Keywords

1 Introduction

India, a country with a population of over 1.2 billion people has witnessed massive economic growth over the past few decades. In spite of a significant increase in the Gross Domestic Product (4.5 times) and per capita consumption (3 times) in the past two decades [1], 194.6 million people are undernourished in India and is home to the largest hungry population in the world (15.2% of the population) [2]. There are around 3000 children in India dying every day from illness related to poor diet [2]. The food grain production reveals a completely different picture. India ranks first worldwide in farm output [3]. It is the largest producer of milk and second largest producer of wheat, rice, sugar, groundnut and inland fish. Producing adequate amount of food does not guarantee food or nutrition security and is not the only criteria to eliminate hunger and malnutrition. The relation between food waste and hunger do not seem to have a one-to-one mapping; food waste indirectly increases the global food prices [40], which in turn affects people with low access to food.

Multiple initiatives have been started, across the world, to address the food waste problem in different stages of production and consumption [4,5,6,7,8,, 5, 6, 7, 8, 9]. Some of these solutions are working well to manage unused or excess food at a production stage, However, leaving food on the plate is a behaviour that needs to be changed at an individual level. We reviewed literature on behaviour theories and persuasive design techniques to understand food waste behaviours.

Behavioural economists suggest that humans are not only irrational, their irrationality is highly predictable. Further to this, informing people about their own behaviour can improve the quality of rational decisions [10]. Researchers in HCI and behavioural sciences are increasingly exploring the use of technology to support behaviour change in various domains such as health and sustainability [11]. HCI researchers have used behavioural theories to make design decisions about the technical systems, to guide qualitative evaluation and to define users [11].

In this paper, behavioural theories were primarily used to address the design of an interactive installation—The Darker Side, focused on lesser known facts on impacts of food waste in India. The installation attempts to address the habitual behaviour problem of leaving food on the plate, which in turn tries to make a case for the use of interactive installations to bring about behaviour change.

2 Related Work

Habitual behaviours are performed without active thinking. These occur when there has been a high-frequency history of the behaviour [10]. People do not pay attention to food waste and its disposal as they perform it as a routine task [12, 13]. Persuasion and motivation [14] have been identified as some of the triggers to break the habitual cycle. According to Fogg’s behaviour model for persuasive design, three factors essential to bring about a behaviour change are ability, motivation and trigger. In case of food waste, while individuals possess the ability to reduce waste, they might not be motivated to change their food waste behaviour. If individuals have both the ability and motivation to do so, timing and placing an effective trigger near an appropriate location might induce the change.

Consolvo et al. [15] derived eight guidelines for designing technologies for lifestyle change. According to them, such technologies, need to be abstract, reflective, unobtrusive, public, aesthetic, positive, controllable, comprehensible to users, and they must include historical data. The BinCam experiment [10] was conducted to improve recycling behaviour within a closed homogeneous group. The authors describe a design which attempted to raise awareness through social influence and aversion effect. They argue that social influence results in self reflection and causes a ‘feeling of shame’. Although BinCam could not effectively influence user behaviour, social influence and gamification have been identified as possible methods to motivate people to change their behaviour.

Similar projects of inducing behavior change on social issues include Cleanly [16], a gamified urban trashducation system, which aimed at creating awareness about environmental pollution. The trash track project [17], created awareness about the waste removal system by using tags to track in real time and visualise the journey of individual trash objects. The World’s Deepest Bin [18, 19] used audio feedback to prompt reflection on waste disposal.

The piano staircase [39] was an attempt to make people do physical activities through playful persuasion. In [20], the authors attempted encourage exercise in public spaces through the use of a Kinect based interactive installation which enabled the users to perform upper body Tai Chi movements.

There have been attempts to motivate people to perform physical activities [21], reducing electricity consumption [22], improved waste disposal and management [10, 16,17,18] using playful persuasive techniques and contextual information displays. Generating awareness about serious social issues such as domestic violence [23] have also been attempted through interactive billboards, where cameras detected faces looking at the billboard and morphed the visuals. Even though most of these installations were successful in engaging the audience and achieving the ‘target behaviour’ at that point of time; there seems to be little work done on evoking self-reflection and sustaining the behaviour change among users.

