From tablet to table: How augmented reality influences food desirability

Augmented reality (AR) technology transforms a user’s visual experience of the physical world in real-time, by allowing the user to, “see the real world, with virtual objects superimposed upon or composited with the real world,” (Azuma, 1997). While precise definitions of AR technology vary (Höllerer & Feiner, 2004; Liao, 2016, 2019), most scholars agree that AR is fundamentally characterized by one integral component—real-time superimposition. In current AR applications, real-time superimposition is typically achieved by using a mobile device’s camera to visually recognize one’s immediate environment, on which digitally rendered images are instantaneously overlaid (Athsani & Kalaboukis 2012; Yim et al., 2017; Oh & Bailenson, 2017). Today, consumers are able to engage with AR technology through various devices (including stationary computers and headsets), but the overwhelming majority access AR applications on mobile devices (e.g., smartphones or tablets; Tankovska, 2020). Such applications typically use either the device’s front-facing camera (to project visual content such as make-up or personal accessories onto a user’s face) or the device’s rear-facing camera (to project visual content into the user’s current space). Regardless of the format, the technology is typically used to complement, supplement, or enhance the surrounding physical world with added visual information.

Although sometimes used interchangeably (albeit incorrectly), it should be noted that AR is both theoretically and practically distinct from Virtual Reality (VR). VR is a technology which entirely immerses the user in an artificial, virtually simulated world (Schmitt, 2019; Tan et al., 2022). One way in which we conceptualize the distinction between AR and VR is by understanding the primary function of each technology: superimposition and transportation, respectively. Superimposition, one of the primary functions of AR, is the process of placing, or laying, something on top of directly viewed real-world scenes so that the two coexist and are both still evident (Milgram et al., 1994). This process utilizes the real-world surroundings as a natural backdrop upon which virtual elements are overlaid, providing the viewer with the advantage of visualizing virtual objects “without the vulnerability of being blind to the real world” (Tan et al., 2022), as is the case with VR. On the other hand, transportation, one of the primary functions of VR, removes many real-world sensations (e.g., obstructing the view of the real world via a head-mounted display) and “transports” individuals to another place that may or may not exist in reality (Sadowski & Stanney, 2002; Tan et al., 2022). Put differently, AR leverages technology to visually alter the user’s immediate environment, whereas VR leverages technology to visually remove the user from their immediate environment and transport them to a different, completely synthetic environment, and these different processes will likely result in distinct user outcomes (see Hilken et al., 2022b for an empirical comparison of AR and VR in an experiential retail setting).

Some scholarly work in marketing has begun to explore the psychological implications of AR-enabled visualizations. In Table 1 below, we highlight a selection of papers that have empirically examined the effect of AR on consumer responses, making particular note of the specific manipulations employed and the contexts in which AR was applied. Notably, the bulk of such work largely falls into two categories. The first body of work examines how specific configurations and settings within AR applications influence consumer responses. For example, scholars have examined aspects including the level of customization (e.g., the ability to personalize the visual content; Carrozzi et al., 2019), degree of interactivity (e.g., the ability to manually manipulate or transform the content; Heller et al., 2019a), sensory modality of control (e.g., touch vs. voice, Heller et al., 2019b), exposure time (e.g., how long the user spends on the app; Hopp & Gangadharbatla, 2016) and product composition (e.g., bundled vs. individual, Hilken et al., 2022a).

The second category of work has almost exclusively focused on “virtual try-on” experiences, which typically involve a front-facing camera that superimposes clothing, accessories, or makeup onto the users themselves (Hilken et al., 2017; Poushneh & Vasquez-Parraga, 2017; Smink et al., 2019; Tan et al., 2022; Yim et al., 2017). Collectively, such work has found that AR can improve consumer responses, including brand attitudes and purchase intentions. While this body of work does involve a form of visual superimposition, it represents a practically and theoretically unique area of investigation, given that such applications superimpose digital content onto the users themselves (as opposed to superimposing content into the user’s environment/space). Thus, AR presentation in such contexts necessarily introduces the confound of simultaneously providing a visual portrayal of the user themself (a factor that previous research has found to have significant effects on consumer attention and attitudes—e.g., Chang and Hung 2018; Cho and Schwarz 2010; Hung and Wyer 2011).

