Nudging interventions on sustainable food consumption: a systematic review

As population growth continues, sustainable food behaviour is essential to help reduce the anthropogenic modification of natural systems, driven by food production and consumption, resulting in environmental and health burdens and impacts. Nudging, a behavioural concept, has potential implications for tackling these issues, encouraging change in individuals’ intentions and decision-making via indirect proposition and reinforcement; however, lack of empirical evidence for effectiveness and the controversial framework for ethical analysis create challenges. This systematic review evaluated the effectiveness of nudging interventions on sustainable food choices, searching five databases to identify the effectiveness of such interventions. Of the 742 identified articles, 14 articles met the eligibility criteria and were included in this review. Overall, the potential of certain nudging interventions for encouraging sustainable food choices were found in strategies that targeted ‘system 1’ thinking (automatic, intuitive and non-conscious, relying on heuristics, mental shortcuts and biases), producing outcomes which were more statistically significant compared to interventions requiring consumer deliberation. Gender, sensory influences, and attractiveness of target dishes were highlighted as pivotal factors in sustainable food choice, hence research that considers these factors in conjunction with nudging interventions is required.


Introduction
Population growth, increased per capita global affluence, urbanisation, increased food productivity and food diversity, decreased seasonal dependence, and food prices have caused major shifts in global dietary and consumption patterns (Lassalette et al., 2014;Tillman and Clark, 2014;Davis et al., 2016). Westernisation of food consumption has occurred in population growth regions over the last 50 years, increased demand for meat and dairy, empty calories and total calories has altered the global nature and nutrient transition scale of food consumption (Kearney, 2010;Tillman and Clark, 2014). The Food and Agriculture Organisation of the United Nations (FAO) suggest that food production will have to increase by 70% to feed an additional 2.3 billion people by 2050, with the majority of this population growth occurring in developing countries (FAO et al., 2020).
The 2019/20 annual global production of cereal grains (2.7 billion tonnes) alone is capable of providing adequate nutritional energy to 10-12 billion people (Cohen, 2017;FAO et al., 2020). However, issues surrounding the allocation and utilisation of cereal grains has led to 43% being used for human food consumption, 36% for animal feed and 21% for other industrial uses such as biofuels. This utilisation can price the most vulnerable people out of the world grain market, limiting food choices, purchases, and human wellbeing (Cohen, 2017). The FAO estimate that 8.9% (690 million) of the global population are undernourished and 9.7% (750 million) are exposed to severe levels of food insecurity (FAO et al., 2020).
Global food production is a significant driver in the anthropogenic modification of natural systems, causing burdens and impacts on both the environment and human health. Externalities including environmental impact (e.g., climate change, biodiversity loss, and natural resource depletion), and negative impacts on human health and culture (e.g., obesity, cancer, diabetes, loss of cultural heritage, impacts on rural businesses, access to green spaces) are generally not included in the price of commodities (Lassalette et al., 2014;Beattie and McGuire, 2016;Benton, 2016;Notarnicola et al., 2017;Schanes et al., 2018;Sustainable Food Trust, 2019;Taghikhah et al., 2019;Viegas and Lins, 2019). Encouraging consumers to adopt more sustainable food behaviour, such as locally sourced foods or diets containing less meat, is essential to reduce the impact of food production and consumption, especially in developed countries (Kerr and Foster, 2011;Schoesler et al., 2014;Hartmann and Siegrist, 2017;Ferrari, et al., 2019;Hedin et al., 2019;FAO, 2019;de Grave et al., 2020).
Sustainable consumption (SC) was first highlighted in the 1992 United Nations Conference on Environment and Development, chapter 4 -Agenda 21 (UNCED, 1992), and defined in the 1994 Oslo Symposium on Sustainable Consumption as: the use of services and related products which respond to basic needs and bring a better quality of life while minimising the use of natural resources and toxic materials as well as emissions of waste and pollutants over the life cycle of the service or product so as not to jeopardise the needs of future generations. (United Nations, 2020, p.8) The 2018 Third International Conference of the Sustainable Consumption Research and Action Initative (SCORAI) in Copenhagen highlighted the role of behavioural economics and related strategies on consumption routines to assist SC (SCORAI, 2018). Hence it is vital to understand human behaviour as complex and influenced by cognitive bias and heuristics (Fischer et al., 2012;Lehner et al., 2016). Kahneman (2011) proposed that human thinking is driven by two systems: • system 1-automatic, intuitive and non-conscious, relying on heuristics, mental shortcuts and biases • system 2-intervening, deliberate and conscious, relying on the availability of information and cognitive capacity to process information to make rational choices Both are susceptible to 'nudges' that encourage behavioural change within civil society (Allcott and Mullainathan, 2010;Kahneman, 2011;Fischer et al., 2012;Marteau, 2017). Richard Thaler and Cass Sunstein first popularised the term 'nudge' in the book Nudge: Improving Decisions About Health, Wealth, and Happiness (2008), in reference to any characteristic of the decision environment "that alters people's behaviour in a predictable way without forbidding any options or significantly changing their economic incentives" (Thaler and Sunstein, 2008, p.6). Sunstein (2013) further suggested that nudges can be promising tools for promoting a broad range of pro-environmental and sustainable consumption behaviours (Sunstein, 2013).