Using design to change user behaviour for social benefits has also been identified as an upcoming trend [21, 24]. ‘Design with Intent’ (DwI) [25] is an approach where patterns help designers to ‘Design for Behaviour Change’. ‘Change the campus with fun’ [26] is an implementation of DwI, implementing ‘fun theory’ in their service design, considering various social and sustainability factors. It suggests that gamification can induce playful triggers, which will not only help to achieve ‘target behaviours’ but also make them habitual or customary. Awareness raising [27] and presenting contextual information [28] has also been identified as a key stage in the processes of behaviour change as it stimulates self-reflection.

Ybarra and Trafimow [29] suggest that sense of group results in conformation to social norms, where the individual is motivated by the desire to obtain social approval and avoids rejection by others [30]. This might not result in true internal change in behaviour as individuals can behave superficially when under social surveillance.

According to Fogg, persuasion should not be coercive, manipulative or deceitful. It should allow individuals to remain in control of their own actions [31]. Although, the use of coercion in persuasive technologies has been questioned and excluded by Fogg, Kirman et al. [32] and Foster et al. [33] have argued that constructive aversive feedback and lighter forms of punishment can support behaviour change. Nonetheless, coercive strategies need to be designed carefully such that they do not disengage or intimidate users.

‘Nudges’, are considered to be soft, unobtrusive type of strategies, which influence people to change their behaviour [34]. ‘Libertarian Paternalism’ helps individuals to use their ‘Reflective System’ of thinking (Rational, Controlled, Self-aware, Rule-following) during decision making.

Through this short review, we can assume that, installations can be used for playful persuasions [21] and can be designed to be non-intrusive, thereby a potential medium to bring about behaviour changes. However, a ‘one-size-fits-all’ approach might not work well [35] and hence performing user studies to understand the audience could bring up insights which help in taking design decisions for the intervention.

3 Primary Research

The university in which the experiment was carried out was a fully residential institute with in-house dining. The selected hostel housed the largest mess in the campus and served around 2000 students. The mess was operated on a contract basis. The services provided by the contractor broadly included cooking, distribution of food, cleaning and maintenance of the mess. This mess was selected for two primary reasons—it was the largest mess in the campus and the mess coordinators were already motivated to reduce food waste. The mess had posters asking people to stop wasting food and a white board where food waste statistics were updated after each meal. These evidences indicated that certain stakeholders were already motivated and a trigger might help in bringing about a behaviour change.

Before commencing user studies, different stakeholders of the mess and their roles were identified, namely, Students, Mess Staff, Mess Managers, Mess Student Coordinators. Semi-structured interviews were carried out and different key questions were designed for each of these stakeholders. Seven students, two mess workers, one mess manager, one mess coordinator were interviewed. Students were asked questions to understand their food waste behaviours, the mess manager and workers were interviewed to understand the supply chain, waste and excess food management.

Students showed resistance to answering questions about food waste. Individuals, wasting food, were selected and interviewed in an attempt to understand their food waste behaviours. The approach was to empathise with them; conversations were initiated by asking questions about the quality of food. Most of them blamed the quality of food for their wastage; a handful of them accepted overestimation of their eating capabilities to be the reason. Students complaining about the taste of food were further asked to objectively categorise their reasons for disliking the taste, as too salty, spicy, bland, sour etc. The responses received were mixed. Each interview lasted for around five minutes, this indicated that the students were not keen to spend time to elucidate their behaviour.

The user journey and layout of the mess were studied to decide the position of the installation. The placement of trigger had two potential options—the food counter and the dustbin. Placing it near the food counter would ideally have motivated them to take food wisely, but the queue for the food and limited space were major drawbacks. A crowd gathering near the counter would add to the chaos during rush hours. Moreover, hungry students would not have wanted to engage with the installation. Hence, the space surrounding the dustbin was selected for the installation.

Major observations and insights drawn from the primary research were: students did not have loss aversion (a tendency to strongly prefer avoiding losses) [34] when it came to wasting mess food as they considered it to be equivalent to free food. While some students were motivated to reduce food waste, they limited themselves to putting up posters, which, probably did not create much of an impact. Most of the students did not consider tasting as an option before taking a larger portion. As the food counters were placed at extreme corners of the mess, they tended to take large amounts of food to avoid coming back for further helpings. These observations and insights were analysed, collated and used to build design ideas. Each design idea intended to solve multiple problems identified during the primary research by using the insights gained through literature review and user studies.