Our research complements and builds upon the existing work in at least two ways. First, as opposed to exploring specific configurations or settings within AR applications (e.g., level of customization or degree of interactivity), we hold such aspects constant, instead empirically isolating and focusing on AR’s fundamental ability to superimpose digital objects onto a consumer’s real-time environment. In other words, we explore whether and how such superimposition of visual content can, in and of itself, influence consumers’ evaluations and purchase likelihood.² Second, we complement extant work by exploring and identifying the underlying psychological mechanisms that might explain such effects. To provide support for our theorizing, we next turn to research on mental simulation.

We focus our empirical investigation within the food domain for both substantive and theoretical reasons, as previously mentioned. Notably, academic research has found that judgments and decision-making with respect to food are often influenced by a consumer’s propensity to engage in mental simulation, particularly by a consumer’s tendency to imagine consuming a designated food (Elder & Krishna, 2012; Hildebrand et al., 2019; Kappes & Morewedge 2016). Neuroscience research has shown that mental simulation can engage parts of the brain associated with tasting, smelling, and hearing stimuli (Krishna, 2012; Schifferstein, 2009), and several consumer researchers have accordingly shown that when consumers mentally simulate (i.e., imagine) consuming a food item, it increases their immediate desire for it (Elder & Krishna, 2012; Hildebrand et al., 2019). While mental simulation has also been shown to improve responses to non-food product categories, research suggests it is most beneficial for highly sensorial and hedonic (as opposed to utilitarian) product attributes (MacInnis & Price, 1987).

Mental simulation can be induced from all sensory modalities, but visual images tend to be perceived the most vividly and therefore are the method most commonly used to induce mental simulation (Schifferstein, 2009). This may have evolutionary roots, as our sense of sight at least partially developed in order to increase our species’ chances of survival by identifying the most nutrient and energy-rich sources of food (Spence et al., 2016). Yet importantly, not all visual imagery is equally likely to induce mental simulation. For example, researchers have found differences across high quality versus low quality pictures (Petrova & Cialdini, 2005; Rossiter & Percy, 1980) and dynamic versus static images (Lutz & Lutz, 1977, 1978; Roggeveen et al., 2015; Schlosser, 2003). Most relevant to the current research is work finding that contextual visual cues can also play an influential role in inducing mental simulation. For example, Hildebrand et al. (2019) demonstrate that displaying foods with an occasion-setting background (e.g., depicting a pizza over a depiction of a pizzeria vs. a solid or incongruent background) can increase mental simulation tendencies, especially for holistic thinkers. Similarly, Elder and Krishna (2012) show that subtly portraying food in a manner more fluent with consumption (e.g., visually placing a fork on the same side as the viewer’s dominant hand) can similarly induce greater mental simulation of consumption.

Given that AR has the ability through superimposition to create vivid illusions of a product’s presence in a user’s immediate real-world environment, we expect AR technology has a high potential to induce mental simulation in consumers. In fact, at the conclusion of their meta-analysis on mental simulation, Ceylan et al. (2022) speculate that augmented imagery might facilitate mental simulation, and explicitly invite future research to explore this phenomenon.

Notably, while much research has treated mental simulation as a unidimensional construct, some work has added nuance to the mental simulation literature by distinguishing between two distinct types of mental simulation, each of which have been shown to influence consumers’ judgments and behavior: process-focused simulations and outcome-focused simulations, respectively (Castaño et al., 2008; Escalas & Luce, 2004; Ringler et al., 2021; Taylor et al., 1998; Zhao et al., 2011). Process-focused simulations, or “how-thinking,” makes salient the process of engaging in an activity through the use of sensory cues to influence product evaluations and behavioral outcomes, including willingness to pay, purchase intention, goal completion, and consumption (Castaño et al., 2008; Escalas & Luce, 2003; Ringler et al., 2021; Taylor et al., 1998). Conversely, outcome-focused simulations, or “why-thinking,” makes salient the outcome from engaging in an activity, without consideration for how that outcome was achieved (Castaño et al., 2008; Escalas & Luce, 2003, 2004). We expect that the mental simulation facilitated through AR presentation will most closely resemble process-focused simulation rather than outcome-focused simulation, as the superimposition facilitated by AR provides users with the opportunity to imagine “engaging” with, or consuming, the food item that has been projected in their immediate space, as opposed to simply imagining a post-consumption outcome (e.g., feeling full, satiated, or any other type of consumption consequence).

