1.1.

Background

The concept of virtual reality (VR) can be tracked back to 1935 when American science fiction writer Stanley Weinbaum envisioned a device resembling goggles that could allow the wearer to experience “sight and sound, taste, smell and touch” and to interact with characters in a story. Today, modern VR technologies implement this concept by employing visual display units and projected environments to generate images, sounds, and other sensations that closely resemble reality to immerse a user in a virtual space.¹ Fully immersive VR is most commonly achieved through the use of a head-mounted display (HMD), which contains small, high-resolution screens positioned in front of the user’s eyes enclosed with a goggle-like apparatus (Fig. 1). Other types of immersive VR utilize projections on screens installed on three or more surfaces surrounding the user within a cube-like room to enable an immersive feeling, a technique known as the cave automatic virtual environment (CAVE). By contrast, non-immersive VR experiences are generally realized through a conventional computer screen and an interface that would allow a user to observe or interact without altering the physical surroundings. For the purpose of this review, our focus is on studies that employ fully immersive VR (iVR) technologies.

Fig. 1

Depiction of immersive VR combined with fNIRS. (a) HMDs and (b) CAVE, compared with (c) non-immersive VR based on computer screens.

The recent miniaturization of HMD-based iVR and the increased affordability of VR technologies have increased its popularity, extending its use beyond recreational purposes into scientific research and healthcare investigations.² Although some early studies question the full reliability of iVR to mimic a real environment in terms of human performance,³ the advances in iVR have been successful at addressing those shortcomings by providing more realistic viewpoints and creating the sense of presence. In particular, iVR provides a useful tool in the study of brain functions and therapy, as perception, vision, and vestibular information to produce the feeling of presence and sense of immersion are constantly collected and analyzed by the user’s brain. Notably, iVR allows the researchers to deliver and precisely adjust multisensory stimulations to the brain in a safe and highly controlled environment that is often not feasible in real-world settings.⁴ Moreover, iVR may offer a cost-effective alternative to conventional stimulation methods to establish a similar realistic feeling, especially under the circumstances in which conventional methods are limited or unavailable (such as in underequipped hospitals/labs or at home).^5,6 These applications encouraged methodological advancements that integrate iVR with neuroimaging techniques, enabling the delineation of users’ brain responses during their immersive virtual experiences.

1.2.

Brain Measures

To date, electroencephalography (EEG) has predominately been integrated with iVR, typically through the installation of a VR HMD directly above the EEG cap.⁷ In cognitive and behavioral research, iVR-EEG has been employed to investigate various domains, including environment awareness, spatial navigation, attention, stress, emotion, and memory functions,^8–11 by analyzing the evoked potential patterns and band power alterations associated with designed tasks in virtual environments. Another substantial body of iVR-EEG literature is focused on the development of neurofeedback systems and brain-computer interfaces, notably in training motor and executive functions for limb control, as seen in neurorehabilitation applications.^12,13 In parallel, other groups have explored the feasibility of conducting functional magnetic resonance imaging (fMRI) scans with the user immersed in a virtual environment. Such studies generally applied MRI-compatible VR HMDs¹⁴ or utilized computer screens/mirrors placed at a close proximity to the user’s head inside an fMRI coil.^15,16 Despite the progress, current iVR-EEG setups often necessitate compromises related to evoked potential signal complexity, reduced monitoring area, and susceptibility to electrical interferences.^17,18 iVR-fMRI faces challenges from the high cost of the MRI console and MRI-compatible iVR devices, as well as the complicated implementation and synchronization requirements.^15,19 Furthermore, EEG and fMRI are vulnerable to motion artifacts, which may be common in many VR applications involving large ranges of head or limb movement.²⁰ Finally, fMRI scans restrict the user to a supine position within a noisy MRI room, potentially diminishing the level of immersion experienced by the user in the simulated virtual environment.²¹