Nudging interventions can play an important role in sustainable food consumption (SFC), helping change consumers food habits in a non-obtrusive, cost-effective manner by modifying the choice architecture in which consumers operate -thus steering their behaviour in preferred directions (Torma et al., 2018;Kácha and Ruggeri, 2019;Vandenbroele, et al., 2019). Hence nudges are the opposite of coercive policy tools which tackle behaviour change through fines or bans (Ferrari, et al., 2019). Blumenthal-Barby and Burroughs (2012) describe the ethical issues surrounding the MINDSPACE framework and identify six principles that can be used to nudge people: defaults (D); ego and commitment (EC); incentives (I); messenger and norms (MN); priming (P); and salience and affect (SA) (BIT, 2010;Blumenthal-Barby and Burroughs, 2012). Descriptive norms, such as incentivising tools for online shopping, can help encourage pro-environmental behaviour and the purchasing of green products (Demarque et al., 2015) whilst social norm interventions, such as those around the use of reusable cups, can help customers avoid wasteful disposable items (Loschelder, et al., 2019). D and P nudges, such as visibility, positioning, display area size and quantity, can shift consumers' purchase behaviour towards more sustainable choices (Coucke, et al., 2019), whereas environmentally friendly food packaging can produce overall positive impacts on consumers' sustainability choices (Ketelsen, et al., 2020).
Nudging is still in its infancy. The UK established the Behavioural Insight Team in 2009 to help develop the concept of nudge units, initiatives and networks, whilst The World Bank, OCED and the EU have supported research to further examine the potential of nudging (Hansen, 2016). Policymakers utilise nudges to help design, implement and evaluate the appropriate policy instruments to assist in devising effective policies to enhance sustainable behaviour and counteract the negative impact of other actors who encourage 'undesirable' behaviours (Lehner et al., 2016;Marteau, 2017). However, nudging has been challenged and criticised on a number of grounds, including the lack of empirical evidence proving its effectiveness, the difficulty in putting theory into practice, and for ethical reasons -i.e. paternalism and reduced human autonomy (Hansen, 2016;Kasperbauer, 2017).
Existing systematic reviews (SR) undertaken on nudging interventions on food choices have mainly focused on human health and diet (Bucher et al., 2016;Wilson et al., 2016;Broers et al., 2017;Taufika et al., 2019;Vecchio and Cavallo, 2019), and the environmental impacts on the supply chain (Ferrari et al., 2019). For example, Ferrari et al. (2019) showed that 'green nudging', especially D, NM, P and SA, has the most significant effect on leveraging more sustainable practices and behaviours of both farmers and consumers, having the potential to be used as tools for environmental policy formulation (Ferrari et al., 2019). Bucher et al. (2016), Broers et al. (2017 and  illustrated how altering placement of food items can produce a moderate significant effect on promoting healthier eating behaviours through healthier food choices. Bucher et al. (2016) further suggested that the strength of the nudge depends on the type of positional manipulation, the magnitude of the change and how far away foods are placed (Bucher, et al., 2016;Broers et al., 2017;.  additionally demonstrated that SA, I and P could increase consumers plant-based choices by 60-65% . Wilson et al. (2016) illustrated that the combination of P and SA enable consistent positive influences on healthier food and beverage choices, making healthier options easier to choose both mentally and physically (Wilson et al., 2016). Furthermore, Taufika et al. (2019) illustrated that the combination of SA and MN could be associated with the reduction of meat consumption (Taufika et al., 2019). Vecchio and Cavallo (2019) suggested that, overall, nudge strategies successfully increased healthy nutritional choices by 15.3% (Vecchio and Cavallo, 2019).
Although these results show that nudges are generally effective in promoting healthier food choices and sustainable practices and behaviours, none of the studies examined the effectiveness of nudging interventions on SFC. There is a knowledge gap on the effectiveness of nudging interventions on sustainable food choices. The goal of this systematic review is to synthesise the empirical findings of existing published academic literature that has investigated the effect of various nudging interventions on these choices and therefore upon SFC in real-life contexts. This paper will: • examine the evidence around the effectiveness of interventions for SFC