As previously discussed, much of the recent marketing research on augmented reality has focused on how consumers interact with AR applications (Tan et al., 2022). These explorations have included manipulations of customizability, interactivity, modality of control, and exposure time (Carrozzi et al., 2019; Heller et al., 2019a, 2019b; Hilken et al., 2020; Hopp & Gangadharbatla, 2016). While such research has provided valuable insights into the optimal configuration of AR, our work instead focuses more fundamentally on how one critical aspect of AR technology itself—the superimposition of digital stimuli onto consumers’ real-time environment—can influence behavioral responses. Motivated by both managerial prevalence as well as prior research establishing the link between visual contextual cues and increased mental simulation in the context of food consumption (e.g., Elder & Krishna, 2012; Hildebrand et al., 2019), we focus our investigation on the effects of AR-induced mental simulation in this consequential and highly-sensory domain. This focus addresses the recent call by Tan et al. (2022) to explore how AR can most effectively be leveraged by the service and hospitality sectors and how this technology may influence consumers’ judgments and decisions.

Given that AR has a unique ability to visually superimpose digital objects on top of one’s real-time visual environment, together with the knowledge that visual contextual cues often serve as critical antecedents of mental simulation (e.g., Elder & Krishna, 2012; Hildebrand et al., 2019), which itself has been shown to increase food desirability and purchase intentions (particularly within highly sensorial product categories; MacInnis & Price, 1987), we formally hypothesize:

While the effects of mental simulation on downstream variables have been well-documented (albeit not in the AR context) in extant literature (Ceylan et al., 2022; Elder & Krishna, 2012; Hildebrand et al., 2019; MacInnis & Price, 1987), little research thus far has explored the link between AR technology and mental simulation. Therefore, it is compelling to consider what specifically about AR might increase consumers’ mental simulation in the first place. One possible explanation comes from AR’s unique ability to visually superimpose a virtual object onto a consumer’s peripersonal space (the immediate space around one’s body which can be touched or manipulated, Holmes & Spence, 2004). Objects which appear in one’s peripersonal space are likely to be perceived as personally relevant, since people tend to surround themselves with objects they enjoy and that are of personal relevance, rather than objects which are not personally relevant. Therefore, it is plausible that by superimposing an object into one’s peripersonal space via AR, that object should be perceived as more personally relevant to the individual, compared to the same object that is not superimposed into one’s peripersonal space. Further, perceived personal relevance of an object or advertisement has been previously shown to increase both message processing (Ajzen et al., 1996) and mental simulation (Buckner et al., 2008; Gutsell & Inzlicht, 2010; Ülkümen & Thomas, 2013). Notably, Ülkümen and Thomas (2013) specifically demonstrate that messages framed with high personal relevance (versus low personal relevance) led participants to spontaneously simulate the action of the message (i.e., process-oriented mental simulation). Stringing these findings together, we argue that the superimposition of a virtual object into a users’ peripersonal space should increase the perceived personal relevance of that object, which in turn will increase the ease of mentally simulating engaging with the superimposed object. Formally, we hypothesize:

Our theorizing is collectively illustrated in Fig. 1 below. Notably, while AR can be presented in a variety of forms, we focus the current investigation on its manifestation through mobile devices, given their ubiquity and dominance as the primary form factor for consumer-facing AR applications (Tankovska, 2020). Behavioral research in marketing has explored how mobile devices can systematically influence consumers’ attitudes and behaviors (e.g., Bart et al., 2014; Grewal & Stephen, 2019; Melumad et al., 2019, Melumad & Meyer, 2020; Song and Sela, 2022), and some work has focused on how specific features of mobile devices play a role. For example, scholars have demonstrated that the touchscreen feature on these devices can influence both psychological ownership and consumer choices (Brasel & Gips, 2014; Shen et al., 2016). Other work by Hadi and Valenzuela (2020) has shown device-delivered haptic feedback can improve consumer responses to communications. Diehl et al. (2016) explored how the camera function on mobile phones allows consumers to increase their enjoyment experiences through increased photo-taking. Our research expands on this prior work by exploring how another unique capability of mobile devices, AR presentation using the integrated camera, can alter and drive consumer responses.