Over the past decade, functional near-infrared spectroscopy (fNIRS) has attracted much attention in iVR studies.^22,23 fNIRS is a noninvasive, flexible, and low-cost brain imaging technique that employs low energy near-infrared light to quantify cortical hemodynamic variations in terms of oxygenated hemoglobin (HbO) and deoxygenated hemoglobin (HbR) concentration changes. Therefore, fNIRS is generally less affected by electrical interference, making it highly compatible with the operation of HMD or CAVE equipment. Many available fNIRS devices feature high compactness and portability, simplifying the iVR-fNIRS setup for use in daily life scenarios (e.g., at home) or within complex clinical settings.²⁴ The higher motion tolerance of fNIRS allows participants to undergo brain measures while maintaining a certain degree of mobility,²³ which, in combination with its silence during operations, can significantly enhance users’ sense of immersion and extend the types of stimulations and tasks being administered.

In this paper, we review published work that integrated iVR and fNIRS in a concurrent setup and discuss the following topics: (1) the design and technical implementation of different iVR-fNIRS systems and studies; (2) major applications of iVR-fNIRS in neuroscience research and therapy; and (3) the advantages, current limitations, and future prospects of iVR-fNIRS. Based on the evidence, we provide an evaluation on the feasibility and usefulness of the combined iVR-fNIRS technique.

3.1.

iVR Implementation

The first attempt to install an iVR HMD along with fNIRS optode arrays was conducted by Seraglia et al.,²⁷ who adapted a heavily modified V8 Research HMD fixed onto a bicycle helmet [Fig. 3(a)]. As VR HMDs have evolved to become more compact and affordable, recent studies predominantly employed commercially available HMDs. In some cases, modifications to the HMD or head strap have been necessary to accommodate fNIRS measurements, particularly in the prefrontal area [Fig. 3(b)]. Among the commercial VR HMD models, the most frequently used were the HTC Vive (HTC Corp., New Taipei, Taiwan), featured in 31% of the studies, and the Oculus Rift (Meta Platform Technologies, Menlo Park), employed in 28% of previously published work (Fig. 4). Standalone iVR HMDs (also known as all-in-one HMDs) such as the Oculus Quest, which incorporate built-in processors without any wired connection to a control computer, may be more favorable in studies that prefer a fully wireless and portable iVR-fNIRS setup⁴⁹ or involve subject movement.⁵⁵ CAVE-like iVR has been generally established with multi-screen displays or wide-angle projectors (Fig. 5). Compared with HMD, CAVE-like iVR-fNIRS studies are less common, likely due to the relatively higher costs and spatial demand. However, CAVE-like environment could offer distinct advantages, such as the ability to accommodate additional equipment or components (e.g., a full-size driving simulator) to further enhance user immersion.^57,61 More discussions about HMD versus CAVE are provided in Sec. 5.2.

Fig. 3

HMD iVR-fNIRS design in previous studies. (a) The first iVR-fNIRS combined setup by Seraglia et al.²⁷ using a V8 Research HMD fixed on a modified bicycle helmet. Reprinted with permission under the CC-BY license. (b) Image of the HMD from Landowska et al.,³¹ in which modifications to the top part of the HMD have to be carried out to accommodate the installation of fNIRS optodes on the forehead. Reprinted with permission from Springer. (c) A virtual classroom environment with controlled distractions implemented by Wiebe et al.⁵⁶ with HMD iVR to study adult attention deficit hyperactivity disorder. Reprinted with permission from John Wiley & Sons. Anyone wishing to use this figure will need to contact John Wiley & Sons publishing company directly.

Fig. 4

iVR products and use. (a) Brands and models of HMDs used in previous HMD iVR-fNIRS studies. (b) Methods for iVR task/stimulus development.

Fig. 5

CAVE-like iVR-fNIRS design in previous studies. (a) Hoppes et al.⁵⁸ utilized CAVE iVR and optical flow stimulations to explore brain responses in visual vertigo patients. Reprinted with permission under the CC-BY license. (b) A CAVE-like iVR-fNIRS setup was employed along with a full-size driving simulator to enhance subject immersion in the study of cognitive demands during driving.⁵⁷ Reprinted with permission under the CC-BY license. (c) de Boer et al.⁶⁶ presented a proof-of-concept design using CAVE-like iVR to create out-of-body experiences. Reprinted with permission under the CC-BY license.