Methodology
A search was conducted to identify published literature that utilised interventional and experimental studies to examine nudging interventions to encourage SFC. The studies were identified using the search strategy and analysed against inclusion criteria, those studies that met these criteria were further synthesised by analysing abstracts and full texts. Type of nudges applicable were D, EC, I, MN, P and SA.

Search strategy
This systematic review was conducted in September 2020. The search terms used to identify literature from data sources were: "nudge*" OR "nudging" OR "nudging theory" AND "sustainable* consumption" AND "food" OR "diet" AND "consumer".
Using these search terms, published literature were retrieved from online databases, Web of Science, Scopus, ScienceDirect, EBSCO (Bournemouth University Library) and Google Scholar 2 . The title and abstracts of the retrieved articles were screened for relevance. The potentially relevant articles were examined for their eligibility to be included in the review, whilst the references of the eligible literature were screened to identify any additional eligible literature.

Language and date restrictions
Publication dates were restricted to between 2010-2020 in order that only material released after the publication of Thaler and Sunstein's (2008) techniques was considered. Only literature published in English were included.

Selection criteria
The inclusion criteria for selecting eligible literature were: • Full test peer-reviewed studies in English language • Primary studies between 2010-2020 • Studies should examine the effectiveness or impact of nudges on sustainable food choices • Randomised control trial studies or have a 'control' to ensure empirical evidence • The study should measure sustainable food choices as one of its outcomes via dietary choices i.e., less meat more vegetables 'Grey' literature such as reports and letters were excluded as they were not peer-reviewed.

Selection process
A total of 742 eligible studies were retrieved from the data sources using the aforementioned search strategy, 6 from Web of Science, 338 from ScienceDirect, 9 from EBSCO (Bournemouth University Library), 379 from Google Scholar and 8 from Scopus. After reviewing titles and abstracts, 614 were excluded (Fig 1). The remaining 128 studies were assessed against the inclusion criteria, resulting in exclusion of 102 studies, leaving 26 for further review. 12 articles were further excluded owing to collection of empirical evidence being conducted in a laboratory setting or online surveys, holding the potential for behavioural bias, resulting in 14 articles that were based in a naturally occurring setting i.e., supermarket/canteen. Searching reference lists of the remaining 13 articles, 1 further article was obtained, creating a total of 14 articles for the SR (Fig 1).