We test our conceptual model (Fig. 1) and hypotheses across four experimental studies,³ which collectively provide empirical support for our proposed theorizing across both large (e.g., tablets; Studies 1–3) and small (e.g., smartphones; Study 4) mobile devices, in the context of indulgent (i.e., dessert in Studies 1 and 2), non-indulgent (Study 3), and both desirable and undesirable (Study 4) food categories. Notably, the first two studies were conducted in field settings to lend external validity to our investigation, while the lab-based nature of the subsequent two studies allowed us to achieve rigor in our measurement of psychological processes. Study 1, a field experiment run at a restaurant, demonstrates that presenting indulgent foods in AR (versus a non-superimposed format) can increase both desirability and consequential downstream variables (i.e., real purchase), supporting H1. In Study 2, a field experiment run at a café, we replicate the positive effect of AR presentation and provide preliminary support for mental simulation as the underlying process, hence supporting H2. In Study 3, conducted in a behavioral lab, we replicate the effect of AR presentation on desirability and purchase likelihood of a non-indulgent food item, use multi-item measures to more robustly support the mediational role of mental simulation, and additionally find support for personal relevance as an anteceding mediator (supporting H3, and fully testing the model illustrated in Fig. 1). In the final study (conducted in a behavioral lab), we find converging support for our overarching model (Fig. 1) using a smaller and more accessible mobile device (i.e., smartphone), extend the generalizability of our findings (by demonstrating that the results hold for both desirable and undesirable food items), and more precisely identify the type of mental simulation at play (i.e., process-oriented mental simulation).

One hundred and one diners⁴ (41% female, 59% male, 0% nonbinary/other; M_Age = 37.95, SD = 13.04) participated in exchange for a £5 discount on their restaurant bill. The experiment used a two-level (presentation format: control versus AR) between-subjects design. We ran this experiment in collaboration with a brasserie-style restaurant in a large international city (see Appendix A for photographs of the restaurant). To manipulate presentation format, we created two versions of the restaurant’s existing paper-based dessert menu (six items, see Appendix B for a list of all options on the dessert menu). For the control condition, we took high-resolution photographs of each dessert, and presented these as a digital menu on a tablet (an Apple iPad) given to diners. For the AR condition, we used a professional 3D scanning photogrammetry kit to take over 400 high-resolution photographs of each dessert from various angles. These images were then given to a professional AR developer (QReal) for conversion into 3D renderings of each dessert that were viewable as AR objects using a mobile app installed on the tablet.

Importantly, while both menu formats were electronic and viewed in a mobile app on a tablet, the control menu displayed two-dimensional desserts on static blank backgrounds, whereas the AR menu displayed three-dimensional desserts superimposed in a diner’s environment (i.e., on the restaurant table in front of them). The two presentation formats are illustrated in Appendix C (while the AR condition included a manipulation of both dimensionality and superimposition in this study, our next two studies attempt to separate these factors).

The experiment was run on two consecutive weekday evenings during the dinner shift (6 pm – 10:30 pm). After restaurant diners completed the main course of their meal, an experimenter blind to the hypothesis and posing as a waiter approached their table with a tablet and informed them that the restaurant had created a “digital dessert menu” and was offering diners a discount off their bill for simply looking at the menu and providing feedback (without any obligation to order anything if they did not wish to).

Each table was randomly assigned to view the dessert menu in either the control or AR format on an alternating basis (we adopted this procedure as opposed to random assignment at the individual level to prevent diners at the same table from becoming aware of the manipulation). While viewing the menu, participants completed a paper survey in which they assessed the desirability of each dessert on a 7-point scale (1 = “very undesirable” to 7 = “very desirable”). After placing their dessert orders (if any) with the experimenter, participants were asked their age, gender, and familiarity with augmented reality technology (on a 7-point scale, 1 = “very unfamiliar” to 7 = “very familiar; for complete list of all measures used in Study 1, see Web Appendix B). Afterwards, all diners were served any desserts they ordered. At the end of the meal, the experimenter collected the corresponding receipts for each table, which indicated all food items and beverages ordered, how much money was spent, and the number of people at each table.