The induced iVR stimuli can be either “passive” (or “observational”) or “interactive.” Passive iVR involves immersing users in a virtual environment through HMD or CAVE, but their activities are limited to mainly observation and exploration. In previous iVR-fNIRS studies, passive stimuli were often delivered through the playback of pre-recorded three-dimensional (3D) videos or display of pre-configured virtual scenes. They were more prevalent in studies with a primary goal of providing an observational experience of distinct virtual environments. Conversely, interactive iVR tasks involved a higher degree of user interaction within VR, including object manipulation, action execution, and the ability to control/influence the course of events using additional hardware such as VR controllers. Those tasks were often employed in studying user behavior (e.g., in neuropsychological studies), as well as cognitive control. Most iVR-fNIRS studies that utilized in-house developed iVR tasks or stimuli opted for the open-source Unity 3D engine (Unity Technologies, San Francisco) and used C# programming language⁶³ [Fig. 4(b)]. This is potentially because of its cross-platform compatibility with various types of VR headsets, including those manufactured by Oculus and HTC; its user-friendly development interface that includes pre-packaged virtual object assets and templates; and the abundance of supportive resources available to developers.⁶⁴

In 59% of the reviewed studies, multiple iVR stimuli were delivered using a block design within one data acquisition session at a relatively constant time interval or through several sessions. Notably, 38% of the previous work employed a single continuous stimulation task, primarily to simulate real-life scenarios without habituating participants to the created virtual environment (e.g., in the study of phobia,^31,41 pleasure,³⁴ or creativity³⁷).

3.2.

fNIRS Implementation

Previous iVR-fNIRS setups incorporated a wide range of fNIRS devices with regards to brands and models, optode types, and montages, demonstrating the adaptability of different fNIRS systems in such study designs. Most of the work utilized a continuous wave fNIRS system, which maintains steady illumination of brain tissue and detection of transmitted near-infrared light intensities, and yielded relative changes of HbO and HbR concentrations through the differential pathlength approach. One study employed a frequency domain fNIRS system (specifically the ISS Imagent from ISS Inc., Champaign).²⁷ However, the study only analyzed direct current component of the optical signals, resulting in relative hemoglobin concentration outputs.

A smaller number of fNIRS channels were installed in iVR-fNIRS studies using HMDs (mean = 22, ranging from 1 to 52) compared with those with CAVE-like iVR environments (mean = 46, ranging from 20 to 78). This may potentially be due to the competition of surface space on subject head between fNIRS optodes and iVR HMDs. Targeted brain regions included primarily the prefrontal cortex (in 34 studies), most often the dorsolateral prefrontal cortex (DLPFC) and the frontopolar cortex (FPC) [Fig. 6(a)]. Other brain regions of interest that were frequently involved were the temporoparietal areas (10 studies), the sensorimotor cortex (seven studies), and the visual cortex in the occipital lobe (eight studies).

Fig. 6

fNIRS measures in iVR-fNIRS studies. (a) Depiction of the major brain regions of interests covered by fNIRS. (b) Data presentation methods in previous iVR-fNIRS studies.

The analysis and interpretation of data in previous VR-fNIRS work largely relied on direct measures of relative changes in HbO and/or HbR concentration amplitudes [Fig. 6(b)]. Comparisons were conducted on either hemoglobin response peak values, areas under curve, or beta values and $t$ statistic values extracted from a general linear model (GLM) analysis assuming a canonical shape of brain hemodynamic response. In addition, four studies explored the changes in functional connectivity of remote brain regions by computing the Pearson’s correlation coefficients between fNIRS signal time courses of different channels.^38–40,43

4.1.