Data extraction and synthesis
From the 14 eligible articles basic descriptive data were recorded to ensure quality assessment, including study design, year of data collection, country of residence, target subjects, sample size and intervention setting. More detailed data extraction included: intervention strategy; outcome measured; data analysis method; main findings; and effectiveness of intervention when evaluated against sustainable food choices i.e., less meat more vegetables.
The simple mnemonic MINDSPACE framework was utilised to identify the nine robust nudges that can influence behaviour: messenger; incentive; norms; defaults; salience; priming; affect; commitments; and ego -MINDSPACE (BIT, 2010). For this SR, they have been grouped into six categories -D, EC, I, MN, P and SA (Table 1) (Blumenthal-Barby and Burroughs, 2012). There are many frameworks that help identify key concepts and nudges to influence behaviour towards healthier choices. The TIPPME framework (Typology of Interventions in Proximal Physical Micro-Environments) aims to reliably classify, describe and enable more systematic design, reporting and analysis of interventions in order to help change behaviour across populations utilising nudges D, P, SA to change selection, purchase and consumption of foods (Hollands et al., 2017). Applying EC, MN, SA nudges, the SHIFT framework aims to encourage consumers into pro-environmental behaviours when the message or context influences psycological factors, such as social influence, habit formation, individual self-feeling, cognition and tangibility (White et al., 2019). Chance et al.'s (2014) The 4P's framework aims to integrate nudges within a dual-system model of consumer choice by targeting possibilities, process, persuasion and person, using nudges D, EC, MN, P, SA. Kraak et al. (2017) extend this framework by suggesting a marketing mix and choice architecture 8P's framework, highlighting the potential to promote and socailly normalise healthy food environments. This works by utilising nudges D, EC, I, MN, P, SA encouraging voluntary changes made to the properties of the environment and food being sold (place, profile, portion, pricing, promotion) and volunatry changes made to the placement of food sold (healthy default picks, priming/prompting, proximity (Kraak et al., 2017).

Study quality assessment
To assess the quality of data obtained from the eligible studies a rating scheme was utilised, ranging from weak (*) to very strong (****). The principles of the ratings were based on study design, selection bias, sample size, duration of study, and risk of bias from missing information ( Table 2). The rating scheme was adapted from a previous study undertaken by Nørnberg et al. (2015) who successfully utilised this method to rate and assess the effectiveness of interventions on vegetable intake in a school setting. Table 2: Definition and explanation of study quality assessment (Nørnberg, et al., 2015) Rating Definition Study Description Design & Methods

* Weak
Three of more of the following details are missing: intervention setting, design, duration, RCT or control, statistical analysis.
Design of intervention or statistical methods are flawed.

** Moderate
One or two missing details and satisfactory presentation.
*** Strong One or two missing details and clearly presented.
Large sample size (>100) and/or longer duration (>1 week). Includes any or all of the following: population randomly allocated or matched for intervention or control and validated assessment.

Overall effectiveness of nudging interventions on SFC
The 14 articles included in this SR all focused on SFC in the form of food choice behaviour and were conducted in North America and Europe. Interventions were conducted at supermarkets, canteens, cafeterias, restaurants or cafeterias at universities, workplace, senior activity centres and the Institute Paul Bocuse. The subjects consisted of students, university staff, workplace employees, retirees, and the general population. Five studies used SA as the core nudge, three used a P/SA combination, two used a D/P combination, one used P, one used D, one used D/SA combination and one used I/MN/SA combination. The intervention strategies, intervention duration and sample sizes were largely heterogeneous across all studies (Table 3).
The different strategies and methods applied to implement the varying nudges illustrated differing effectiveness ( Table 4). The studies utilising nudges SA (Gravert and Kurz, 2019;Kurz, 2018) were not statistically significant. One study which utilised P/SA combination (Ohlhausen and Langen, 2020) showed statistical significance with regards to SA but not P, whilst a P/SA combination (Kaljonen et al., 2020) and I/MN/SA combination (Becchetti et al., 2020) illustrated marginal statistical significance, highlighting the potential use of these combinations.

Data quality assessment
Twelve studies were randomised control trials (RCT), duration of interventions varied considerably, ranging from 1 day to 3 years, three studies did not specify the intervention duration. All studies, bar one, had a large sample size (>100) and one lacked sufficient statistical analysis. The quality rating of the included studies was strong to very strong with a mean rating of 3.2 and standard deviation of 1.08 (Table 5).