4 Design of the Installation

Positive and negative (‘Pleasure’ and ‘Pain’) motivations were explored while developing the message to be put across. Playful persuasion and gamification were also considered as possible options for intervention. In order to raise awareness through the installation, the option of presenting facts related to food waste and hunger were also examined. The motivated students could also be made a part of the design solution to create social influence on the less motivated students.

A small scale pilot installation was implemented to validate initial design ideas, test the hardware, get initial feedback and suggestions. It also helped in understanding the space and finalising the position of the installation.

The design decisions focused on facilitating awareness about the ill effects of food waste. The strategies applied were ‘gaze effect’ (Gaze Effect: the awareness of any object can induce an awareness of also being an object) [36] and ‘panopticon’ (Panopticon: a feeling of being observed, under surveillance) [37]. A poor hungry Indian child staring at the camera was selected to be the subject of the visual, as people are generally more concerned about children than grown ups [38]. It also induced among the audience, a sense of being observed as a subject. The messages were designed to evoke pain and give a moderate level of aversive feedback. The messages changed from low to high level of aversion depending upon the amount of food thrown (wasted) by an individual. The changes in the messages were kept subtle to make the installation incidental in nature. The statements used for negative motivation were an attempt to articulate the behaviour of students towards food as understood during primary research. The quantitative data captured before and after the pilot-installation was not significantly different but it attracted a lot of attention. The audience found the concept of reducing food waste through an installation intriguing. They were also interested in the technical nuances, enthusiastically gave suggestions as to how the hardware and visuals can be further improved. Taking cues from the pilot implementation, the main installation was put up on a larger display with improved visuals, messages and hardware.

4.1 The Final Installation

The final implementation was designed to detect proximity and faces of the audience. The food waste data was measured through a weighing machine on the dustbin which was connected to a Raspberry Pi and projected on a wall mounted display. In a college mess, the minimum screen size was decided to be 40 inch which was an observation made during the pilot implementation. A flat arrangement for the screen and the dustbin was aimed to facilitate self reflection through the awareness facts, contextual information and visuals. The installation was named ‘The Darker Side’ highlighting the lesser known facts about food waste and hunger issues in India.

The visuals projected on the screen changed only for users who (wasted) threw above a certain amount of food into the dustbin. For others, the screen displayed only the total amount of food wasted. A minimum threshold of 100 grams was decided to eliminate instances of inedible leftovers being dropped. The threshold was decided after analysing the data captured during the pilot implementation.

The visuals had three components: Awareness facts, Contextual data, and Interactive visuals. The awareness facts were related to the plight of farmers and malnourished children in India. They raised questions about the attitude of the students towards mess food, and reminded them that they were a privileged section of the country. The messages (Fig. 1) were aimed at giving aversive feedback and were constructed such that they highlighted the undesirable behaviour. The contextual data projected the total food wasted in the mess and food wasted by that individual at a particular point of time. This data reinforced the need for cognisance and the urgency to change the behaviour of wasting food on plate.

Fig. 1
figure 1

Messages with facts, aversive feedback, highlighting in red the desired behaviour

Full size image

The images (Fig. 2a) portrayed the underprivileged population and complemented the awareness messages. The subjects, being close up portraits of people staring at the camera, if closely observed were not static. The image changed based on the amount of waste thrown into the dustbin. This contributed to the incidental attribute of the installation. The facial expression of the subjects changed (Fig. 2b) (happy to sad, healthy to malnourished) according to the amount of food thrown which was thrown into the bin. The time of the transformation was manipulated by observing the average attention span of the audience; five to eight seconds for the mess under consideration. The background was white and the images were grayscale. This was done to reduce elements which might distract the audience.

Fig. 2
figure 2

a (left) Static visual of a healthy child. b (right) The visual changes, the face of the child morphs into a malnourished child when food is thrown in the bin. Source Flickr

Full size image

5 Evaluation Protocol

In order to understand the effectiveness of the installation, both quantitative and qualitative evaluation approaches were adopted. The experiment was designed to collect individual and total food waste in pre (control data) and post installation phases. The data collection was carried out in three stages: pre-installation, installation and post-installation. In the pre and post installation stages, food waste data was collected without the installation during lunch and dinner (3 hours for each meal). In the installation phase, the installation was put up for lunch and dinner. Different set of visuals were chosen for the two meals to maintain audience engagement. An online survey was prepared and administered in the post-installation stage. This survey had 15 closed and 2 open ended questions. Some of the questions in the survey were:

  • How interesting did you find the installation? (evaluating installation as a medium)

  • Do you think that the installation was effective in spreading awareness? (effectiveness of installation in spreading awareness)

  • Do you think it motivated you not to waste food? (effectiveness of installation in motivating audience not to waste food)

  • Do you think the system has helped you to take informed decisions regarding food waste? (effectiveness of installation in breaking habitual behaviour cycle)

The overall impact of the installation and its sustenance were measured through:

  • Do you get reminded of the installation when food waste is being discussed?