One hundred and thirty participants (composed of business school students and staff) participated in this study in exchange for a free dessert. We used a two-level (presentation format: control versus AR) between-subjects design. For this study, we worked closely with the university catering team that wanted to showcase three new dessert items developed by their chef for the café (see Appendix D for pictures of these desserts). We again used a professional 3D scanning photography kit to take over 400 high resolution photographs of each dessert from various angles and worked with the same AR developer to convert these photographs into 3D renderings. To manipulate presentation format in this study, we again created two versions of a digital dessert menu. This time, both conditions featured food items which were visually identical in every possible way (e.g., in scale, resolution, dimensionality, etc.). In the control condition, the menu featured the 3D renderings of each dessert over a static blank background. In the AR presentation, as in Study 1, the same 3D renderings for the control condition were superimposed onto the viewer’s real-time environment using the tablet’s camera. Accordingly, the only difference across conditions was the background behind the dessert item: the dessert was featured on a static background for the control condition, or the dessert was visually superimposed in real-time in the AR condition. In both conditions, participants were equally able to interact with the stimuli by using their fingers on the touchscreen of the tablet to rotate, reposition, and resize the featured foods as they desired. The two presentation formats are illustrated in Appendix E.

The experiment was run in the café on a weekday afternoon during lunch hours. An experimenter blind to the hypothesis approached students and staff who were seated and appeared to have just finished having lunch and informed them that the catering team was testing new desserts. Participants were offered a free dessert in exchange for providing feedback on a brief paper survey about the new menu items. Participants were then randomly shown either the control or AR digital dessert menus, done on an alternating basis. After viewing the menu, participants were asked to choose the item they would like to receive. They then indicated the desirability of the dessert (“I am craving the dessert I chose”) and responded to a one-item measure of mental simulation (“When viewing the dessert, I could imagine myself eating it,” taken from Elder and Krishna, 2012; we use more comprehensive scales in Study 3), both measured on a 7-point scales (1 = “strongly disagree” to 7 = “strongly agree”). After completing these items, participants were served the dessert they had selected. After consumption, they indicated how much they enjoyed the dessert (“I enjoyed the dessert, measured on a 7-point Likert scale 1 = “strongly disagree” to 7 = “strongly agree”; see Web Appendix C for complete list of all measures).

One hundred and eight volunteers from a university setting (46% female, 54% male, 0% nonbinary/other; M_age = 29.73, SD = 6.61) participated in this experiment in exchange for monetary compensation. The study employed a 2 cell (presentation format: control versus AR) between-subjects design. Presentation format was manipulated as in Study 2 (keeping scale, resolution, dimensionality, and interactivity constant): the control condition featured the 3D food item over a static blank background, while the AR condition superimposed the 3D food item onto the viewer’s real-time environment), except this time the target stimuli was a lamb shawarma (see Appendix F for videos and photographs of the stimuli).

This study was conducted during the COVID-19 pandemic in a university that was allowing students to attend classes in person with appropriate protection measures and social distancing in place. To comply with local COVID-19 government regulations, we modified a lecture theater to serve as a laboratory (see Web Appendix E for complete list of precautions taken to ensure participant safety and regulatory compliance). Once participants arrived at the lab, they scanned a QR code with their mobile phone to access a mobile survey that included the consent form and survey questions. Participants were told that they would view and evaluate lamb shawarma on a tablet, but before viewing the item, they were asked to indicate their prior experience with the food (“I have eaten lamb shawarma before,” with the options, “yes,” “no,” or “unsure”). Then, all participants were given a tablet to view the lamb shawarma in either the control or AR presentation format, according to their randomly assigned condition. Importantly, due to social distancing requirements, participants were sat far enough apart as to not be able to interact or engage with one another.

