (multiple linear regression) and odds of being a diverse consumer (logistic regression) were estimated with Bayesian inference (Bendtsen, 2018) in R (rethinking::quap).Expected liking was contrasted across the two herring concepts using multilevel regression with normal priors centred on the middle of the hedonic scale (μ = 5, σ = 1).Means of the posterior distributions are used to estimate effect strength, and the probability of association in the direction of the mean is provided.For study 1, all coefficients received standard normal priors centred on zero.The priors used in analysis of the 4Ns data in study 2 were updated to reflect the posteriors obtained in study 1.For study 2, FNS-A and CAFPAS-short scores were additionally included in the regression models to estimate their influence on seafood consumption (standard normal priors assigned).Associations between the psychographic variables were assessed with correlation analysis, using Bayes factors (BF 10 , evidence in favour of H 1 over H 0 ) for inference.
Data obtained from CATA were analysed using McNemar tests to determine significant differences in citation frequency of expected sensory attributes between the traditional and alternative preparations of herring.Testing was performed with α = 0.05 using XLStat ver.2022.4.1.

Evaluating the seafood 4Ns: exploratory factor analysis
The underlying structure of the 4Ns survey was evaluated for consumption of meat and seafood in study 1.Most of the variance was explained on the first factor for both meat and seafood versions of the questionnaire at 51% and 41% respectively, see Table 2.The structure of the loadings was also similar between the meat and seafood versions, indicating structural alignment.RV coefficients calculated between the meat and seafood 4Ns supported high similarity when analysed on the first factor (RV = 0.95, p < 0.001) as well as on four factors (RV = 0.844, p < 0.001).
Table 3 shows the internal reliability of each scale and their subscales for both studies.The majority of (sub)scales showed satisfactory internal reliability (Cronbach's α > 0.70), although some, e.g., the 'Normal' subscale for the 4Ns for both meat and seafood, performed fairly poorly.This subscale has previously shown low internal reliability (e.g., Collier et al., 2023;Piazza et al., 2015) which likely reflects the lack of clarity in what respondents should use as a reference for 'normality' (tight friend group vis-a-vis wider society).The internal reliability of the 'Natural' subscale varied somewhat between the two studies.Participants in study 2 typically had low-medium FN scores, which has a theoretical range of 1-7, and moderately high food agency.
Contrasting scores for the full seafood 4Ns scale and its subscales for the same individuals ~3 weeks apart (test-retest reliability) revealed that although there was a systematic tendency for scores to be higher at follow-up, the most likely (mean posterior) differences were very small (<1 pt), see Table 4.Because each subscale has a theoretical range of 4-28 pts (grand score of 112 pts for the full seafood 4Ns scale), a 1 pt difference does not represent a large shift in score.Complementary repeated-measures ANOVA with subscale and test time as factors showed that although the identified systematic increase in scores from initial testing to follow-up was statistically significant (p = 0.022, η p 2 = 0.042), no individual subscale differed significantly across testing times (all p > 0.99 except 'Necessary' where p = 0.07).The ANOVA table and accompanying raincloud plots are shown in the supplementals (Table S1 and Fig. S2).Overall, the test-retest reliability of the seafood 4Ns scale can be considered good.

Associations between 4Ns and consumption patterns: consumption frequency
Consumption frequency for the different seafood groups probed is shown in Fig. 1.In both studies, fish was by far the most consumed seafood category, followed by roe, crustaceans, and shellfish.That the consumption of roe and crustaceans was relatively high (second most consumed) was unsurprising due to the ubiquity of these products in Swedish food culture.Small pelagics were also reported as consumed relatively often in study 2, and it is assumed that 'fish' in study 1 includes considerations of small pelagics.The least consumed categories were cephalopods and echinoderms.Overall, consumption patterns were relatively similar across the two study cohorts.
Multiple regression revealed that the 'Nice' subscale had by far the strongest association with seafood consumption frequency (posterior mean = 0.76, probability of association >99%): all else being equal, stronger agreement that seafood is sensorially enjoyable predicted more frequent consumption of seafood, see Fig. 2 (top).Endorsement of seafood as 'Necessary' for good health also showed a positive association with consumption frequency (posterior mean = 0.18, probability of association = 99%), whilst higher ratings of seafood as 'Natural' was negatively associated with consumption frequency (posterior mean = − 0.17, probability of association = 96%).The 'Normal' subscale showed a somewhat positive association with consumption frequency (posterior mean = 0.07, probability of association = 74%).
A similar pattern was identified in study 2: of the 4Ns, the strongest association was detected for 'Nice' (posterior mean = 0.72; probability of association >99%), see Fig. 2 (bottom).Rating seafood as 'Necessary' was also associated with higher consumption (posterior mean = 0.16; probability of association = 99%) but, again, to a weaker extent than the association seen for 'Nice'.'Normal' (posterior mean = 0.04; probability of association = 67%) showed no marked associations with higher consumption of seafood, whilst 'Natural' showed a negative association with consumption (posterior mean = − 0.17; probability of association = 98%), suggesting that agreeing that seafood consumption is natural could have a somewhat detrimental impact on consumption frequency.

