One core technique that is integrated in biofeedback training is operant learning. In biofeedback training, feedback is given about physiological changes that occur. The complexity of biofeedback can vary from presenting raw signals of physiological activity (eg, heart rate variability [HRV], respiration, and electroencephalography [EEG] signals) to a moving frequency analysis [58]. In clinical and educational settings, additional aids are often provided by presenting feedback in the form of graphs, images, or sounds [59]. However, simply showing individuals their activity is insufficient. Instead, it is vital that individuals are actively taught, using operant (or instrumental) learning, how to change their physiological state, specifically by providing real-time feedback and reinforcement while regulation attempts are made [60-64].

In operant learning, positive reinforcement is used to shape an individual’s behavior by strengthening adaptive behavior through the use of rewards, which makes the original behavior more likely to occur [65]. In a typical biofeedback paradigm, an individual is placed in front of a computer screen on which feedback regarding their physiological state is presented. When the appropriate activity increases or inappropriate activity decreases, this change is followed by a pleasant response (eg, a pleasant tone). As sessions are repeated, the thresholds for receiving a reward are gradually modified, thereby stimulating the display of healthy physiological activity [66,67].

In addition to operant learning techniques, biofeedback training often incorporates self-regulation techniques that are similar to meditation and mindfulness practices. Meditation is a practice consisting of various exercises that are meant to shift one’s attention primarily to internal stimuli to achieve better well-being and emotional balance [31,68]. Mindfulness meditation likewise aims to shift one’s attention to present experiences, but puts an additional emphasis on regarding experiences in a nonjudgmental and accepting manner [69,70].

Furthermore, similar mechanisms of change can be identified in both biofeedback and meditation. For instance, similar physiological patterns of change are found in response to biofeedback as well as mediation [71], such as a coherent cardiac rhythm [72,73]. Another similarity is that both practices encourage individuals to shift their attention to internal experiences. In biofeedback, this is achieved by using biosensors to provide individuals with feedback on changes in their physiological activity, whereas in mindfulness training, this is achieved by guided mediation exercises where individuals are prompted to shift their attention to the present moment and present experiences [69,70,74]. Both approaches also often train individuals to use slow, diaphragmatic breathing [58,75-79]. However, although both approaches try to enhance individuals’ internal attention, biofeedback specifically focuses on changes in physiological activity, whereas meditation aims for a broader focus on the present moment, including all present experiences. Furthermore, mindfulness meditation explicitly encourages individuals to regard observed sensations in a nonjudgmental manner [69,70], whereas biofeedback training does not. The core difference between meditation and biofeedback, however, is that meditation training does not provide feedback on how the person is doing in terms of regulating their stress. In contrast, biofeedback training gives participants continuous feedback on how they are doing in terms of physiologically regulating their stress. Thus, although there are various similarities between mediation practices and biofeedback, we argue that the feedback provided in biofeedback training is crucial to help people gauge their progress and to keep them engaged in the training.

Various mechanisms have been proposed to explain the effectiveness of biofeedback (Figure 1). One of the core assumptions is that biofeedback leads to increases in interoceptive awareness, which in turn helps individuals to better regulate their physiology (Figure 1, paths A and B) [80-82]. Interoceptive awareness is our ability to sense and interpret internal and physiological signals, which is an important part of emotion regulation [13,15,36,83,84]. Scientific research on interoceptive awareness started in the field of psychophysiology back in the 70s and 80s, followed by a new wave of interest in the 90s sparked by the introduction of the somatic marker hypothesis [36,83]. According to this hypothesis, the interpretation of bodily sensations is closely associated with emotional processing and decision making [10,85]. These early days of interoceptive awareness research also gave rise to the idea that biofeedback training could help individuals improve their interoceptive awareness [36], and this assumption is still at the core of biofeedback research today [35,38,41,43,47,59,60,86,87]. The support for this claim, however, seems to be mostly theoretical in nature, as there is very little direct empirical evidence showing that biofeedback training systematically improves awareness of physiological states (Figure 1, path A). Equally important is that it is unclear if and when this internal acuity is adaptive [36,88,89].