Participants responded to survey questions on their mobile phones while continuing to view the lamb shawarma on the tablet (should they wish to). To take advantage of the laboratory setting, participants were asked to respond to a battery of measures (see Web Appendix F for a full list of items). Our dependent variables of interest in this study were desirability (3-item scale including those from Studies 1 and 2; α = 0.89) and purchase likelihood (“After viewing the lamb shawarma, how likely would you be to order it if it was offered on a menu?”). Our proposed mediators were personal relevance (2-item index; e.g., “This is similar to other foods I eat”; r = 0.56) and mental simulation (6-item scale adapted from Hildebrand et al., 2019; e.g., “I could imagine myself eating the lamb shawarma displayed”; α = 0.89). In addition, we measured potential alternative process explanations including: presence (3 items adapted from Slater et al., 1994; e.g., “The lamb shawarma felt like it was on the table in front of me”; α = 0.68), attention to background (“I paid more attention to the lamb shawarma than I did the background behind it,”), perceptual fluency (2 items adapted from Labroo et al., 2008, e.g., “It was easy for me to evaluate this food item”; r = 0.85), psychological ownership (2 items adapted from Atasoy & Morewedge, 2018; e.g., “I felt like the lamb shawarma was already mine”; r = 0.65), spatial and temporal distance (items adapted from Elder et al., 2017; spatial: “How close do you think the restaurant offering the lamb shawarma is located?”or temporal: “How quickly do you think the restaurant could deliver this lamb shawarma to you?”), enjoyment of the experience (2 items, e.g., “It was fun to view this item”; r = 0.77), and realism (4 items, e.g., “This lamb shawarma looks real”; α = 0.86). Finally, participants indicated their mood (2 items, e.g., “I am in a good mood right now”; r = 0.67), familiarity with AR Technology (as measured in Study 1), gender and age.

One hundred and seventy-three volunteers (44% female, 0% Non-binary/other; M_Age = 29.87, SD = 7.00) from a university setting participated in this experiment in exchange for monetary compensation. This study employed a 2 (presentation format: control vs. AR) × 2 (food item: undesirable vs. desirable) between-subjects design. Presentation format was manipulated as in Study 2 and 3 (keeping scale, resolution, dimensionality, and interactivity constant): the control condition featured the 3D food item over a static blank background, while the AR condition superimposed the 3D food item onto the viewer’s real-time environment. To manipulate food-item desirability, participants either viewed fermented trout (determined as “undesirable” in the pretest reported in Web Appendix H), or parmesan fries (determined as “desirable” in the pretest reported in Web Appendix H). Images of the stimuli are presented in Appendix G.

Participants responded to survey questions on their mobile phones while continuing to view the food item on the mobile phone provided (should they wish to). Our dependent variable of interest in this study was product evaluation (using the 4-item scale from Study 3; α = 0.96). In an attempt to measure and replicate the sequential mediation observed in Study 3, participants responded to the same personal relevance (r = 0.72), and mental simulation (α = 0.91) measures that were employed in Study 3. To further investigate the type of mental simulation participants might engage in, they were also asked to respond to a 2-item index measuring process-oriented simulation (2-item measure adapted from Castaño et al., 2008; e.g., “I thought about how I would eat this food item,” “I thought about the process of eating this food item”; r = 0.72), and a 2-item index measuring outcome-oriented simulation (2-item measure adapted from Castaño et al., 2008; e.g., “I thought about why I would eat this food item,” “I thought about the benefits I would gain from eating this food item”; r = 0.34). Participants indicated the realism of the stimuli (α = 0.85), their overall mood (r = 0.74), and their familiarity with AR Technology, all measured as in Study 3. Finally, they indicated their gender and age (see Web Appendix I for complete list of measures).

The current work addresses recent calls to explore how AR affects sensory perceptions, decision-making, attitude formation, and pre/post purchase behavior and evaluations (Tan et al., 2022) and how new immersive consumer technologies—including AR—can affect consumers’ mental simulation and imagery generation (Ceylan et al., 2022). While existing literature on AR has advanced our understanding of what optimal settings within AR applications might look like, we empirically isolate what we consider the core, unique facet of AR presentation: real-time superimposition of visual objects in one’s environment. Specifically, we demonstrate that due to its ability to visually superimpose products onto a consumer’s real-time environment, AR presentation uniquely able to generate perceived of personal relevance and mental simulation above and beyond visual stimuli that is not superimposed, contributing to this literature by both identifying and measuring the psychological processes driving the positive effect on desirability and purchase likelihood.

We also add to a growing body of behavioral research in marketing that explores how technological features in mobile devices can systematically influence the way consumers process information and behave. While previous scholars have examined the touchscreen (Brasel & Gips, 2014; Shen et al., 2016), haptic feedback (Hadi & Valenzuela, 2020) and photo-taking (Diehl et al., 2016) functionality of mobile devices, the current research demonstrates that AR superimposition using a mobile device’s camera can also alter consumers’ perceptions and influence real-world behaviors.