Associations between 4Ns and consumption patterns: diversity
Respondents were designated as either diverse or non-diverse consumers as described in Section 2.3: 24% and 12% respondents fulfilled the criteria of 'diverse consumer' in study 1 and 2 respectively.Logistic regression indicated that considering seafood consumption to be 'Nice' was associated with markedly increased odds of being classified as a diverse consumer (median OR = 1.20, probability of association >99%), see Fig. 3 (top).Increased agreement with 'Natural' was related to a decreased likelihood of being classified as a diverse consumer (median OR = 0.94, probability of association = 89%).Here, neither the 'Necessary' (median OR = 1.02, probability of association = 66%) nor 'Normal' (median OR = 1.01, probability of association = 42%) seemed strongly associated with diverse consumption.
Overall, study 2 suggested similar results: 'Nice' showed the strongest association with being classified as a diverse consumer, whereby individuals with stronger agreement that seafood is sensorially enjoyable were more likely to report consuming a wider diversity of seafoods (median OR = 1.13, probability of association >99%), see Fig. 3 (bottom).Again, 'Necessary' showed no marked association (median OR = 1.00; probability of association = 57%), and 'Natural' showed an inverse association (median OR = 0.94; probability of association = 95%) with consumption diversity.Compared to study 1, endorsement of seafood consumption as 'Normal' in study 2 (median OR = 0.93; probability of association =95%) was associated with decreased odds of being a diverse consumer, showing a difference across the two cohorts regarding the relevance of this subscale.

Expected liking and sensory attributes for the two herring concepts
Multilevel regression revealed that traditional pickled herring was expected to be liked more (mean posterior difference = 1.83, probability of association >99%) than the alternative, minced herring, see Fig. 4. As indicated by a positive correlation, individuals who expected to like the pickled herring also expected to like the minced herring (r = 0.46, BF 10 > 100).
Analysis of the CATA data revealed that many of the attributes were relevant for the expected sensory experience of both the pickled and minced herring (Fig. 5).Overall, for both pickled and minced herring, there were high citations for 'seasoned', 'salty', and 'fishy' as expected in the sensory character.Attributes such as 'neutral', 'bitter', and 'none of the above' were among the least cited for both herring presentations.Differences between the pickled and minced herring were also noted and more attributes were expected in the pickled herring (N = 9) compared to the minced (N = 5) that differed significantly (p < 0.001).Based on the images presented, consumers expected the pickled herring to be more 'seasoned', have more intensity in several basic tastes ('salty', 'sweet'), and have more textural character such as 'firmness', 'juiciness', 'chewiness', as well as 'sliminess'.In contrast, minced herring was expected to be more 'mushy', 'fishy', 'grainy', have 'small bones' and a 'dry' texture.These observations indicate that pickled herring may be expected to have more complex sensory characteristics encompassing basic taste, flavour, and texture/mouthfeel unique for this product in comparison to minced herring, which seems to mainly be characterised by texture and one notable flavour characteristic ('fishiness').Logistic regression was used to explore whether endorsing seafood as 'Nice' affected the odds of expecting different sensory attributes across the two herring concepts.Individuals with higher 'Nice' scores were more likely to expect pickled herring to be 'firm', 'seasoned', 'meaty', 'sweet', and 'juicy', and that minced herring should be 'firm', 'seasoned', and 'meaty', see Fig. 6.These might be considered sensory aspects that contribute to expected liking of each concept (OR > 1 with high probability of association relative to other effects).On the other hand, these individuals were less likely to expect 'slimy', 'bitter', 'neutral', 'small bones', 'sour', 'mushy', 'dry', and 'grey' attributes for pickled herring, and perceptions of 'slimy', 'bitter', 'sour', and 'sweet' for minced herring.These attributes may contribute to expected disliking among those who enjoy seafood (OR < 1 with high probability of association relative to other effects).