iVR-fNIRS in Cognitive Neuroscience

iVR-fNIRS has been employed in cognitive neuroscience research, offering insights into the neural correlates of various processes including cognitive control,^27,39 prospective memory,²⁹ working memory,⁶⁵ and attention.⁴⁹ iVR can create controlled testing environments that isolate the subjects from external interferences; therefore it was particularly useful in studies that required high levels of subject attention and engagement. For example, Zapała et al.⁴⁹ designed attention and working memory tasks in iVR and reported a higher accuracy in distinguishing users’ attention state and resting state with fNIRS-measured PFC signals compared with previous studies that did not utilize iVR. Conversely, iVR-fNIRS could facilitate the study of attention and cognition-related deficits through controlled distractions. The recent work of Wiebe et al.⁵⁶ sought to assess adult attention-deficit hyperactivity disorder (ADHD) with combined iVR, EEG, and fNIRS by immersing their patients in a virtual classroom with induced visual, auditory, or audiovisual distractions [Fig. 3(c)]. A proof-of-concept neurofeedback system was designed in another study²⁸ to train adults who were highly impulsive for ADHD using cognitive control tasks in a simulated virtual classroom. Their results showed a significant reduction in subject impulsive behaviors during follow-up tests as well as improved abilities to regulate PFC activities.

Several iVR-fNIRS studies have investigated brain-level interactions related to cognitive loads in dual or multiple task experiments, leveraging the ability of iVR to precisely deliver multisensory stimulations in a controlled environment. One such application explored the brain resource demands associated with driving^57,61 [Fig. 5(b)]. Using a CAVE-like iVR setup and a driving simulator, healthy participants were requested to engage in multiple subtasks involving visuospatial attention (lane width change) and working memory (vehicle speed adjustment) functions during realistic highway driving scenarios.⁶¹ Simultaneous fNIRS measurements revealed that brain activity changes in the DLPFC and the parietal lobe were dependent on both participants’ visuospatial attention levels and working memory loads in high-demand driving situations, suggesting significant interactions among the underlying neural processes and competition for brain resources. In a separate study, Stojan et al.⁶² investigated brain activation changes from single-task to dual-task walking in older adults with working memory and inhibitory control tasks. They reported increased activations in the ventrolateral prefrontal area and parietal lobe accompanied by deteriorated task performance, indicating neural inefficiency in older adults under heavy cognitive loads.

iVR perhaps offers an unparalleled platform for researchers to manipulate a user’s perception of space and time. In Cho et al.,⁴⁴ the authors described their virtual prism adaption platform integrated with iVR and fNIRS for correcting unilateral spatial neglect in stroke patients. In this setup, a virtual hand was created and was intentionally misaligned with the user’s actual hand. The virtual hand was then used to direct the user’s hand to point at targets placed at the neglected side of space. Validations in healthy subjects revealed significant activations in the DLPFC and the frontal eye fields, both components of the dorsal attentional network. Other intriguing applications included the use of VR and live streaming images to induce visual illusions for studying out-of-body experiences⁶⁶ [Fig. 5(c)]. These discussions highlighted the potential of combining iVR and imaging techniques in delineating brain functions in rare and unpredictable scenarios that may be challenging to replicate in real-world environments.

4.2.

iVR-fNIRS in Behavioral Research

iVR-fNIRS finds applications in the study of various human behaviors, such as decision-making,^36,42 creativity,³⁷ and self-expression.⁴⁷ In an iVR-simulated food selection scenario, Cheah et al.³⁶ explored the role of the inferior and orbital PFC in regulating user’s choices between high-nutrition-density and low-density-nutrition foods. Two recent studies investigated how emotion may influence decision-making regarding exercise and physical activities.^34,52 Their findings showed that individuals exposed to more pleasant exercising environments exhibited weaker brain activations in the DLPFC, reflecting less cognitive effort in retaining a positive affect and exercise interest. In the study of creative behaviors, Wang et al.³⁷ placed a virtual wall in iVR to obstruct a corridor, and discovered that subjects encouraged to break the wall performed better in subsequent creativity-demanding tasks and exhibited a lower level of brain activations in the medial PFC during the tasks. Similarly, Kaimal et al.⁴⁷ reported reduced activations in the anterior PFC during creative self-expression drawing in iVR compared with rote tracing. These results implied that the frontopolar area might potentially be involved in rule-based and self-restrictive behaviors.

