Keywords

1 Introduction

Persuasive technology (also called “persuasive system”) has drawn increasing interest from researchers in the past decade and has been used to modify human behaviors in the fields of health, sustainability and more [1]. However, few studies in the literature have proposed theoretical frameworks for the design and evaluation of persuasive systems, and most theories currently available are not fully developed. A recent evaluation of a famous persuasive systems design framework pointed out that the model does not explicitly describe how to include persuasive design principles and user context in the content development process [2]. A recent literature review of persuasive systems found that many studies of persuasive systems investigated the behavior change effects of the designed persuasive system as a whole instead of analyzing how individual design features and functions of the system caused intended behavior changes [1]. These findings show that current literature does not fully investigate the mechanism by which the user changes behavior or attitude after interacting with the persuasive system, or in other words, the users’ persuasive experience occurring when a persuasion is attempted by the system. Without a framework to describe the details about the interaction between the users’ persuasive experience and components of persuasive systems, it is impossible to fully evaluate and compare the persuasiveness of different persuasive systems as each system would be a “black-box.”

The goal of this paper is to investigate the mechanism of users’ persuasive experience and the implication of the persuasive experience in the design and evaluation of persuasive systems. Famous health behavior change theories were reviewed to conceptualize persuasive experience, and a new framework for designing and evaluating persuasive systems was developed accordingly. The framework can be used for defining successful persuasive systems and for developing meaningful metrics for evaluating the persuasiveness of systems.

2 Method

Recent literature on health behavior change theories was searched on the PMC database and used to identify the most widely adopted theories for behavior change in the health domain. Identified theories were then reviewed, and content that may explain the persuasive experience of users was summarized. Current definitions of persuasive systems and limitations of currently available theoretical frameworks were also summarized. Keywords that were used to search persuasive systems design frameworks were “persuasive system/technology,” “persuasive design,” “frameworks of design persuasive systems” and “design principles of persuasive systems.”

3 Results

3.1 Health Behavior Theories Identified

A literature review in 2015 found that the most widely studied behavior change theories are the Transtheoretical/Stages of Change Model, the Theory of Planned Behavior/Reasoned Action, the Social Cognitive Theory, the Health Belief Model and the Self-Determination Theory [3]. Among these theories, the Self-Determination Theory describes the nature of motivation, and the Transtheoretical Model describes the behavior change process of a target audience being persuaded while other theories mainly describe the factors that cause the behavior change. The Precaution Adoption Process Model (PAPM), which also attempts to explain the behavior change process and closely very resembles the Transtheoretical Model, was also identified during the process of reviewing relevant literature that describes the Transtheoretical Model [4]. In addition, Fogg’s Behavior Model (FBM) and the Behavior Grid [5, 6] are also considered health behavior theories in this paper.

3.2 Self-Determination Theory (SDT)

The Self-Determination Theory (SDT) introduced by Ryan and Deci is the only theory among all reviewed theories that addresses the nature of motivation and how different types of motivations are influenced. In this theory, two major types of motivation, intrinsic and extrinsic, are identified. The former is defined as “the doing of an activity for its inherent satisfactions rather than for some separable consequence;” the latter is defined as “doing something because it leads to a separable outcome” [7]. Authors also explain that when a person lacks an intention to act, he or she is said to be in the state of a motivation. Extrinsic motivation is further divided into four taxonomies: external regulation (performing behaviors or complying because of extrinsic rewards or punishments), introjection (performing behaviors to “enhance or maintain self-esteem and the feeling of worth”), identification (recognizing the importance of the activity to one’s personal values), and integration (fully assimilating identified regulations into one’s personal values and needs). When the state of one’s motivation changes from amotivation to extrinsic motivation to intrinsic motivation, the process is called internalization. The authors argue that internalized motivation promotes more persistent and engaged behaviors [7]. The theory suggests that internalization can be facilitated by increasing one’s sense of relatedness, perceived competence and a feeling of autonomy.