Table 3
Mean scores, standard deviations, and Cronbach's α for all scales and subscales used in studies 1 and 2.  (*asked only in study 2, **it is likely that respondents in study 1 counted 'small pelagics' in the 'fish' category).

Associations between consumer characteristics, expected liking, and consumption
Table 5 summarizes how consumer characteristics (demographics, 4Ns, FN, and food agency) were related to seafood consumption frequency and expected liking of the two herring concepts, according to multiple linear regression.Older individuals tended to report greater total seafood consumption and higher expected liking for the pickled herring than younger individuals, but no influence of age was detected for expected liking of minced herring.Women showed somewhat higher seafood consumption and expected liking for pickled herring than men, and markedly higher expected liking for minced herring.
The conditional associations between the 4Ns and seafood consumption were largely unaffected by including additional consumer characteristics in the model, reinforcing their unique contribution to understanding consumption patterns.The relationships between the 4Ns and expected liking for the two herring concepts were also similar to those for total seafood consumption, except the weaker (but persistent) influence of 'Nice', and the emergence of 'Normal' as a contributor to expected liking of both herring concepts.FN exerted a strong negative impact on seafood consumption, and slightly weaker but consistent inverse associations with expected liking of both concepts.Though not particularly strong effects, it is interesting thatall else being equalfood agency was associated positively with expected liking of the familiar pickled herring, but negatively with the less familiar minced product.
Correlations between consumer characteristics were also evaluated (Table 6).Food agency was positively correlated with age (r = 0.21, BF 10 > 100) and inversely correlated with FN (r = − 0.39, BF 10 > 100).Scores for 'Nice' and 'Necessary' were also higher among older participants (r = 0.29 and r = 0.24 respectively, both BF 10 > 100).FN correlated negatively with the 'Nice' subscale (r = − 0.21, BF 10 > 100), suggesting that higher FN individuals agreed less that seafood is sensorially enjoyable, but not with the other subscales.Table 6 also reveals that scores on the 4Ns were all positively correlated, an issue which is addressed in the discussion section.