Sustainable food choices
In total, five of the studies utilised the nudge SA to encourage SFC (Kurz, 2018;Gravert and Kurz, 2019;Salmivaara and Lankoski, 2019;Slapø and Karevold, 2019;Piester et al., 2020). The main strategy utilised consisted of signage, ranging from descriptive menus to visual environmental information. Gravert and Kurz (2019) suggested that introducing two different menus, 1 x meat and fish dishes 1 x vegetarian and fish dishes -meat or vegetarian choices were available upon    request. Meat dish choice decreased by 38% with the vegetarian and fish menu and vegetarian choices increased (3.9%) with the meat and fish menu -no compensatory effect was monitored (p<0.01). Kurz (2018) found that by changing the position of vegetarian dishes in a menu order, and allocating equal visibility of vegetarian dishes with meat dishes in the purchasing environment, purchase of vegetarian dishes increased by 40% (p ≤ 0.01). Weekly sales of vegetarian dishes increased by 0.8%-0.9% after the intervention ceased (p ≤ 0.01) (Kurz, 2018). Slapø and Karevold (2019) found that impementing traffic light labelling (red, yellow, green) on dishes to indicate the environmental friendliness of a dish, encouraged a 7%-9% reduction in meat sales (p=0.10), although having just singular green or red labels had little to no impact (p>0.1). Salmivaara and Lankoski (2019) indicated that activating injunctive norm message signs at point of purchase had no significant effect on sustainable food choice (p=0.6263), whilst Piester et al. (2020) found that implementing sustainability labels on menus marginally influenced women's uptake of more sustainable choices (p=0.11) but not for men (p=0.23). Piester et al. (2020) identified the intention-behaviour gap, highlighting that only 45% of people bought the items they intended to purchase.
Three studies utilised P/SA combination (McBey et al., 2019;Kaljonen et al., 2020;Ohlhausen and Langen, 2020), applying signage with availability and accessibility to help encourage more SFC. Kaljonen et al. (2020) suggested that increasing the availability and accessibility of vegetarian dishes in a buffet line, placing vegetarian dishes at the front, increased sales by 10%. Climate labels attached to the dishes had limited effect, although women were more susceptible (42%). Ohlhausen and Langen (2020) found that DNLs were statistically significant when in combination with a DE (unattractive meal dish) (p≤0.001), while DNLs were 10% more influential than the DE. McBey et al. (2019) proposed that environmental labelling is crucial for framing the disconnection between food choice and the environmental consequence, and the physical layout of retail stores can be a powerful tool in promoting SFC to consumers.
A D/SA combination (Campbell-Arvai et al., 2014) which implemented 'appealing' vegan/vegetarian dishes on a menu assisted significantly with the prediction of food choices made by consumers (p<0.001), and when applied into a default menu (appealing dishes positioned at top) sales increased significantly (p<0.001); however providing information-only menus promoted a decrease in meat-free purchases (p=0.534). Becchetti et al. (2020) utilised a combination of I/MN/SA, implementing three small posters/labels in-store, one promoting environmental responsibility and two labels implementing a 5% and 10% price increase on organic items. Overall, the intervention increased sales by 2% (p≤0.01), with the 5% and 10% labels increasing sales of organic items by 5% and 4.3% respectively, supporting the theory that higher environmental concern can induce the purchase of organic foods, and can induce the purchase of organic food despite its greater cost. Garnett et al. (2019) utilised P, doubling the quantity of vegetarian dishes (25% to 50%) available in three university cafeterias. The intervention increased vegetarian dish uptake by 60.4% across the three cafeterias, positively impacting consumers whom previously had low levels of vegetarian purchases (p≤0.001) with no rebound effect. Zhou et al. (2019) used 'Dish of the Day' as a D intervention, providing statistically insignificant results (p=109-0.865). However, they highlighted that the default dish was chosen when concerns such as security (e.g., safety, harmony, and stability of society, of relationships, and of self) and universalism (e.g., understanding, appreciation, tolerance, and protection, for the welfare of all people and for nature) were strong (p=0.11 and p=0.008 respectively). D/P combination (Coucke et al., 2019;Vandenbroele et al., 2018) provided statistically significant results. Couke et al. (2019) increased sales of poultry by 13% (p<0.05) and decreased sales of other meats by 18% (p=0.001) via enhanced visibility and quantity of poultry available at a butcher's counter. When the intervention ceased, poultry sales decreased significantly (p 0.001). Vandenbroele et al. (2018) illustrated that altering the portion sizes of sausages (150g, 125g, 100g) increased the purchase of 125g and 100g portions marginally (52%), with no compensatory effect on customers purchasing extra portions of the same size (p=0.001). The intervention decreased overall meat (kg) purchased by 13%, however compensatory purchases of other meats did not differ amoung buyers of all portion sizes (p=0.62).