Interoceptive awareness has been positively related to decision making [90,91] as well as various forms of emotional processing, such as emotional memory [92], emotion recognition [93], and emotion regulation [15-18]. Importantly, it seems that increased interoceptive awareness is not always better. For example, heightened interoceptive awareness has been linked to increases in anxiety [94,95] and has also been linked to various anxiety disorders such as panic disorder and social anxiety disorder [89,96-99]. In individuals with panic disorder, an increased focus on and the subsequent misinterpretation of physiological sensations is the main cause of panic attacks [96,100]. For individuals with social anxiety disorder, this vigilant attention to physiological sensations likewise impairs their ability to process information from their immediate social environment. Socially anxious individuals often interpret bodily sensations of anxiety as a confirmation that they cannot function well in social situations. Furthermore, these anxious individuals believe that their bodily sensations are clearly noticeable by others, which plays into their fear of being humiliated [89,97,101,102]. Thus, whether or not high levels of interoceptive awareness are adaptive seems to, at least in part, rely on how internal sensations are interpreted.

Biofeedback training teaches individuals how to regulate their physiological activity [35], and the benefit of biofeedback is often attributed to the acquisition of these self-regulation skills. This belief is rooted in the large body of research dedicated to linking specific emotional states to specific physiological changes [36]. The strong physiological ties of anxiety make it a particularly fitting target for biofeedback, as biofeedback training addresses a wide array of physiological processes. The most prominent forms of biofeedback that are used for anxiety treatment and stress management are HRV biofeedback and neurofeedback.

Despite the fact that teaching individuals to change their physiology is central to both HRV biofeedback as well as neurofeedback, there is no convincing evidence as to whether these changes are indeed essential for positive treatment outcomes (Figure 1, path C). In fact, early evidence suggests that physiological changes may only account for a small percentage of variance in clinical outcomes [133,134]. In current efficacy studies, physiological changes are treated as secondary to clinical changes [39], with many studies failing to check whether subjects have been trained to criterion, thus lacking a demonstration that physiological responses have truly been altered [43]. When changes are monitored in neurofeedback, many individuals seem unable to effectively control their physiological activity [135]. However, even in the absence of significant physiological changes, or even when the change was in the wrong direction, significant clinical improvements can still be found [131]. In HRV biofeedback training, similar findings indicated improvements in symptoms of depression and anxiety, although no significant changes in HRV could be established [136]. In addition, improvements in anxiety were found even when sham feedback was given [137]. This could indicate that certain expectations or perceptions of control in relation to biofeedback training may be important determinants for the treatment outcomes of biofeedback [62,135]. However, empirical evidence regarding specific working mechanisms and possible influencing factors related to physiological changes in biofeedback training is lacking.

In the current biofeedback literature, the efficacy of biofeedback is attributed to the implementation of operant learning reinforcement and meditation techniques as well as changes in interoceptive awareness and physiology (Figure 1). Although there is indeed a wealth of theoretical speculation about the importance of these factors, the empirical evidence linking these processes to biofeedback outcomes remains scarce. Equally important is the fact that there are likely important mechanisms that mediate the effectiveness of biofeedback that have not yet been investigated. Specifically, we propose that there is a set of cognitive appraisals that acts as a mediator between previously identified processes (eg, interoceptive awareness and physiological change) and anxiety reduction from biofeedback training. In the section Cognitive Mechanisms Underlying Biofeedback Processes, we introduce this set of cognitive mechanisms and discuss the importance of including them in a more comprehensive framework for understanding biofeedback training.

When considering appraisal mechanisms by which biofeedback may work to regulate anxiety, appraisals about the self are particularly relevant [140,147]. For example, someone’s uncertainty regarding their ability to cope with a situation is highly predictive of anxiety [148]. Whether someone succeeds in self-regulation may therefore strongly depend on their self-efficacy, in other words, the belief that they can do it. Various meta-analyses have shown that self-efficacy is important for self-development, adaptation, and change [149]. Moreover, high self-efficacy is linked with better emotion regulation skills and general psychosocial functioning [150]. In contrast, low levels of self-efficacy are accompanied by high levels of anxiety [151-153]. In youth, self-efficacy has been found to be predictive of the development and maintenance of affective disorders [154,155]. Furthermore, self-efficacy is an important predictor of treatment outcomes for panic disorder [156]. Given the importance of self-efficacy in anxiety regulation, it is likely a key mechanism for effective biofeedback training.

