Table 3.

Studies classification according to each regulation strategy.

Driving	Reg.	Authors (year)	Title	Stress induction	Implemented technique	Experimental conditions	Participants N (age)	Physiology measures	Subjective measures	Main outcomes
No	AD	Amores and Maes (2017)	Essence: olfactory interfaces for unconscious influence of mood and cognitive performance	–	Odors emitted by a necklace: tea tree, peppermint, rose	Use the necklace during 3 days.	4 (mean = 29) Not naive	–	Comfort, usability, distraction, pleasantness, satisfaction, relaxation, debrief	Participants described the technology as seamless, effortless and relaxing.
No	AD	Ansems et al. (2011)	Smart photo frame for arousal feedback	Video game (snake)	Enlighted photo frame	Participants had to play to a snake game while having a small light ball that displayed colors.	Unspecified	–	Debrief	The users did not feel more stress when seeing a certain color. Colors did not influence the emotions of the users.
No	AD	Daher et al. (2020)	Reduce stress through empathic machine to improve HCI	TSST	Ambient blue light	2 conditions: TSST with light and TSST without light. Participants followed all conditions (counterbalanced).	17 (21–63)	HR HRV	Stress scale	Increase of HRV with blue light, significant reduction for no-light condition when compared to baseline. Participants under blue light reported lower stress.
No	AD (music) RM (biofeedback)	Williams et al. (2015)	Swarm: an actuated wearable for mediating affect	–	Multimodal stimulation (heat, vibration, music) delivered by a scarf according to emotional state	Brief on each module of the scarf, then participants could wear the scarf and evaluate each module	9 (18–61) Not naive	–	Semi structured interview, Usability	Interest of the participants to have a biofeedback device. Participants with disabilities rated the scarf as more useful.
No	RM	Azevedo et al. (2017)	The calming effect of a new wearable device during the anticipation of public speech	TSST	Haptic false biofeedback: wrist worn	2 groups: one with wristband OFF; one with wristband ON, vibrating with a frequency 20% lower than the rhythm measured at rest. Participant are naïve on the purpose of the study.	52 (mean = 26.4) Naive	HR EDA	STAI Ya	Lower EDA for slow HR group during the task. Lower anxiety for slow HR group after speech preparation. Task was also rated as less stressful.
No	RM	Ban et al. (2018)	Relaxushion: controlling the rhythm of breathing for relaxation by overwriting somatic sensation	–	Haptic stimulation: cushion that can adjust its size	2 types of breathing tempo: 7 resp/min or 15 resp/min. Participants are naive of the purpose of the study.	5 (unknown) Naive	BR	Debrief	Participants BR adjusted to the motion of the device. Low level of distractions.
No	RM	Bergstrom et al. (2014)	Using music as a signal for biofeedback	–	Music tempo and volume modulation according to heart rate	3 conditions: listening to pre-recorded music; Sonification biofeedback; Musical biofeedback. Participants are instructed to increase or decrease their arousal level in each condition.	24 (mean 28.2) Not naive	HR BR	Bodily Awareness Questionnaire, Debrief	Facilitating effect of musical biofeedback to modulate arousal through BR, when compared to sonification or music alone.
No	RM	Choi and Ishii (2020)	Ambienbeat: wrist-worn mobile tactile biofeedback for heart rate rhythmic regulation	Physical exercise	Haptic biofeedback: wrist worn	3 tasks: sit still; Sit still after physical exercise; draw with a mouse. Participants had a tactile, visual or auditive stimulation at 60 or 120 beats per minute (bpm).	12 (18–60)	HR HRV	Modality preference	Better effects of tactile stim to reduce stress level: increased HRV and fastest HR decrease after jumping with the tactile stimulation. Higher preference to tactile and lower disturbance level.
No	RM	Costa et al. (2016)	Emotioncheck: leveraging bodily signals and false feedback to regulate our emotions	TSST	Haptic biofeedback: wrist worn	2 groups: one with wristband ON, one with wristband OFF. For the group ON, 3 conditions: vibrations at 60 bpm with naïve participants (vibration group); Vibrations at 60 bpm with participants informed that the vibrations represented HR (slow HR group); Vibrations at real HR (real HR group).	67 (19–30) Naïve/not naive depending on the condition	HR	STAI Ya, STAI Yb, Distraction	Lower anxiety scores for slow HR group. The intervention was rated as not distractive.