  • Do you get reminded of the installation when someone wastes food?

Demographic information was also recorded with the possibility of drawing inferences based on change in behaviour according to age, education etc. The objective questions were in the form of a 5 point likert scale which were quantified and evaluated using standard statistical methods.

6 Results

The installation was put up for a limited period of time and there were multiple variables contributing to the food waste for any particular day at the mess. Collecting food waste data for a longer period of time would have neutralised the outliers within the dataset. One of the major challenges was that the menu was different across the meals making it difficult to evaluate the effectiveness of the installation through quantitative data. The total and individual food wasted collected across the meals during the pre and post assessment phases were not significantly different. Hence, survey data was given more weightage during evaluation. An anonymous online survey was circulated within the hostel and we received over 250 responses. These responses were cleaned and statistical tests were performed on around 200 responses.

The quantitative results from the survey data indicate ‘towards positive’ results:

The responses to questions related to, ‘Motivating the audience not to waste food through installation’ with the responses to questions on the ‘effectiveness of installation in spreading awareness’ had a Pearson correlation of 0.7245.

The question: ‘How interesting did you find the installation?’ had a median score of 4, on a scale of 1 to 5 (1 = not at all interesting, 5 = extremely interesting). 95% confidence interval of the mean was 3.45–3.75 with standard deviation (SD) = 1.09.

The question: ‘Do you think it motivated you not to waste food?’ had median score of 4, (1 = Strongly Disagree, 5 = Strongly Agree). The 95% confidence interval of the mean was 3.3–3.64 with SD = 1.20.

The question: ‘Do you think that the installation made you more informed about food waste?’ had a median score of 4 (1 = Strongly Disagree, 5 = Strongly Agree). 95% confidence interval of the mean was 3.31–3.63 with SD = 1.29.

Some questions did not indicate a statistically significant trend but the qualitative feedback was encouraging, as the students voluntarily provided multiple design ideas in the open ended question. Students also volunteered to join the initiative of reducing food waste. Some of the feedback received was:

Figures showing empty plates—ticked as ‘You have done a good thing’; Plates with a little remaining food-ticked as ‘OK’; Plates with a considerable amount of food remaining as ‘You are doing it wrong’. Little quotes may also help near the food distributing table like: ‘Take what you want, eat what you take’.

A big projector screen should be used instead of an LED screen so that people can see what’s going on while having their food. This will certainly make them not to leave food in their plates. A video can be made about food wastage with real footage (without showing faces) and circulated among residents.

More than 50% students mentioned that they were reminded of the installation when food waste was being discussed or someone wasted food after the installation was permanently removed from the mess. People gathering near the installation, audience engagement, motivated the students to explain the concept of the installation to fellow students and added to the positive feedback. The quantitative results of the survey and the qualitative feedback reinforce the possibility of using interactive installations to bring about behaviour change.

7 Conclusion

The main aim of the experiment was to understand whether a positive correlation can be drawn between interactive installations and behaviour change. The uniqueness of the installation designed lies in the use of aversion and moderate coercive feedback combined with persuasive techniques. The experiment was conducted on a heterogeneous group of students having diverse socio-economic and cultural backgrounds. The installation was put up for a short period of time and the statistical results of the post-experimental survey indicate a ‘towards positive’ trend. The preliminary statistical inferences along with the qualitative feedback indicates that interactive installations have the potential to bring about behaviour change.

This experiment is just the first step towards bringing about a behaviour change pertaining to social and environmental issues. Some of the persistent challenges include, finding a sustainable solution to the food waste problem, expansion and scaling up in different contexts, attracting and retaining attention of the audience when implemented over a larger span of time. The future scope of work includes implementing an ecosystem around the installation, increasing the sustainability of the solution and cost reduction of the setup.