An additional way in which self-efficacy may be tied to biofeedback is by changing the locus of control. Locus of control refers to the degree to which individuals believe that they themselves have control over the outcome of events in their lives (internal locus) as opposed to forces outside of their control (external locus) [157,158]. People with high self-efficacy generally have an internal locus of control, believing that their own actions and decisions shape outcomes. In contrast, people with low self-efficacy have an external locus of control, often viewing their lives as being beyond their control [149]. Control is an important factor that determines distress and anxiety in a given situation. Specifically, the less someone feels like they are in control, the more anxious they become [159-163]. Moreover, perceived control has been shown to predict outcomes of mental health therapy, with stronger feelings of control being linked to better outcomes [164]. The amount of control someone feels in biofeedback paradigms may therefore influence the success of the training [60,162,164].

Over the course of biofeedback training, with repeated practice and continued feedback on physiological changes, participants in training may increasingly believe that they themselves can influence the outcome of the intervention, in turn leading to a decrease in distress and anxiety. If biofeedback training indeed leads to an increased internal locus of control and locus of control directly influences anxiety, this may also explain why positive therapeutic outcomes of biofeedback can still be present even in the absence of significant physiological changes or when sham feedback is given.

Appraisals of the self often interact with more specific appraisals regarding the situation at hand to determine how anxious someone is and whether they are able to regulate their anxiety. For instance, when personal resources are perceived as greater than the situational demands posed by a stressor, the situation is likely to be appraised as a challenge. However, when situational demands are perceived as exceeding personal resources, the situation is more likely to be appraised as a threat [36,165-167]. Therefore, someone with high self-efficacy is more likely to view difficult tasks as something to be mastered (ie, a challenge) rather than something that should be avoided (ie, a threat) [168].

Challenge and threat appraisals are both related to activation in the SNS. However, they differ in how they prepare the body for action. When individuals enter a threat state, the body responds by activating the hypothalamic-pituitary-adrenal axis, leading to increases in cortisol production. In addition, downstream vascular resistance increases and cardiac efficiency decreases. These changes prepare the body for damage of a physical or social nature. In contrast, when individuals are in a challenge state, there is an increased activation of the sympathetic-adrenal-medullary axis, which results in increased oxygenation of the bloodstream to the brain and peripheral sites and vasodilation, leading to increased cardiac efficiency. These changes prepare the body for approach-oriented behavior [166].

In biofeedback training, the goal is to change a person’s physiological activity from an erratic state to one where there is increased cardiac efficiency and synchrony between the parasympathetic and sympathetic branches of the ANS, that is, a state of autonomic balance or homeostasis [39,58,72,118]. However, the extent to which homeostasis is maintained may not solely depend on practice and feedback but may also be determined or moderated by how physiological activity is appraised over the course of biofeedback training. According to the threat-challenge model, these appraisals are key to changing physiology, cognition, and behavior in stressful situations [166,169]. Indeed, reappraising stress arousal as helpful rather than harmful has been shown to effectively reduce attention bias to threat cues and improve physiological functioning, resulting in decreased vasoconstriction and increased cardiac efficiency [170]. Thus, another way in which biofeedback may work is by helping individuals to shift from interpreting their physiological arousal as being indicative of a challenge rather than a threat.

Figure 2 summarizes our integrative model of biofeedback effectiveness on a real-time scale. We propose that the relationship between interoceptive awareness and anxiety regulation could be mediated by changes in a person’s moment-to-moment appraisals (Figure 2). Focusing on these real-time processes could potentially address conflicting results in past research on biofeedback. On the one hand, increasing interoceptive awareness through biofeedback is thought to facilitate anxiety regulation. On the other hand, we also know that an increase in physiological awareness can lead to an increase in anxiety [94,95]. Examining how cognitive appraisals may interact in a feedback process in real time with interoceptive awareness helps us make sense of these contradictory findings.