Discussion
The primary aim of the present work was to better understand consumers' motives for eating seafood and how these relate to the frequency and diversity of the seafood species consumed.Given its previously established adaptability from meat to other food categories such as dairy (Collier et al., 2023), the 4Ns questionnaire was used and its suitability for the seafood category was evaluated in study 1.The analysis suggested that the 4Ns conception of drivers underlying meat consumption (i.e. that it is 'Natural', 'Normal', 'Necessary', and 'Nice') is also applicable to seafood, with good test-retest reliability.The meat version measured in the current study also aligned well with the original version developed by Piazza et al. (2015).A similar degree of transference was  previously reported from meat to dairy (Collier et al., 2023) and indicates that motives for consumption of different categories of animal products seem, thus far, highly comparable.The results and their implications are discussed in detail in the sections that follow.
Scores on the subscales of the 4Ns were highly correlated, and indeed EFA analysis in study 1 suggests that the seafood 4Ns is relatively unidimensional.This may raise concerns regarding whether the 4Ns survey captures four distinct reasons for consuming meat (or dairy or seafood).However, as highlighted by Hopwood et al. ( 2021), individuals who are motivated to consume meat for one reason appear to also endorse other reasons more strongly.In other words, although distinct reasons for consuming meat exist, endorsing one seems to increase the likelihood of endorsing others.Thus, although the constructs captured by the 4Ns scale (or the MEMI) are strongly related to each other, they retain important qualitative distinctiveness.
As observed here, and previously (Collier et al., 2023;Piazza et al., 2015), this distinction may be particularly relevant when investigating associations with other factors such as consumption patterns.Sensory enjoyment of seafood (higher 'Nice' scores) was the most powerful motive for more frequent seafood consumption detected here.This supports previous work indicating that sensory aspects are drivers of both liking and disliking of seafood (Carlucci et al., 2015;Saidi et al., 2022) and aligns with findings that hedonic appreciation of the taste of fish can increase consumers' willingness-to-pay for it (Morales & Higuchi, 2018).In addition to being the strongest predictor of total seafood consumption, endorsement of seafood as 'Nice' was also associated with increased likelihood of consuming a more diverse repertoire of seafood.As such, sensory enjoyment could also lead to a more diverse selection of options within a given food category, possibly driven by sensorial curiosity and pleasure seeking.The 4Ns framework has thus been successfully applied to seafood and its explanatory value for consumption patterns extended to dietary diversity.
Seafood consumption patterns were in line with previous findings for Swedish consumers (Seafood in Sweden, 2019).In fact, that the proportions of diverse seafood consumers were below 25% in both studies presented here reinforces that seafood consumption in Sweden remains typically below national dietary recommendations and limited to relatively few species.The sensory characteristics of seafood do not currently seem 'Nice' enough for many consumers, and thus the sensory aspect remains a major barrier for seafood consumption (Govzman et al., 2021).
Rating seafood as 'Necessary' for good health was also positively associated with consumption frequency, suggesting that messages highlighting the health aspects of consuming seafood may be valuable for increasing consumption in general.Indeed, evidence in favour of campaigns focusing on health increasing willingness-to-buy fish above other meats has been previously reported (Morales & Higuchi, 2018).Nonetheless, this approach may be less relevant for encouraging consumers to diversify their consumption as no association between 'Necessary' and diversity of consumption was found.Instead, sensory marketing and reinforcing the enjoyable sensory attributes of less conventional or unfamiliar seafood species may be an effective approach here.Birch, Memery, Johns, and Musarskaya (2018) suggested that fun, fast-food-inspired, and snack-like seafood products could help stimulate seafood consumption among UK adolescents.Appearance was noted as particularly important for this group, specifically that such products should appear 'tasty'.Visual appeal also seems important for encouraging more neophobic adult consumers who may be particularly sensitive to visual cues conveying texture, at least where oysters are concerned (Costa, Wrange, et al., 2023).
That agreement of seafood as 'Normal' contributed positively to expected liking of both herring concepts, but showed no marked influence on seafood consumption frequency, could be because of the place of Fig. 3. Median odds ratios and 95% CoIs showing the relative impact of each of the 4Ns on the likelihood of being classified as a 'diverse consumer' in study 1 and 2. Higher median odds ratio above 1 indicates higher likelihood of having diverse seafood consumption given an increase in 4N score.herring in Swedish cuisine.In some way, this may indicate an influence of 'familiarity' rather than 'normality', or at least reinforce that these concepts can be strongly interlinked for consumers.Similarly, concepts such as 'natural' and 'traditional' (Etale & Siegrist, 2021), as well as 'processed' and 'unhealthy' (Hässig, Hartmann, Sanchez-Siles, & Siegrist, 2023) can become conflated.The nuanced ways in which such heuristics affect consumers' choices both at the point of purchase and in a broader behavioural sense is relevant for improving communication with consumers.
The issue of heuristics and potential conflation of psychographic constructs prompts discussing the choice of scale for this study in more detail.Here, we assessed how well an adapted version of the 4Ns scale describes motivations for seafood consumption; however, the 4Ns is not unique in its scope.Scales such as the original 4Ns and the MEMI attempt to quantify specific psychographic aspects of omnivorous consumers' relationship with meat and capture the most prevalent reasons given for continuing to eat meat.Other survey instruments such as the Food Choice Questionnaire (FCQ; Cunha, Cabral, Moura, & de Almeida, 2018;Steptoe et al., 1995), provide a broader view of the factors affecting food choice by encompassing aspects such as price, convenience, and environmental friendliness, alongside naturalness, familiarity, and sensory enjoyment.It is worth noting that the concept of 'natural' as described in the 4Ns differs from that described in the FCQ: the former is grounded in the perceived innate biological inclination to consume meat, while the latter targets product content (i.e., no additives, no artificial ingredients).
However, the FCQ does not typically specify the food category of interest, opting instead for broader item phrasing.It seems plausible that specifying the relevant food category can unearth otherwise easily missed motives for their consumption, for example an individuals' agreement that meat is sensorially 'Nice' does not necessarily imply sensory enjoyment of, nor more frequent consumption of, seafood.Our results support this assertion.For example, during development of the 4Ns survey, Piazza et al. (2015) tested how 4N scores for meat   consumption related to consumption of other categories, including fish and seafood.There it was found that only the 'Necessary' subscale of the original 4Ns showed a significant correlation with consumption of seafood (though both 'Necessary' and 'Normal' associated with consumption of fish).Similarly, Onwezen, Reinders, Verain, and Snoek (2019) reported lower R 2 values when regressing frequency of fish consumption on the 'sensory' subscale of the FCQ than on the 'health' subscale (both the original and their own single-item version), implying that consuming food for sensory enjoyment in general may not be reliably associated with fish consumption frequency.In contrast, here, endorsing seafood as 'Necessary' for good health was associated with consumption frequency but an even stronger association with 'Nice' was detected.It is thus possible that adapting the FCQ or its single-item version (Onwezen et al., 2019) to specific food categories could provide deeper understanding on how both internal and external factors influence consumption patterns of those foods.
FN and food agency were also investigated here.FN was negatively associated with seafood consumption frequency and 'Nice' scores but no other seafood 4Ns subscale.These results are consistent with previous findings that higher FN individuals tend to reject seafood (Jaeger et al., 2017) and that FN is associated with higher responsiveness to sensory properties and lower levels of food enjoyment (Prescott, Chheang, & Jaeger, 2022).FN has a negative impact on food acceptance, indirectly via increased levels of emotional arousal (Costa, Niimi, et al., 2023) that can be elicited by characteristics beyond novelty (e.g., flavour intensity, Jaeger, Hedderley, & Prescott, 2023).Attempting to identify specific sensory attributes sceptical consumers expect from seafood products could support product design by elucidating those attributes that should be avoided to reduce the risk of reinforcing negative expectations.The attribute 'sweet' associated differently with 'Nice' scores across herring concepts and was less likely to be expected for the minced format by individuals endorsing seafood as 'Nice'.As such, 'sweetness' should be avoided in minced herring, as it could contribute to disliking among individuals who enjoy seafood.Since these individuals are those with the most frequent and diverse seafood consumption, they are an important consumer group to satisfy.Replicating the finding that FN correlates negatively with food agency highlights the potential for addressing low dietary diversity due to higher FN by attempting to increase cooking skills and perceived behavioural control regarding food (Costa, Niimi, et al., 2023;Niimi et al., 2022).