3.3 Stage Theories of Behavior Change

The Transtheoretical Model (TTM) and the Precaution Adoption Process Model (PAPM) use the stage model to describe the behavior change process. The TTM model is the most widely used and studied theory in health behavior change [3], and it mainly consists of four constructs: the Stages of Change, the Processes of Change, the Decisional Balance and the Self-Efficacy. The Stages of Change construct states that when a person changes his or her behavior, he or she will linearly or non-linearly progress through a series of six stages: precontemplation (having no intention of taking action anytime soon), contemplation (intending to change behaviors in the next six months), preparation (intending to take actions soon, measured as within the next month), action (having made specific, observable modifications in lifestyle within the past six months), maintenance (having made additional specific, observable modifications in lifestyle and trying to prevent relapse) and termination (perceiving no need for further behavior change and maximum self-efficacy achieved) [4]. The Processes of Change, Decisional Balance and Self-Efficacy constructs describe the principles that promote behavior change.

The TTM model is criticized for a number of reasons. First, when it is used to produce complex behaviors (i.e. behaviors that consist of multiple behavior patterns instead of a single behavior), such as physical activity and low fat intake, the assessment of stage allocation may become imprecise as people may think they are in line with the proposed complex behaviors while their actual behavior patterns differ from the recommended patterns. Similarly, if people think they are not in line with the recommendation when they actually are, the model categorizes people under the precontemplation stage. A second criticism of the TTM is that the difference between the action stage and the maintenance stage can be measured only by arbitrary methods. Third, future behavior change (i.e. behaviors that occur after completing the initial behavior change) may be produced by psychological changes, such as having a more positive attitude towards a particular behavior, rather than the progression of the stages only. Fourth, the effective behavior change process needs not only stage-based intervention but also individually tailored intervention [8].

In contrast to TTM, which uses days and the state of the behavior change process as the stages of change, PAPM identifies seven stages or mental states: S1-Unaware of Issue, S2-Unengaged by Issue, S3-Undecided About Acting, S4-Decided Not to Act, S5-Decided to Act, S6-Acting and S7-Maintenance [4]. The main difference between TTM and PAPM is that PAPM can distinguish between those who are unaware of the behavior change opportunities and those who are aware of the issue but undecided about changing behaviors. The theory suggests that different communication strategies should be taken to approach the two different populations because the former may not hurry to make a conclusion from the information provided while the latter is ready to make a decision of acting or not [4].

3.4 Constructs of Health Behavior Change

Reviewed health behavior change principles identify what causes a behavior change. The Integrated Behavior Model of the Theory of Planned Behavior/Reasoned Action, the Social Cognitive Theory and the Health Belief Model suggest that the key constructs that influence one’s decision to behave include one’s actual behavioral capability (knowledge and skills to perform the behavior), self-efficacy (perceived confidence in successfully performing the behavior), perceived control (perceived control over the behavior), social norms (others’ expectations and behaviors), attitude (feelings and beliefs about the behavior), perceived susceptibility (perceived likelihood of acquiring the proposed condition), perceived severity (perceived risk of maintaining the current condition), expectations/perceived benefits (perceived positive consequences that the behavior would produce) and cues to action (something that triggers the action) [4].

Fogg’s Behavior Model (FBM) simplifies behavior change by suggesting that a specific behavior would easily occur if one’s level of motivation, ability to perform the behavior and the right trigger were merged [5]. Fogg emphasizes that if the behavior is simple enough to be performed, one will need only basic motivation and an adequate trigger (cues of action) to start the behavior. He argues that the general strategy to behavior change is to start with a very simple behavior change that needs little effort to perform and gradually build on small successes of behavior change to achieve the ultimate behavior change goals.

Fogg’s Behavior Grid identifies 15 types of behavior (lowered from its original 35 types) categorized according to the duration that each behavior needs to be performed and the nature of the behavior being performed [6]. As Fogg explains, there are dots (behaviors done one-time and measured in seconds, minutes or hours), spans (behaviors performed for days and weeks) and paths (habits or permanent behavior changes). It is suggested that dot behaviors are the easiest to perform. The difficulty of performing the behavior increases when the duration of the behavior change period increases.