At the start of biofeedback training, participants with heightened anxiety symptoms may begin with low interoceptive awareness, and they may not yet be able to effectively self-regulate. However, as biofeedback training continues, awareness of physiological signals may increase as participants begin to detect changes in their physiology. Importantly, however, this increase in awareness may only have a positive effect on self-regulation when certain appraisals are elicited. For instance, when a person becomes aware of physiological sensations and changes related to their anxiety but they have low self-efficacy, their ability to effectively self-regulate may be impeded and give rise to further anxiety. Alternatively, a person with higher self-efficacy who becomes increasingly aware of changes in their physiology may feel more competent to focus on the training and regulate that physiology. Thus, appraisals such as self-efficacy and locus of control may influence the real-time effects of biofeedback training within individual sessions. However, we argue that for lasting improvements in anxiety to be realized through biofeedback, repeated sessions are needed for locus of control, self-efficacy, and threat-challenge appraisals to change and stabilize into more resilient patterns.

Some mechanisms of change may already occur within a single biofeedback session, but it is likely that repeated exposure is essential for changes to become automated and internalized. Specifically, we suggest that only when practice is provided in such a way that the proposed changes are optimized will changes generalize to outside the context of the training and influence developmental outcomes such as mindset, stress reactivity and recovery, and trait anxiety (Figure 3).

Even if someone may become increasingly aware of their internal state while engaging in biofeedback training, it may still be difficult to notice changes in physiology in someone’s daily life. Although continued practice may increase the likelihood that changes and skills transfer to outside the training context, this process may be further enhanced by using wearable self-monitoring technology. Wearable technology is becoming increasingly easy to integrate in our daily lives [9]. There are already a large number of mobile phone apps available that are either linked to internal sensors (eg, global positioning system or pedometer) or external sensors (eg, heart rate sensors), which can generate real-time information or provide daily, weekly, or even yearly overviews of a person’s activities, mood, and physiology. Using these wearables may not only increase someone’s interoceptive awareness but may also provide information regarding the contexts in which certain internal changes are most likely to occur. Furthermore, providing an overview of how someone has changed over a certain period may foster a growth mindset as it emphasizes their malleability. Finally, reports provided by wearables may also be valuable for clinicians as they can more closely monitor their clients’ improvement or deterioration and identify contexts in which certain exercises or interventions may be most effective.

At this point, it is important to emphasize that although interoceptive awareness is an important aspect of emotion regulation, merely increasing this awareness may actually result in an increase in anxiety [94,95]. Continuously providing feedback on whether a desired level of physiological activity is reached may lead to extreme attention vigilance and be counterproductive for those anxious people who already focus too much on their internal states. Therefore, fostering a balanced awareness may be a better approach for biofeedback training for anxiety reduction. To achieve this balance, the training should be designed in such a way that other elements in the environment provide opportunities to disengage from attending to physiological activity. For instance, in the biofeedback video game Dojo [49-51], there is a figure of a seated human with a heart displayed in the corner of the computer screen that changes color based on the player’s HRV. Specifically, the heart changes in color from red to orange to green, with green being the most optimal level of HRV. Using this type of stimulus clearly communicates changes in the physiology of the player; however, having it continuously present may result in a singular focus on their internal state. Thus, making the training context more engaging in different parts of the game and leading participants’ attention away from the heart representation for short periods is likely a more optimal strategy. It may actually be most effective to move away from explicit representations of heart rate, pulse, and other bodily representations of stress and arousal.