Limitations and future directions
Firstly, not very many consumers in this study could be categorised as 'diverse'.Although this likely reflects the reality of Swedish seafood consumers at the time of writing, it could impact the reliability of the inferences made here.Relatedly, the measure of seafood diversity used here accounted only for main seafood groups and not specific species.Each of the subcategories of seafood assessed here includes numerous aquatic species, and future measures of seafood diversity should consider this.For instance, separating squid and octopus could have impacted diversity categorisation because squid is more commonly consumed in Sweden than octopus but here both were categorised under 'cephalopods'.Further, some of these groups are difficult to find in Sweden (e.g., echinoderms) and diversification towards these species thus remains a challenge in a Swedish context.
In a similar vein, the consumption frequency categories used here could be refined in the future, for example participants could estimate portions or grams consumed per week/month.Nonetheless, it seems likely that the direction of the effects detected here would remain after such methodological changes, although the relative strength of the tested factors could be impacted.Non-rational factors are known to be a major driver of choices (see behavioural decision theory which acknowledges that choices are not always rational or even purposive, Einhorn & Hogarth, 1981) and of the 4Ns, 'Nice' is almost certainly the least cognitive or rational.Given this, and similar findings for other

Table 5
Conditional associations (means of the posterior distributions and 95% CoIs) for demographic and psychographic characteristics on seafood consumption and expected liking of the two herring concepts.The probability of association in the direction of the mean is given in italics.*Man -woman, negative associations indicate higher scores for women than men.
1 Short version of the Cooking and Food Provisioning Action Scale (Karlsson et al., 2023).