No	RM	Costa et al. (2019)	Boostmeup: improving cognitive performance in the moment by unobtrusively regulating emotions with a smartwatch	Mental calculation	Haptic biofeedback: Applewatch	2 groups: one with slow biofeedback (slow HR); one with increased biofeedback (fast HR).	72 (18–25) Not naive	HR HRV	STAI Ya, Distraction	HRV increased for slow HR group. HR decreased for fast HR group. Performance increased for slow HR group. Increased reaction times for slow HR. The technique was not distractive for the participants. Fast HR increased anxiety. Slow HR reduced anxiety.
No	RM	Cuijpers et al. (2019)	Psychophysiological stress control via heart rate entrainment	TSST	Heartbeat-like sound	3 groups: one with slow biofeedback; one with real biofeedback; one with no biofeedback.	30 (unknown)	HR EDA	STAI Ya	No significant results.
No	RM	Fedotchev et al. (2017)	Effects of musical acoustic signals controlled by the subject's eeg oscillators	–	Sound modulated according to EEG oscillations	2 types of sound: one that varies according to EEG, another presenting the same variations together with a 1-Hz rhythm.	17 (23–55)	HR HRV EEG	WAM scale (well-being, activity, mood)	Increase of alpha and beta waves for the condition with the 1-Hz rhythm. Significant increase in well-being and mood.
No	RM	Hamon et al. (2018)	Exploring biofeedback with a tangible interface designed for relaxation	N-back task	Visual biofeedback: ambient light from a flower	4 groups: focused attention required while the light varied according to HR and BR (dynamic light); focused-static light (the light vary at a fixed rhythm of 6 bpm); ambient	36 (mean = 23.8)	HR	STAI Ya USE survey (usability)	No significant results.
						(no focused attention required)-static light; ambient-dynamic light.
No	RM	Harris et al. (2014)	Sonic respiration: controlling respiration rate through auditory biofeedback	–	Music modulation according to breathing rate	2 types of music manipulation: audio tract layering (manipulation of the quality of audio or noise addition).	6 (20–59) Not naive	BR	Survey on attitude toward the technique	Lower BR with audio modification. Participants preferred noise addition over track layering.
No	RM	Kim et al. (2018)	Affective and autonomic response to dynamic rhythmic entrainment: mechanisms of a specific music therapy factor	Stroop task, Quesions, Mental calculation	Music tempo modulation according to heart rate	2 groups listened to music: one group with adaptative tempo that follow participant's HR then gradually decrease; One group with fixed tempo at 70 bpm. Both groups had control condition with no music.	30 (mean = 26.9)	HR BVP	Stress scale, Well-being scale	Adaptative tempo group: Strongest increase in the peripheral blood flow (indicating better stress recovery). Strongest increase in global well-being. Slight stress reduction (not significant).
No	RM	Leslie et al. (2019)	Engineering music to slow breathing and invite relaxed physiology	Oddball task	Music recommendation according to breathing rate	4 conditions: silent; fixed tempo; personalized tempo; personalized envelope.	19 (19–55) Naive	HR HRV BR BRV	–	Decrease in BR across the conditions (baseline > fixed tempo > personalized envelope > personalized tempo). BRV increased when compared with baseline. EDA reduction for personalized tempo. Increase in z-scored interbeat intervals for the condition fixed tempo when compared to baseline.
No	RM	Lopes and Campos (2019)	SCAARF: a subtle conditioning approach for anxiety relief facilitation	Various daily stressors	Haptic biofeedback and sound conditioning: scarf	1 group of participant has to use the scarf at least 10 min/day during 3 weeks. During the first phase, when the user gets stressed, an app can guide him through respiratory exercise, which is conditioned with a sound. During the second phase, the participant only has the sound.	7 (mean = 31)		Debrief	Technology rated as subtle (scarf and sound). Less stress was felt by participants for the conditioning phase.
No	RM	Sato and Moriya (2019)	Respiration rate change induced by controlling the phasic relationship between melodic sound and respiration	–	Music modulation according to breathing rate	6 types of target phases, defined by their shift with breathing patterns, 3 melodic sounds (pop, rock, classic).	10 (20–50) Naive	HR BR BRV	–	Significant differences in BR depending on the phase selected. Higher BR when the target phase is different from participant breathing pattern.