If the digital context where self-regulation is practiced is engaging enough (eg, by offering thrilling or beautiful environments to explore where stimuli respond to changes in physiology), these design features may craft a more balanced attention landscape. Practicing in this type of landscape may change from a vigilant form of interoceptive awareness to a state of concentrated relaxation in which participants are aware of their arousal, but instead of fighting it or losing themselves in it, they are in control and move along with it. Gamers have described this state of concentrated relaxation as feeling completely focused and concentrated but at the same time feeling like they are able to let go of both [182,183]. For example, the game Nevermind [52], in addition to using explicit representations of HRV (a heart changing in color), also uses atmospheric changes that reflect the player’s level of arousal. Players find themselves in an eerie environment that becomes more or less unsettling depending on how well the player is able to self-regulate their arousal. The screen becomes more distorted or ominous noises become even more pronounced as the person feels more anxious. Moreover, specific scenic elements are used, such as the kitchen that floods with milk, the more anxious the player becomes. In addition, the game also includes challenging puzzles that require the player’s attention, which means they cannot singularly focus on changes in their HRV [53]. Another exemplary biofeedback app is DEEP, which integrates atmospheric forms of biofeedback [54]. In DEEP, players receive feedback on their breathing by means of a circle that enlarges and shrinks in accordance with the player’s breath. Although this circle is always directly in the player’s line of sight, the game environment is designed in such a way that it grabs the attention and invites players to explore the beautiful underwater world with its vibrant flora and fauna that respond to players’ breath. For example, some of the plants mirror players’ in and out breath by growing and shrinking or becoming more or less bright [55,56]. As these examples demonstrate, there is a great deal of promise in integrating biofeedback into engaging environments where participants are made aware of their internal states but feel competent and able to focus on the task at hand.

One of the foremost sources of information from which we derive our sense of self-efficacy is the genuine experience of mastery [184]. Specifically, our feelings of competence and confidence are largely based on earned failures and successes [184,185]. A sense of mastery can be felt when a task’s level of challenge is aligned with our perceived level of competence. In the developmental literature, this ideal balance between challenge and guidance is known as the zone of proximal development [186]. Putting players in this sweet spot of challenge to keep them motivated and engaged is something that video game designers excel at, as is demonstrated by the large number of individuals that play games and by the number of hours that they play them [187-189].

This balance of challenge and difficulty experienced in games is also referred to as a state of flow. Flow is experienced when individuals are immersed in a task, with energized focus, full involvement, and enjoyment in the process of the activity. This flow state coincides with a high sense of control and a loss of self-consciousness [181]. Notably, flow has also been linked to increased self-esteem and decreased anxiety [190]. To achieve a flow state, clear goals must be set beforehand and there must be a possibility to monitor progress. In addition, clear and immediate feedback needs to be given with the aim of balancing the perceived challenge and one’s perceived skills [180]. Some of these conditions are already an inherent part of biofeedback, such as immediate feedback and the ability to monitor progress. However, most biofeedback training programs are not specifically designed with flow states in mind. Video games are designed in such a way that flow states are likely to occur as the difficulty level dynamically adapts to the player’s current level of mastery, ramping up whenever the skill of the player increases [191]. Although some biofeedback programs do in fact dynamically adjust the level of difficulty, participants may still not experience mastery. To accomplish a true sense of mastery, it is not only important to balance the level of challenge but also to balance rewards and acknowledgments of success. Providing rewards when appropriate instils a sense of accomplishment and pride. However, if challenges are too easily overcome and rewards are disproportionately doled out, the same sense of accomplishment is not felt due to ill-gotten gains [192]. When designing biofeedback interventions, it may therefore be beneficial to focus on facilitating flow and genuine mastery experiences, thereby creating experiences where participants feel competent, in control, and focused on the goals in the training but are still aware of their physiological arousal.

Clinical interventions often focus on decreasing physiological arousal as increases in arousal (eg, quickening of the breath and an increase in heart rate) are often associated with negative emotional states such as anxiety. Interventions that focus on changing appraisals of stress from a threat to a challenge do not focus on reducing arousal or inducing relaxation per se, but instead aim to change the way the arousal is interpreted and experienced [169,170,193]. Interventions aimed at changing threat-challenge appraisals either directly target arousal interpretations or target higher-level belief systems such as growth mindset. When directly targeting threat-challenge appraisals, participants are often provided with information in a written, oral, and sometimes video form, which demonstrates that arousal is not harmful but rather a functional response of the body that aids performance in challenging and stressful situations. Interventions focused on fostering growth mindset use similar strategies but have an additional emphasis on people’s ability to change (for a complete review of threat-challenge manipulation strategies, refer to the study by Jamieson et al [193]). Appraising arousal in a positive manner has been shown to lead to more adaptive stress responses [166] and even better academic performance [169,175].