3.5 Frameworks for Persuasive Systems

The literature search identified three widely adopted frameworks that intend to guide the design and evaluation of persuasive systems. Fogg’s original work that conceptualized persuasive technology is very well known, and the work has been cited by many scientists who study persuasive systems [9]. In 2009 Fogg also provided an eight-step design process of persuasive technology that was integrated with FBM [10]. A more systematic design and evaluation framework, Persuasive Systems Design (PSD) framework [11], was developed by Oinas-Kukkonen and Harjumaa in 2009, and this framework has been used and evaluated by many other studies [3]. Later, Oinas-Kukkonen called PSD-based systems a Behavior Change Support System (BCSS) [12]. Lastly, the concept of motivational affordance was promoted by Zhang, which emphasized that persuasive systems should incorporate the design features and functions that fulfill the needs of user motivation [13]. Karanam et al. later listed types of motivational affordances that are used in a health game [14].

3.6 Design and Evaluation Framework

In Fogg’s framework of persuasive technology, Fogg identified the major persuasion principles that are used in persuasive technology, but he did not provide specific guidance on how to implement these principles into the actual design of the system [9]. Fogg addressed this issue in his eight-step design process [10]. The process suggests that designers need to identify the behavior to cause, the audience of the persuasion and what FBM factor prevents the target behavior. Then designers select a technology channel based on the analysis. After initial analysis of the context and selection of the technology channel to be used, the process suggests the design team find relevant persuasive technology examples and imitate effective design patterns of the examples. Then the team should test and iterate the system quickly to improve and comprehend the system. It should be noted that what makes Fogg’s approach unique from other methods is that the design process does not directly generate system features and functions based on existing persuasion principles but instead, first identifies the existing design patterns that are already successful in terms of applying the persuasion principles (i.e. design patterns that already have generated persuasion effects, preferably). Fogg says this approach is taken because “there is no need to reinvent the wheel.” Therefore, Fogg’s approach focuses more on building successful persuasive technology products than on investigating the general cause and effect of developed persuasive systems.

Oinas-Kukkonen and Harjumaa provide more of an “academic” framework for the design and evaluation of persuasive systems in their PSD model. The model suggests that key steps for designing persuasive systems are to identify the intent; analyze use and user and technology context, choose direct or indirect routes to deliver the persuasion message, select appropriate persuasion principles to implement, translate principles into system quality requirements and implement a design based on the requirements [11]. To evaluate the system, designers can use checklists to evaluate if the system quality requirements are fulfilled. Problems with the PSD model were later identified by Harjumaa and Muuraiskangas when they implemented the PSD model in two persuasive systems [2]. First of all, some persuasion principles provided by the PSD model were perceived to be overlapped by the participants of the study, and the list of persuasion principles might be incomplete. Secondly, the model appeared to be too generic as it did not offer thorough guidance to designers on how to perform the design activities suggested by the model.

Oinas-Kukkonen later identified critical questions that need to be answered in future research of Behavior Change Support Systems (BCSS) developed using the PSD model [12]:

  • How does one measure (behavior) changes caused by the system?

  • How does one conduct experiments to pinpoint a change to a specific software feature?

  • Which software features, or combinations of software features, have the greatest impact in different settings?

  • Which modes of interaction are more persuasive than others? How does one measure the fit between the modes and persuasion effects so that different modes can be compared?

Lastly, motivational affordances refer to “the properties of an object that determine whether and how it can support one’s motivational needs” [13]. This is more of a supplemental concept that can be used to identify the design principles and features that satisfy generic motivational needs of users [13, 14].