As mentioned above, our documentation of the positive effect of AR presentation on consumer responses to foods appears to be relatively robust (generalizable across different device sizes, food categories and consumption contexts). However, given our delineation of the underlying process, we can make some logically informed inferences about the generalizability of our findings apply beyond the food domain. Namely, since our research demonstrates mental simulation is one of the critical mechanisms through which AR exerts its effects on consumer responses, positive consumer responses should theoretically manifest in other contexts where mental simulation is considered advantageous. Given that previous research suggests mental simulation is most beneficial for highly sensorial/hedonic (as opposed to utilitarian) product attributes (MacInnis & Price, 1987) and our finding that the specific mental simulation at work is likely process-focused (as opposed to outcome-focused) in nature, AR presentation should theoretically improve responses when consumers are focused on hedonic attributes that arise while using a product. For example, if a consumer is shopping for a kitchen appliance, AR presentation might improve the product’s desirability if the consumer is focused on how fun it would be to use the appliance, as opposed to being focused on a utilitarian outcome (e.g., the result of using the appliance). We expand on these potential extensions in the “future research” section further below.

As mentioned in the introduction to this paper, some food and beverages establishments brands have begun experimenting with AR applications, but it is far from being an ubiquitous practice. Our findings suggest that offering consumers the ability to view products in an AR format may be a worthwhile investment, and this can be implemented in a number of ways. In the context of the food/restaurant industry that we have focused on, many dining establishments (e.g., restaurants, cafeterias, bars) already offer patrons digital menus via handheld tablets (Anindita, 2018). These digital menus can easily be upgraded with AR-enabled renderings of the menu items, allowing patrons to visually preview dish on their table before placing an order (similar to the procedure we used in Studies 1 and 2). Catering firms and bakeries that provide custom offerings (e.g., personalized cakes) can use AR technology to facilitate potential customers’ ability to visualize what yet-to-be-created products will look like, before going through any irreversible production process. Our research suggests that such efforts should make the viewed food items more desirable and lead to increased purchase likelihood. Further, given our finding that AR previewing improves consumers’ evaluation of the consumption experience itself, it is likely to increase customer satisfaction (itself a critical determinant of marketplace behaviors such as repurchase, recommendation, and willingness to pay; Anderson & Sullivan, 1993; Homburg et al., 2005) and result in fewer returns and complaints.

Additionally, because AR technology simply requires a camera-enabled mobile device, these implications carry over to consumers’ at-home viewing of foods. This has become an increasingly relevant given that online food ordering (e.g., Uber Eats, Seamless, Deliveroo) and online grocery shopping have both been on the rise (Kats 2019, Littman 21,019), and even more so since the Covid-19 pandemic (Venkataramakrishnan 2020). AR technology can give consumers the opportunity to view food offerings in their own homes, on their own dining room tables, before placing any orders. In fact, given that one of the underlying mechanisms we uncover is personal relevance, AR presentation might prove even more beneficial in such circumstances. That is, superimposing the featured products not only in a user’s peri-personal perimeter, but also within the user’s intimate household settings might exaggerate the effects on perceived personal relevance, and accordingly amplify the effects on mental simulation and downstream variables accordingly.

In addition, the camera-enabled nature of mobile devices means these implications can also extend to mobile marketing efforts. Brand managers can use AR tools embedded within Snapchat, Instagram, or TikTok filters to present foods in consumers real-time environments and can even enable transactions through those channels (as Domino’s pizza did; Swant, 2018). These opportunities will likely become more common in the future given the developments of AR glasses by the aforementioned tech giants (Swanner 2019).

Aside from the implications for practitioners, it is worthwhile to consider what the current research might mean from a consumer welfare perspective. At first blush, the notion that AR presentation increases food desirability and purchase may suggest that the practice is potentially detrimental for consumer well-being, particularly given the already-widespread tendency for consumers to overeat and the related obesity epidemic (Gao et al., 2022; Scott et al., 2008). Indeed, several health organizations and scholars have attributed the increasingly obesogenic environment to the ubiquitous and compelling nature of food media (Bublitz et al., 2010). However, the implications of our findings are likely more nuanced than such a conclusion would suggest. In particular, the results show that AR presentation does not only increase the desirability of indulgent and unhealthy foods (e.g., the desserts in Studies 1 and 2), but also functioned to increase the desirability of a non-indulgent food (the lamb shawarma in Study 3, as per pretest results in Web Appendix D) and even an otherwise undesirable food item (e.g., the fermented trout in Study 4, as per pretest results). This suggests that AR presentation may very well increase the desirability and purchase of healthy foods as well, which could, in the right contexts, have a beneficial influence on consumer health and well-being. Furthermore, AR presentation may also be a means of encouraging consumers to more readily imagine consuming foods that they are less familiar with (by creating the perception that they are more personally relevant) and could thus possibly provide benefits related to epicurean exploration and learning. This notion represents an interesting route for further investigation, more of which we discuss next.