To encourage challenge appraisals, feedback messages could be implemented in biofeedback training before and throughout biofeedback training, which frames arousal as adaptive. However, a downside to integrating these types of messages is that they are fairly explicit, which may clash with the more implicit, bottom-up type of feedback that is delivered through biofeedback. Therefore, if feedback messages are integrated, they should nicely blend into the environment. One successful example of this type of integration can be found in the phone-based companion app called #SelfCare [194]. In this app, which aims to provide relaxation and mindfulness training, users are invited to stay inside their virtual bedroom where they can interact with objects in the environment and play some minigames. When interacting with the app, messages pop up on screen that encourage the user to observe how they feel, such as “Is there tension somewhere?,” “Where does this tension come from?,” and “What can we learn from it?.” These types of messages frame bodily sensations and arousal as informative, similar to threat-challenge manipulation strategies [193]. Furthermore, the user is addressed in a comforting and encouraging way, such as “There’s no hurry, we can stay as long as we’d like.” However, the messages do not feel patronizing or forced because they are phrased in a personal and relational way and because they blend in with the soothing and safe atmosphere of the environment. Thus, although biofeedback training may benefit from integrating feedback messages that frame arousal as functional and encourage a growth mindset, we argue that caution must be given so that it is well balanced with the design.

In previous sections, we posited that merely interacting with biofeedback could already help individuals to appraise increasing arousal from a perspective of challenge as they witness that arousal as something they can control. However, the environments in which participants train are often fairly neutral, and participants are rarely given the opportunity to practice in situations that increase their arousal [57]. We argue that especially when it comes to anxiety, it is important that individuals practice in contexts where feelings of anxiety are actually triggered, similar to exposure exercises in anxiety treatment [195]. When practicing self-regulation within contexts that evoke arousal, the likelihood increases that skills are transferred outside of the training to stressful situations where these skills are needed most. Moreover, when practicing within these contexts, reappraisals of arousal are more likely to occur than practicing in a neutral or relaxing setting [196,197]. Some biofeedback-based games such as Nevermind [53] and Mindlight [198] have already adopted the use of anxiety-inducing settings. The latter game, Mindlight, has also been shown to be as effective in relieving anxiety as CBT [199]. Ideally, stressful training environments are designed to increase arousal, but they also allow individuals to directly interact with and exert control over the environment so that they feel immersed, in control, and motivated to continue.

Although biofeedback training may already implicitly change participants’ mindset as feedback with regard to change and progress is constantly provided, incorporating design elements that explicitly facilitate a growth mindset may further strengthen the training’s effects. Various approaches have been used in previous interventions to foster a growth mindset, such as providing individuals with excerpts from scientific texts, showing them educational videos, or giving them writing exercises. All these strategies focus on priming individuals with or explicitly displaying information that emphasizes people’s ability to change [175,200]. These mindset interventions have shown positive results in improving physiological responses, work performance, affective responses, and health outcomes (for overview, refer to the study by Jamieson et al [193]).

Although biofeedback training may already foster a growth mindset by providing feedback with regard to physiological change, the training may further benefit from including elements that are explicitly designed to emphasize a person’s malleability. For instance, feedback messages in initial biofeedback sessions could be adapted so that they not only indicate how someone is currently doing but also highlight their rate of improvement and that they have the potential to further improve. Furthermore, participants can be provided with regular opportunities to check their progress throughout the course of the training. For instance, they could be shown a visual representation of all changes that they went through from the first session onward. Specifically, feedback messages in the initial session can be used to demonstrate the possibility of change, and showing explicit evidence of a person’s progress can serve as a more concrete representation of improvement. Furthermore, for these growth mindset microinterventions, it is essential to structure the specific feedback messages such that they reinforce and focus on the participant’s efforts instead of emphasizing that the participant is a competent person [176]. For instance, messages should not focus on personal traits such as “You’re awesome!” or “You’re so smart!” Instead, these messages should highlight progress and effort such as “You’re doing great, it’s clear that you are improving!” or “Keep at it, you’ll figure it out!” Adding these types of feedback messages could result in someone adopting a growth mindset regarding their ability to self-regulate.