No	RM	Yu et al. (2018)	Delight: biofeedback through ambient light for stress intervention and relaxation assistance	Mental calculation + time pressure	Visual biofeedback: environment light modulation according to HRV	Three groups: control without biofeedback, biofeedback with warm light, biofeedback with cold light.	12 (25–35) Not naive	HR HRV	Stress scale, relaxation scale, distraction	Lower change of HR when compared to control condition. Higher change in HRV for warm BF compared to both control and cold BF. Stress increase more for the control condition than for both BF. Stress increase more for warm than for cold BF. Participants prefer cold BF and find it more relaxing and less distracting
No	RM	Zhou et al. (2020)	The calming effect of heartbeat vibration	–	Haptic biofeedback: vibrations according to heart rate	Participants had haptic biofeedback during 3 sessions composed of resting phases and stimulation phases.	21 (mean = 35.7)	HR HRV	STAI Ya	HRV increased and HR decreased compared to the two rest conditions, in all three sessions. There is a main effect of the sessions, translating into an increase in RMSSD and decrease in HR.
Yes	AD	Hu et al. (2015)	Safedj: a crowd-cloud codesign approach to situation-aware music delivery for drivers	–	Music recommendation according to emotional state	4 groups: 1 without music; 1 with music recommendation by default player; 1 with music recommendation according to user's mood.	48 (unknown)	HR HRV	–	Increased reduction of the fatigue and of the negative moods experienced when the system recommended music.
Yes	RM	Balters et al. (2020)	Calm commute: guided slow breathing for daily stress management in drivers	TSST Heavy metal song	Respiratory guidance through haptics in the seat	2 groups: 1 with haptic guidance; 1 without (control). Counterbalanced order of stress induction methods.	24 (mean = 40.2) Not naive	HR HRV BR	Stress scale, Physical tension, Affect grid (arousal and valence)	Normal driving: decrease of 15% of BR (no significant). Post stress driving: reduction of 25.3% of BR, increase in HRV. Participants expressed to have calming effect and low distraction from the guidance.
Yes	RM	Bhandari et al. (2015)	Music-based respiratory biofeedback in visually-demanding tasks	Video game (racing)	Music modulation by adding of a white noise according to breathing rhythm	4 conditions: no music no feedback (control), music only, auditory biofeedback (only white noise reflecting BR), musical biofeedback (music with white noise).	28 (22–35) Not naive	HR HRV BR EDA	Valence, Calmness	Lower arousal (BR & EDA decrease and higher HRV) for musical biofeedback. Lower EDA for musical biofeedback compared to auditory biofeedback.
Yes	RM	MacLean et al. (2013)	Moodwings: a wearable biofeedback device for real-time stress intervention	Driving context	Visual biofeedback: moving wings of a butterfly on the wrist according to arousal	6 conditions: 2 modalities for butterfly (actuated, stationary) × 3 driving actions (easy, maneuvering, unexpected event). Participants are informed of the purpose of the technology.	11 (unknown)	EDA	Stress scale	Driving performance was better in the actuated condition (driving more safely). Users felt more stress when mood swings was in actuator mode.
Yes	RM	Paredes et al. (2018b)	Just breathe: in-car interventions for guided slow breathing	Driving context	Respiratory guidance through haptics in the seat or voice guidance	2 groups: Manual mode and autonomous mode. Two guidance modalities: voice or haptics, both with a frequency lowered by 30% according to breathing baseline. Two driving scenarios: one stressful urban driving, one highway driving.	24 (18–64)	HR HRV BR EDA	Stress scale, Distraction	No significant differences for EDA. Decrease in BR & increase in HRV with both audio and haptic guidance. No difference in driving performance. Stress reduction trend. Haptic is rated as subtle but participants noted they would not focus on the device if the situation is too stressful. Participant preferred haptic modality.
Yes	RM	Zepf et al. (2020)	Empathicgps: exploring the role of voice tonality in navigation systems during simulated driving	–	GPS voice modulation according to emotional state estimated via EDA	3 voice modalities: congruent (biofeedback adapted to the arousal level); incongruent (BF unadapted to the arousal level); neutral (constant calm voice).	18 (22–58) Not naive		Stress scale, affect grid (arousal and valence), pleasantness of the voice	Brake intensity increased in the incongruent condition. No differences between congruent and neutral. Incongruent voice is associated with lower arousal and higher stress. Congruent voice is associated with higher pleasantness.

AD, attentional deployment; CC, cognitive change; RM, response modulation; SM, situation modification; SS, situation selection; HR, heart rate; BVP, blood volume pressure; HRV, heart rate variability; BR, breathing rate; BRV, breathing rate variability; EDA, electrodermal answer; EEG, electroencephalogram.