iVR-fNIRS was also used to unveil the functional aspects associated with learning of knowledge/skills using new VR-based teaching tools in education. Lamb et al.³² compared brain activity levels across different types of teaching methods in a simulated biology class and observed higher HbO changes in the DLPFC and better learning outcome during iVR-assisted interactive practice than the more conventional video lectures. Another study on the training of spatial visualization abilities in iVR reported similar activations in the DLPFC and orbitofrontal areas, which were positively correlated with task difficulty and modulated after practice and familiarization of the task.⁵¹ These results demonstrated the role of DLPFC in critical thinking, memory, and motor control.⁶⁷ Several studies have employed iVR-fNIRS to assess user performance in job-related skill learning, such as in basic life support training,³⁵ industrial shutdown maintenance,³⁸ and firefighter pipe operations.⁴³ Those investigations reported a consistent increase in brain activities in the DLPFC during skill acquisition and retrieval, which were enhanced under induced stress (e.g., adding a time limit) and attenuated after repetitive training.³⁵

iVR has been applied in treating specific phobias and anxiety disorders by setting up virtual environments that expose the patients to their feared objects or environments in the absence of actual harm. Using fNIRS to simultaneously record the brain responses, previous iVR-fNIRS work has explored the neural correlates of acrophobia (i.e., fear of height)^31,46,60 and public speaking anxiety.⁴¹ In both cases, exposure to feared situations led to greater activations in the DLPFC and the medial PFC, which might reflect modulated emotional processing in the PFC and subcortical areas. With a similar idea, a study employed iVR environments to trigger addicted behaviors.⁵⁰ Their work on individuals with kleptomania revealed distinct PFC activation and connectivity patterns in response to 3D videos of shops/markets compared with healthy controls.

4.3.

iVR-fNIRS in Postural Control and Locomotor Abilities

iVR-fNIRS offers a versatile platform to assess balance and related motor functions with either a treadmill⁴⁸ or the use of optical flow^33,58 [Fig. 5(a)]. Optical flow induces an illusion of movement to a stationary observer by moving objects in a virtual scene relative to the observer.⁶⁸ With a CAVE iVR setup, Hoppes et al.⁵⁸ compared fNIRS-measured brain signals in healthy subjects during exposure to optical flow versus an unchanged visual field. They observed higher brain activation levels in the fronto-temporo-parietal area and the occipital lobe when the subjects viewed optical flow on a fixed surface. These changes were presumably associated with related vestibular activities for postural stabilization. Extending their study to patients with visual vertigo, the same research group reported similar brain activations in the temporal and occipital regions but deactivations in the middle prefrontal area in visual vertigo patients compared with healthy controls, which they attributed to vestibular hypofunction.⁵⁹ Interestingly, Hinderaker et al.³³ employed optical flow stimulations through a iVR HMD and discovered reduced brain activations in the fronto-temporo-parietal areas and the frontal cortex in older adults compared with young adults, particularly under fast optical flow speeds. These brain changes might be associated with reduced ability to process visuosensory information and to maintain postural equilibrium in visual vertigo patients and older adults.

Cybersickness, a specific form of motion sickness triggered solely by visual stimuli (i.e., illusory of self-motion),⁶⁹ has been the subject of two investigations with iVR-fNIRS. In Gavgani et al.,³⁰ the authors immersed healthy volunteers in a virtual rollercoaster ride and observed elevated HbO concentration levels in bilateral temporo-parietal regions among participants who experienced strong motion sickness symptoms. Pöhlmann et al.⁵⁴ conducted a similar experiment involving a virtual rollercoaster scenario but introduced simultaneous working memory tasks within the iVR environment. They observed activations in the right DLPFC of their participants during both the memory tasks and periods when cybersickness was experienced. They proposed that user engagement in cognitive tasks might distract iVR users from cybersickness symptoms, potentially through a competition of attentional resources in the brain. Motion sickness and cybersickess are important issues for iVR.⁷⁰ In Sec. 5.3, we provide further discussions on their implications in iVR-fNIRS research.