Effectiveness of nudging interventions on SFC
Out of the 14 studies reviewed, ten provided statistically significant results, supporting the positive effectiveness of nudging interventions in encouraging sustainable food choices (Table 6). SA as a nudge Gravert and Kurz, (2019) suggested that the simple and inexpensive rearrangement of menus in terms of convenience could reduced meat consumption by 38% and increase vegetarian and fish dishes sold by 200% (p<0.01). Kurz (2018) further supported this theory by suggesting that increasing visibility and changing menu position could encourage a persistent shift in consumption behaviour (p≤0.01) whereas Löfgren et al. (2012) proposed that experienced participants were harder to nudge than inexperienced participants. The heterogenous effects of the nudge in relation to the type of vegetarian dish served identified that the target dish(es) offered had to be more attractive to meat eaters, hence vegetarian burgers/patties had the most successful impact. With that said, disentanglement of which nudge (visibility/position) caused the vegetarian dish increase was not undertaken.
Conversely, Piester et al. (2020) found the effectiveness of sustainability labels with additional information on a menu was not effective, possibly due to the unknown duration of the intervention and information overload of having messages that combine different types of information (Carfona et al., 2019). Women were more likely to purchase vegetarian dishes with sustainability labels (p=0.11), and with additional information this increased (p=0.23), this is consistent with past research emphasising gender influence in SFC (Andersen and Hyldif, 2015;Zhou, et al., 2019). Piester et al. (2020) highlighted that only 45% of participants purchased what they intended, hence the intention-behaviour gap of consumers is crucial in understand purchasing habits (ElHaffar et al., 2020;Rausch and Kopplin, 2021). Slapø and Karevold (2019) provided marginally significant results utilising traffic light labelling, supporting the theory of the 'compromise effect' (choosing the middle option) (Carroll and Vallen, 2014). Initially there was 7-9% reduction in meat consumption (p=0.10), this behaviour declined over time and almost reverted back to the control period behaviour after period 1; providing evidence that consumers can develop "label fatigue" (p=0.38) (Thorndike et al., 2014). Single red and green labels had no signficant impact (p>0.1), possibly due to lack of available environmental information (Ratner et al., 2008), limited previous knowledge regarding the connection between food choices and environmental consequences (Hartmann and Siegrist, 2017;Lea et al., 2006) and perceived needs of consumers in the choice situation (i.e. focused on sensory factors rather than environmental preservation) (Andersen and Hyldif, 2015;Slapø and Karevold, 2019). Salmivaara and Lankoski (2019) concurred with these results, suggesting that activating injunctive norm messages to promote sustainable food choice is an ineffective measure (p=0.6263), possiblly due to the 1-day intervention duration and exclusion of vegetarian and vegan participants. Nevertheless, this intervention could help identify potential subgroups of consumers who are sensitive to the intervention, i.e. older educated women influenced more by the message of "ecological wellbeing". Multiple norms could have complex casual interactions and joint effects, i.e. messages of ecological wellbeing and local food could be combined to have greater impact than the message used independently; a topic requiring further attention (McDonald et al., 2014). Ohlhausen and Langen (2020) were able to identify the SA nudge (DNLs) as 10% more effective in increasing vegetarian dish choice (p≤0.001), especially the use of "sustainability" and "regional" (20% and 15% respectively), proving consistent with past research (Morizet et al., 2012). Whereas, in constrast to previous non-food related literature, the P nudge (DE) lowered choice frequencies of sustainable choices overall (p=0.23) (Simonson, 1989;Doyle et al., 1999;Masicampo and Baumeister, 2008). This study supports Kurz (2018) theory that nudging interventions are not only influenced by the type of nudge or setting but by other variables (i.e. target dish), hence based on systematic assessment of similarities and difference between dishes, careful selection and grouping of target dishes and competitor dishes is required (Ohlhausen and Langen, 2020).