4 Persuasive Experience Framework

4.1 Necessity of Persuasive Experience Concept

Discussion of persuasive experience is necessary because the currently available frameworks of persuasive systems do not address the change process occurring to users when being persuaded. When designers determine what intention to realize by persuasive systems, some may propose to change complex behaviors, such as weight-loss management and chronic disease management, while others may propose to change simple and specific behaviors, such as sharing a document and clicking the order button. If the intention is about changing simple behaviors that occur only once, the design and evaluation of persuasive systems may be easy because the effects of the system are easily observable, e.g. measured the frequency of the behavior. However, if the intention is about changing complex behaviors or forming habits, which consist of multiple behaviors performed in different situations over long periods, the previous approach may not be applicable for two reasons. First, multiple design features or functions may work together to cause different behaviors in different situations. This makes it difficult to observe the interaction between specific features or functions and the occurrence of target behavior change. Second, forming a habit or mastering complex behaviors is a process that takes time. Designing a system that intends to change complex behaviors requires a design framework that explains how design features and functions of persuasive systems interact with and change users over time, which is not provided by current frameworks. Therefore, it is important to incorporate behavior change theories such as SDT, TTM and PAPM into the current design framework of persuasive systems as they describe the needs of users at different stages of behavior change and how the nature of users’ motivations influences their behavior.

4.2 Persuasive Experience Model

Persuasive experience is an incremental or decremental transition that occurs in between the user-behavior relationship stages. In other words, a persuasive experience occurs when the nature of a user’s relationship towards specific behaviors changes. The Persuasive Experience Model (PEM) illustrates the mechanism of persuasive experience using stage-based theory (Fig. 1). The model is developed by integrating SDT, TTM and PAPM.

Fig. 1.
figure 1

Persuasive Experience Model

Full size image

The PEM consists of two components: the user-behavior relationship stages and the persuasive experience phases. The concept of the user-behavior relationship stages posits that there are six stages in one’s psychological relationship with the defined behaviors when the behavior change process occurs. They include (S0) not known, (S1) recognized, (S2) interested, (S3) motivated, (S4) engaged, (S5-1) obsessed and (S5-2) habitualized. When user relationship stage moves towards a later stage, the user’s level of motivation towards performing a behavior also increases. The persuasive experience phases describe the nature of the persuasive experience occurring. The phases are divided into three parts: awareness, action and maintenance.

The PEM model explains all psychological outcomes that users will gain through interaction with specific design features and functions of persuasive systems. The PEM model divides ultimate behavior change (usually complex behaviors) into multiple simple behaviors that can be achieved at different stages of the user-behavior relationship. Each design feature and/or function is intended to achieve a certain simple behavior or attitude change to move users forward in the behavior change process. Therefore, the PEM model posits that persuasive design features and functions may not necessarily cause the ultimate behavior change directly, but as long as they support incremental stage transitions, the ultimate behavior change will occur eventually. To determine whether particular design features and functions are effective, the designer needs only to know whether they contribute to particular persuasive experience. Persuasive experience can be measured by the frequency of particular behavior or user perception of and attitude towards specific behavior/opinion. Therefore, this approach differs from the traditional approaches that directly connect all design features and functions to the ultimate behavior change(s) and thus are less helpful for explaining which features, functions, or combinations of them caused the occurrence of the ultimate behavior change.

The definitions of the user-behavior relationship stages are described below:

  • Stage 0: Not known. Target users are unaware of the opportunity to change behaviors. At this stage, designers should focus on strategies to reach and get the attention of the target audience and deliver the information that is relevant to the behavior.

  • Stage 1: Recognized. Target users recognize the opportunity to change behaviors. They may already know the pros and cons of the proposed behaviors as they may have learned about them via the Internet, social interactions, or other channels unintentionally. At this stage, target users may still remain uninterested in the behavior. The users at this stage will not proactively search for information relevant to the behavior if they do not find anything further interests them (i.e. behaviors are relevant to their personal lives, values and benefits, etc.). Thus, persuasion strategies at this stage should focus on increasing users’ interests to either perform the behavior or learn more about the behavior. Of note is that users may recognize the proposed behaviors to be either positive or negative at this stage. If they perceive the behaviors to be strongly negative, they may refuse to perform the behaviors.