Additional studies have explored the perceptual and executive processes associated with sensorimotor functions in various iVR scenarios. For example, Kim et al.⁵³ simulated an open sky space that allowed their participants to “fly freely” with few restrictions on the extents and angles of limb movements. They demonstrated a positive correlation between ranges of participant movements in iVR and the HbO concentrations in their right orbitofrontal cortex. Tian et al.⁵⁵ utilized fNIRS to assess brain conditions in older adults with mild cognitive impairments while they engaged in an iVR video game requiring large extents of upper limb movement. They observed significant lower brain activation levels in the prefrontal and occipital areas in the patient group with lower Montreal cognitive assessment scores, reflecting potentially more impaired motor control abilities.

4.4.

iVR-fNIRS in Pain Management

It has been reported that VR might modulate users’ perception of pain by providing effective distractions and reducing the unpleasantness/distress associated with pain experiences.^71,72 Two iVR-fNIRS studies aimed at delineating the underlying neuronal processes in pain reduction. In Deng et al.,⁴⁵ an immersive video game was employed to divert the attention of study participants while electrical pain stimuli were applied to their back. The results showed significant brain activations particularly in the DLPFC and the premotor cortex. Both regions were suggested to be involved in attention orientation and top-down antinociceptive control.⁷³ Hu et al.⁴⁰ used iVR to explore the brain mechanism of mindful breathing in the modulation of induced thermal pain at the trigeminal nerve. Their findings revealed that meditation raised subject pain thresholds potentially through the enhancement of the brain functional connectivity particularly within the anterior PFC, as well as between the PFC, premotor cortex, and auditory/visual regions. These brain areas are believed to play important roles in regulating attention and high-level integration of multisensory information.⁷⁴

5.1.

Use of iVR in fNIRS Research

The most prominent advantage of using iVR to deliver stimulations/tasks to evaluate associated brain responses is its ability to improve ecological validity in method assessment.⁷⁵ Fully immersive VR setups, such as HMDs or CAVE environments, are capable of providing complex, three-dimensional and realistic testing conditions that are highly comparable to users’ daily environments. Traditional tests on cognitive functions and behaviors have often been criticized for their lack of ecological validity, which can lead to discrepancies between test results and real-life performances.⁷⁶ For example, several assessments of ADHD in children revealed that laboratory assessments of inattention, impulsivity, and overactivity showed only low-to-moderate consistency with measures conducted in more natural at-school or at-home settings.⁷⁷ By contrast, ecological approaches (e.g., conducting a study during lectures given in a real classroom) lack quantitative/normative data, offer less specific assessments, and have a low reliability due to many factors that cannot be controlled.⁷⁸ The recent development of iVR-fNIRS methods, on the other hand, allowed the researchers to bring ADHD patients to a virtual classroom, assessing their attention and memory functions during virtual lectures with precisely controlled distractors.⁵⁶ The use of iVR significantly enhances the verisimilitude and veridicality of the study by providing measurements of brain functions that are representative in users’ normal living conditions and predictive of their daily behaviors outside the test environment,⁷⁹ while maintaining scientific rigor and reproducibility.

iVR can create simulated testing environments or deliver stimulations that may be restricted in real world situations or “impossible” based on the physics laws of nature.⁸⁰ This flexibility allows researchers to explore brain activities during various neurological and neuropsychological processes that were previously deemed difficult or risky in a laboratory setting.⁸¹ Several iVR-fNIRS studies have implemented fully immersive virtual environments and stimulations that could span over multiple dimensions to modify subjects’ visual, auditory, haptic, and other sensory inputs, inducing illusions of flying, driving, out-of-body experiences, or emergency situations.^43,53,61,66 Another application is the study of brain responses during exposure therapy for individuals with phobia or anxiety disorders.⁶⁰ iVR allowed for complex object/situation presentations while offering precise adjustments on exposure type, duration, and dose, overcoming the current barrier of ethical or tolerability concerns in such studies.⁸² Although less explored in the current iVR-fNIRS literature, iVR has the potential to facilitate brain research involving patients with reduced mobility or cognitive impairment by providing tailored virtual environments and stimulations that accommodate specific test requirements, offer new experiences, or be used as alternatives to existing stimulation modalities.⁵ This may hold particular promise for neurorehabilitation, which aims to use repetitive training to promote neural activations in neurology patients to restore motor and executive functions after brain disorders or trauma.⁸³ Indeed, physical therapy combined with iVR exhibits greater improvements in gait and balance than traditional rehabilitation approaches.^84,85 The combination of iVR and brain imaging techniques may lead to further development of patient-specific approaches as a standalone or complementary tool for evaluating and rehabilitating brain functions in these populations.⁸⁶