P/SA combination as a nudge
Both Kaljonen et al. (2020) and McBey et al. (2019) undertook qualitative studies that used descriptive labels as the SA nudge. Kaljonen et al. (2020) suggested that climate labels are a restriction to menu and recipe development, whilst McBey et al. (2019) suggested that how descriptive messages are framed is crucial, i.e. comparing meat products with sources of environmental pollution. Kaljonen et al. (2020) further suggested that availability and accessibility, by changing the food order available in a buffet line (P nudge), helps to encourage more vegetarian dish choices (+10%). Coinciding with McBey et al. (2019) who suggested that the physical layout of supermarkets play a pivotal role in highlighting the 'otherness' of alternative food choices (i.e. plant-based), creating a 'not for me' implication. Both studies agreed with past research that more qualitative research is required in understanding SFC (Lehner et al., 2016), the complex and multi-faceted nature of food choice means that what holds true in controlled conditions may not work in every day life (Kahneman, 2011).

D/SA combination as a nudge
Campbell-Arvai et al.'s (2014) D/SA combination suggested that by placing less environmentally-friendly food choices in slightly less convenient positions on a menu (i.e. bottom) the default menus increased the probablility of consumers choosing a meat-free dish (p≤0.001). This was consistent with other research (Downs et al., 2009;Just and Wansink, 2009). The attractiveness of menu dishes had a significant influence on food choice enabling prediction of the choice (p≤0.001), whereas the presence of information on a default menu provided statistically insignificant interactions (p=0.534). Additional information is less effective at motivating behaviour change at an individual-scale and with real time choices due to immediate or intuitive factors that dominate decisions, especially when time pressure and distractions conspire to prevent personal deliberation (Shiv and Fedorikhin, 1999;Ariely and Loewenstein, 2006 (McDaniel et al., 2001;Feldman et al., 2011). The unappealing nature of the veggie balls could have resulted from a lack of detailed information accompanying the dish and the equality it was given amongst the other two dishes, lowering participants' attention to the default dish. Females from the UK and Denmark were more likely to choose the D target dish, especially when more importance was given to sensory factors and universalism (p=0.042 and p=0.033), supporting the view that peoples' concern about nature could be effective for SFC (Worsley et al., 2016). Zhou et al. (2019) highlighted that default-based interventions can be important tools in motivating pro-environmental behaviour and serve to complement information and educational efforts over the long-term. However, this could be seen as underhanded and choice constraining, limiting freedom and autonomy of decison makers.
P as a nudge P as a nudge has the potential to encourage SFC, it is a relatively cheap and easily implemented strategy that generally goes unnoticed by consumers. Garnett et al. (2019) highlighted that meal selection is neither fixed nor random but rather partially determined by availability. By increasing the proportion of vegetarian choice uptake significantly increased, reflecting past research (Holloway et al., 2012;Lombardini and Lankoski, 2013;. The greatest impact was measured amongst participants who were least likely to chose vegetarian dishes before the intervention (p≤0.001), corresponding with Scarborough's findings (2014).