  • Stage 2: Interested. Target users are interested in learning about the opportunity for performing the behavior. If the target users have an opportunity to perform the behavior, and the behavior is simple enough, users may try the proposed behavior once even if they are not strongly motivated. Otherwise, users at this stage may proactively search for relevant information about the behaviors if the interest level is high and there are opportunities to do so. However, interest can fade soon if no effective reinforcement strategies are implemented, and users may experience decremental persuasive experience (i.e. their user-behavior relationship retreats to Stage 1) when they lose interest in performing the behavior.

  • Stage 3: Motivated. Target users are ready to or getting ready to perform the proposed behaviors. Users at this stage are usually motivated enough to perform the behavior once or for certain periods, but such motivation may decrease rapidly if the difficulty of performing the behaviors increases. Hence, persuasive systems should simplify the behaviors, aid the removal of barriers or improve a user’s ability (knowledge and skills) so that the proposed behaviors are performed at least once. Persuasive systems may persuade users to start from some simple behaviors instead of complex behaviors at this stage. For example, if the proposed behavior is to form a habit of running, the system may encourage users to start forming such habit by trying to run for only short amounts of time with moderate intensity in the beginning to help users avoid experiencing muscle pains and feeling uneasy.

  • Stage 4: Engaged. Target users are willing to integrate the proposed behaviors into their habitual/daily life routine, or at least they are willing to perform the behaviors for a period of time as long as they have a passion for the behaviors. Users are highly motivated at this stage, and they can perform the behaviors even if they face some obstacles to do so. However, users may gradually discontinue performing the behavior if they fail to integrate the new behaviors into their current lifestyles and rely on external sources to maintain their motivation. Possible cases of failure may include situations when users are no longer supported socially, or their lives get busier, and they fail to manage time for running. To increase engagement and move to the next stage, designers should propose persuasion strategies that facilitate the internalization of users’ motivation and promote sustainability of the behaviors at this stage.

  • Stage 5-1: Automated 1-Obsessed. Target users are obsessed with the proposed behaviors. This type of motivation largely depends on the internal feeling of users. If users find target behaviors enjoyable, they will automatically perform the behaviors even if there are obstacles to doing so. However, the behaviors may not be performed permanently if users perceive the behaviors to cause negative impacts on their lives (i.e. the behaviors are useless, meaningless and/or harmful). Also, if users find the behaviors to be no longer enjoyable, they may stop performing them.

  • Stage 5-2: Automated 2-Habitualized. Target users have successfully integrated the behaviors into their lifestyles (i.e. formed a habit), so they will no longer need to deliberately maintain their motivation. Users may not stop performing the behaviors easily at this stage. Users may no longer need persuasive systems to facilitate the behavior change anymore. However, persuasive systems can still provide value to users by continuously facilitating the internalization of users’ motivation.

The definitions of the persuasive experience phases are described below:

  • Awareness phase refers to the process of raising awareness among target users. Persuasive experience that occurs within the range of Stages 0 to 2 belongs to the awareness phase. Critical factors that raise awareness of users may include attention, curiosity, emotion and rational judgment of benefits and risks.

  • Action phase refers to the process of engaging users in the proposed behaviors. Persuasive experience that occurs within the range of Stages 2 to 4 belongs to the action phase. Persuasion strategies used in this phase focus on removal of barriers, simplification of behaviors and improvement in users’ knowledge and skills to perform the behaviors. Perceived usefulness, meaningfulness and enjoyment may also play key roles at this phase as they facilitate the internalization of motivation.

  • Maintenance phase refers to the process of aiding users in sustaining the behaviors. Persuasive experience that occurs within the range of Stages 4 to 5-1 or 5-2 belongs to the action phase. This phase promotes the automation of behavior by either largely enhancing the enjoyment of the behavior or supporting the habit formation. Some behaviors are inherently enjoyable while others may not generate that much fun. However, it is possible that persuasive systems can alter the nature of the behavior by gamifying the behavior.