Table 6
Pearson's correlations for expected liking ratings of each sample and consumer characteristics.
animal product categories, we expect that sensory enjoyment would remain the strongest predictor of seafood consumption.Third, the 'Normal' subscale performed sub-optimally in terms of internal reliability as per Cronbach's α, and as such results isolating this subscale should be interpreted with caution.Perhaps researchers could adapt the items in this subscale to specify a reference of 'normality' to participants that is most relevant for their research question.This may limit generalizability and comparability across studies to some extent but could improve inference for specific applications.Finally, it would be beneficial for future research to more comprehensively profile individuals who tend to consider seafood sensorially 'Nice' e.g., to what extent is this associated with factors such as household income and social economic status given that certain seafood species are often perceived as luxury goods as opposed to everyday products.

Conclusions
Consumers motives for eating seafood were quantified using an adapted version of the 4Ns questionnaire and investigated in relation to frequency and diversity of seafood consumption.While it was found that sensorial enjoyment of seafood was associated with both greater consumption frequency and diversity, beliefs that seafood is 'Necessary' for health were only associated with consumption frequency and not diversity.This suggests that communicating the health aspects of seafood may be conducive for increasing consumption frequency but may be less relevant for encouraging diversification in the seafood species consumed.
This study also evaluated consumers' hedonic and sensory expectations towards two herring concepts (traditional pickled against novel minced) in relation to the 4Ns.Consumers expected to like the familiar (pickled herring) more than novel concept (minced herring), although they did not seem to expect to intensely dislike the latter.Some curiosity towards the novel concept thus seems to exist, especially among consumers who currently enjoy the unique sensory profiles of aquatic foods, making this group important to impress upon first tasting.For example, it was noted that seafood enjoyers do not expect 'sweetness' in the minced herring concept (despite expecting this in traditional pickled herring), and therefore perception of sweetness should be avoided in this product.
In addition to the 4Ns, other relevant consumer characteristics such as FN and food agency were examined.As expected, FN was inversely related with seafood consumption, expected liking of both concepts, and food agency.That FN was negatively associated with the 'Nice' subscale, but with no other 4Ns subscale, reinforces that sensory aspects might be critical to increase and diversify seafood consumption, even among more sceptical consumers.All in all, the 4Ns appears to be a relevant tool to uncover motivations for seafood consumption, amongst other animalbased food categories such as meat and dairy, contributing to a better understanding of consumption patterns.
E.S.Collier et al.

Fig. 1 .
Fig.1.Self-reported consumption frequency for each seafood category probed in study 1 and study 2. (*asked only in study 2, **it is likely that respondents in study 1 counted 'small pelagics' in the 'fish' category).

Fig. 2 .
Fig. 2. Posterior distributions for the conditional associations between endorsement of each of the seafood 4Ns and self-reported seafood consumption.
E.S.Collier et al.

Fig. 4 .
Fig. 4. Posterior distributions for expected liking of each herring concept image: traditional pickled and novel minced.
E.S.Collier et al.

Fig. 5 .
Fig. 5. Left: Citation frequency for attributes presented in the CATA task for the images of traditional pickled and alternative minced herring concept.Right: Differences in citation frequency (pickledminced) across the two herring concepts.***p < 0.001.

Fig. 6 .
Fig.6.Median odds ratio of selecting each sensory attribute for each product as a function of endorsement of seafood as 'Nice'.Error bars show 95% CoIs.Higher median odds ratio above 1 indicates higher likelihood of expecting a sensory attribute given an increase in 'Nice'.
E.S.Collier et al.

Table 2
Structure of 4Ns measured within subject for rationalisation of meat and seafood consumption on the first two factors in study 1 for each item under the four subscales (N = 304).
Kock et al., 2022)2023ooking and Food Provisioning Action Scale(Karlsson et al., 2023). 2 Alternative Food Neophobia Scale (deKock et al., 2022).Bayesian analysis of the test-retest reliability of the seafood 4Ns scale: The mean posterior difference provides a point estimate of the most likely difference, and 95% compatibility intervals (CoIs) are shown.The posterior probability of any difference between initial testing and follow-up, a difference of 0.5 pts on the scale, and a difference of 1 pt on the scale in the direction of the mean difference are also provided.
a Initial testfollow-up: negative values indicate higher scores at follow-up.