The burgeoning nature of AR technology and its expanding marketplace applications pave several exciting avenues for potential future research, both within the food domain and beyond. For example, while the current work focused on how AR technology can be used during consumers’ previewing and ordering of foods, it would also be interesting to examine how AR can also be applied during consumers’ consumption of foods and beverages. For example, at Sublimotion, the world’s most expensive restaurant, diners wear headsets and are treated to a 15-course gastronomic show combining gourmet cuisine with AR intended to, “play with emotions, the senses, the set, the aromas, and the taste to be able to create absolutely unique experiences for each scene [course],” (Strause, 2015). In the beverage space, Australian wine company 19 Crimes created an AR app allowing consumers to bring the wine bottle’s label to life through their mobile devices: criminals featured on each bottle become animated and tell their story, enriching the drinking experience by simultaneously engaging both the mind and taste buds (Stone, 2017). It is likely that such applications can significantly transform consumers’ consumption experiences, turning them into highly interactive and experiential episodes.

It is also worth reiterating that this investigation leveraged an AR application where the device’s rear-facing camera (i.e., the camera on the back of the phone, on the side opposite from the screen) projects visual content into the user’s current space. As outlined earlier, other AR applications utilize the device’s front-facing camera (i.e., the camera on the screen-side of the phone, sometimes referred to as the “selfie” camera) to superimpose visual content (e.g., clothing, accessories, or makeup) onto the users themselves (Hilken et al., 2017; Poushneh & Vasquez-Parraga, 2017; Smink et al., 2019; Tan et al., 2022; Yim et al., 2017). It is interesting to consider cases where such applications might visually “transform” the user themself in an effort to show the consequences of food consumption (e.g., a recent Instagram filter shows users’ faces getting fuller if they repeatedly indicate a preference for unhealthy foods). It seems plausible that outcome-oriented mental simulation might become more relevant than process-oriented mental simulation in such cases, and it would be interesting to explore whether such presentations can motivate healthier eating behavior and/or how they might modulate users’ self-image perceptions. This represents a potentially fruitful area for further investigation.

While applications in the food and beverage domain provided a theoretically and externally valid area to examine the effect of AR’s real-time superimposition on consumer responses, it could be interesting and worthwhile to examine how our uncovered effects and mechanisms may or may not apply across other product categories. As alluded to earlier, this might involve systematic investigations into whether the effect and process we found in the current work differentially apply to other hedonic versus utilitarian categories and/or contexts. Further, while we found one dominant sequential mechanism explaining the effect in our studies (personal relevance and mental simulation, respectively), the effect is likely driven by multiple processes (including perceived presence, attention to the background, and enjoyment of the experience), and these variables may become more or less relevant in alternative contexts that future researchers may wish to explore.

Shifting further afield, it is compelling to consider how future research might move beyond products altogether and explore how AR presentation influences consumer responses to the real-time superimposition of other human beings. For example, Google recently created AR “stickers” of the Grammy nominated rapper Childish Gambino, giving users the ability to see a visual depiction of the artist performing in their current environment (Holt, 2019). It is interesting to consider whether and how such applications might influence consumer connections to the individuals who are visually superimposed into their spaces, and how that may or may not extend to the companies and brands sponsoring the content.

While mobile AR is here now (enabled on the billions of Android and Apple smartphones worldwide), the imminent fusion of AR technology into wearable devices (i.e., glasses) will make it an even more permeating phenomenon. One tech executive described this future by saying, “The world is about to be painted with data,” (Fink, 2018). In such an analogy, marketers may very well be holding the paintbrushes, and we hope to see more research exploring how AR can best be used to enhance the customer experiences and marketing outcomes accordingly.