The PEM model suggests that during the behavior change process, as users’ motivation levels increase, users may not only perform the proposed behaviors but also other relevant behaviors that are not specified by designers. For example, users may be motivated to care for their health more when they try to form a habit of running. This increased motivation turns users into proactive entities pursuing better health, and thus, they may also start performing other health-related behaviors such as eating more vegetables and playing tennis.

It should be noted that a persuasive system may not necessarily cause all phases of persuasive experience. Some persuasive systems may be designed to mainly raise awareness instead of engaging users in the behaviors. The decision of which phases to include in a single persuasive system largely depends on the context of the users who use the system and the technology the system is built upon.

4.3 Advantages of PEM over TTM and PAPM

The PEM model visualizes users’ psychological relationship with the proposed behaviors during behavior change whereas the TTM and PAPM attempt to visualize stages of the behavior change process without accurately describing the actual psychological status of users.

The TTM model suggests that an individual in the precontemplation stage has no intention of taking action anytime soon, but the model does not specify “why” one has no intention. This issue is addressed later in the PEPM by dividing the precontemplation part into the “Unaware of Issue” stage and the “Unengaged by Issue” stage. There is a major difference between users who already know the benefits and risks of certain behaviors and users who do not recognize the problems, and those users will need different interventions to raise awareness. The PEM model addresses this issue better, because it labels each user differently to accurately and precisely describe their psychological statuses (not known, recognized and interested).

Another issue of the TTM and PAPM is that they do not differentiate users who are performing the behaviors at different levels. In the case of running, some may run once a week with no intention of increasing the frequency after the behavior change process, while other users may run three times a week and are still interested in increasing the frequency and improve their running. The former and latter groups both changed their behaviors, but the TTM and PAPM may either categorize both of them under the same action stage or put the former back to the precontemplation stage as they do not meet the required level of activeness. However, users who have already performed the behaviors are different from those who have never performed the behavior or are unaware of the issue. The PEM addresses this issue by differentiating users who are acting and by identifying whether they are motivated or engaged in the behavior.

The PEM also separates two different psychological statuses of users at the maintenance phase (described as one single stage in the TTM and PAM). Users may maintain the behaviors either because they feel inherently satisfied or enjoy performing the behavior, or because they have successfully made the behaviors their habit. Original models do not specify such differences in the nature of maintenance.

Therefore, stages described in the PEM model clarify the psychological statuses of users themselves during the behavior change process. This approach yields more relevant information about users compared to the approach of just identifying the behavior change processes that users are experiencing. The PEM has the potential to help designers better tailor specific interventions to appropriate user populations.

4.4 Designing and Evaluating Persuasive Systems Using PEM

The design process of persuasive systems using the PEM model differs slightly from Fogg’s eight-step design process and the PSD model. Designers may still analyze the context of persuasion by researching users and technology, but they may also identify which persuasion experience(s) should be generated by the system before they directly apply persuasion principles to the context. Designers should then identify the measureable outcomes that characterize the success of each persuasive experience the system aims to generate and match persuasion principles to achieve the identified outcomes. After this process, designers should translate the persuasion principles into design elements (design features and functions). With these three steps, documenting each design element, clarifying persuasion principles an element is implemented upon, and defining what measureable user outcomes a design element should yield, designers can clearly see the effectiveness of each design implementation when they evaluate the system. When evaluating the system, designers may check whether each design element and the persuasion principles match the context of users and the technology the system is built upon, and check if the design elements are effectively yielding the predicted outcomes, as well as measuring if the persuasive experience has occurred while users interact with the system. Possible methods to collect data may include system logs, observations, surveys and interviews.

5 Conclusion

This paper reviewed the current literature on behavior change theories and persuasive systems and described a new framework of persuasive systems. As a result, the Persuasive Experience Model is developed to guide design and evaluation of persuasive systems. Further investigation into the concept of persuasive experience is needed, and the PEM is subject to testing through empirical studies to examine its validity. However, the persuasive experience framework certainly brings a new perspective to the understanding and development of persuasive systems.