D/P combination as a nudge
Both Coucke et al. (2019) and Vandenbroele et al. (2018) provided statistically significant results for encouraging sustainable food choice (p 0.05 and p=0.001 respectively), however the studies lacked information on either sample size or duration. Vandenbroele et al. (2018) suggested that nudging consumers at point of purchase, rather than at moment of consumption, led to a 13% reduction in meat (kg) purchased and helped to change consumers purchase behaviour, concurring with previous research (Arno and Thomas, 2016;Vermeer et al., 2010). Coucke et al. (2019) supported this theory by suggesting that increasing the display size and quantity of more sustainable meat products (poultry), increased sustainable choices (+13%). When the intervention was removed sales of the sustainable meat product decreased, highlighting that visual cues can have an impact on consumers behaviour (Van Kleef et al., 2012;Wilson et al., 2016;Helmefalk and Berndt, 2018). Overall, D/P combination is an effective nudge for promoting and encouraging consumers to change their behaviour to more SFC practices.

Conclusion
Overall, this review has established the potential of certain nudging interventions for encouraging sustainable food choices and SFC. Strategies that required little involvement (system 1) from consumers, produced higher statistically significant outcomes compared to nudging interventions which required more deliberation (system 2). Gender, sensory factors, attractiveness, and type of target dish played a pivotal role in encouraging sustainable food choices. Females were influenced by interventions significantly more than males. Proximity, placement, and information encouraged consumers to adopt more sustainable food choices and the overall presentation, portion size and choice of sustainable alternatives played a key role in encouraging consumers into SFC. Successful nudges included P, D/P combination, SA, D/SA combination and I/MN/SA combination. These five nudges utilised intervention strategies that enhancing availability and accessibility, promoted consumers environmental responsibility, altered portions sizes, offered food alternatives upon request, and targeted appealing dishes in combination with a default menu. Five studies that utilised D, SA combination and P/SA combination all provided insignificant results. Interventions such as 'Dish of the Day', activating injuctive norms and sustainability labels, with additional information, proved ineffective tools for promoting sustainable food choices. The effectiveness of nudging is optimal when utilsied together with information campaigns, economic incentives and education, and hindered by factors including bias, intentionbehaviour gap and external influences such as social norms, environmental determinants and financial status (Broers et al., 2017;Taufika et al., 2019).
This SR had several limitations. The search terms "nudges, nudging or nudge theory" may have lead to many undetected studies being left out, as well as "behavioural interventions" not being included in the search strategy may have limited the outcome. The studies were mainly heterogeneous with different interventions measured. Participants were mainly students or staff and the intervention settings were primarily universities, restricting greater external validity. All of the studies were undertaken in developed and highly westernised countries, hence further research should be undertaken in developing countries to allow for better understanding of the effectiveness of nudging interventions. Only English papers were eligible, hence a possibility of missing important relevant studies in other languages. Furthermore, this SR has been conducted by a single reviewer which could potentially cause bias on screening, rating and synthesis of the studies.
All of the studies, bar one, focused on short-term effectiveness of nudging and thus more research should be undertaken to understand if nudging is effective in the long-term. Further research regarding gender, sensory influences, dish attractiveness, multiple norms, intention-behaviour gap and tinkering could be addressed in conjuction with nudging interventions to better understand how more sustainable eating can be achieved in real-life situations, strengthening evidence and knowledege of how nudging might encourage SFC.
Further qualitative research should also be undertaken to enable greater understanding of what occurs in non-controlled environments. Ethical consideration of nudging and transparency is required in any future use of the technique in order to address the issue of freedom or autonomy in decision-making.
The number of people that can be supported within planetary boundaries in part depends on their choices (Cohen, 2017). The massive environmental impact of agriculture and the food industry mean that food choices will become of increasing importance. People are at the centre of sustainable development and with global population projected to increase to 9.7 billion by 2050 (United Nations, 2019), individual and collective human choices coupled with environmentally sustainable practices will be key drivers to enable a sustainable expansion in food production (Cohen, 2017). Nudging may play a role in changing behaviour toward habits of sustainable food consumption.