Finally, iVR offers a compelling advantage in enabling researchers to control the timing and intensity of multiple sensory stimulations while isolating research subjects from unwanted interferences in an immersive and enclosed environment. This provides a robust platform to study the integration and interaction of brain processes associated with simultaneously delivered stimulations or tasks.^87,88 Past iVR-fNIRS studies have focused on assessing the interaction of cognitive control, visuospatial attention, working memory, and motor functions in scenarios such as distracted driving^57,61 and multitask walking⁶² to explore the resource demand and brain performance during demanding tasks.

5.2.

HMD iVR Versus CAVE iVR to Combine with fNIRS

HMD and CAVE represent two distinct approaches to establish iVR.⁸⁹ When compared with CAVE, HMD excels in terms of cost, ease of system setup, and the ability to create a personalized virtual experience with the environment responding to the user’s head movement. HMD iVR headsets are generally affordable, usually ranging from under one thousand US dollars to a few thousands, whereas a complete CAVE setup can cost ten or a hundred times more depending on factors such as size, projection surface, and intended use.⁹⁰ HMD iVR systems do not require large, dedicated study spaces or the complex installation of projectors/screens, making it ideal for applications that prefer a fully portable and flexible iVR-fNIRS setup. Additionally, HMD iVR may induce less ambient light interference with fNIRS signals, as the display screens are normally fully enclosed within the VR goggle.³¹ By contrast, CAVE iVR implements motion tracking cameras to track body movements or as a means of interaction with the virtual space.^91,92 Such cameras (e.g., time-of-flight depth cameras⁹³) often use infrared light, which represents a source of interference for fNIRS. It is worth noting that caution should also be exercised in the case of HMD iVR with the eye tracking function enabled as the tracking cameras may also employ infrared light,⁹⁴ posing potential interference with fNIRS measurements.

On the other hand, CAVE iVR is often considered to be better in creating highly immersed experiences as it offers a wider viewing angle, higher screen resolution, and more freedom of user movement (which may, however, introduce more motion artifacts in the fNIRS data at the same time).⁹⁰ CAVE iVR provides a more natural sense of embodiment as users can see their own bodies during the iVR experience.^31,95 Its spacious environment can accommodate additional pieces of equipment, such as a driving simulator or a flight cockpit simulator, to offer realistic visual and haptic feedback. Unlike HMDs, which are generally intended for a single user, CAVE VR can simultaneously immerse multiple individuals in the same environment,⁹² facilitating fNIRS research that aims at exploring brain coupling among users, as seen in hyperscan setups.²⁴ CAVE iVR also has fewer issues with regards to the competition for space over the user’s head. As seen from previous iVR-fNIRS publications, CAVE iVR permitted higher numbers of installed fNIRS channels (an average of 46 versus 22 with an HMD) and larger sampling areas.

Comparison studies reported mixed results over user behavior and task performance during iVR using HMD or CAVE systems. CAVE iVR was favored in a few early studies because of higher reported levels of presence and stronger emotional responses in subjects,^96,97 whereas others found that participants rated HMD higher in terms of presence and showed better task performance.^98,99 Some studies suggested a minimal difference between the two systems regarding user attention, engagement, and comfort.^95,100,101 These inconsistencies might be partly attributed to the continuous evolution of iVR technologies, particularly the HMDs, which can dramatically impact user experiences. Only one study has directly compared HMD and CAVE setups in the context of fNIRS applications; however, no definitive conclusion was drawn.³¹

5.3.

Limitations in Current iVR-fNIRS Studies and Future Work