The protocol was developed according to the SPIRIT (see Additional file 1) and CONSORT (see Additional file 2) guidelines for randomized controlled trials (46). The schedule for this study is displayed in Figure 1., while the general procedure is shown in Figure 2.
Trial design
This study is a cross-over (within-subjects) randomized controlled trial (RCT) conducted at the Smart-Aging Research Center (Institute of Development, Aging and Cancer, Tohoku University), Sendai City, Miyagi Prefecture, Japan. All participants will perform both, experimental and control condition (in other words, all participants will act as their own controls). This study was registered with the University Hospital Medical Information Network (UMIN) Clinical Trial Registry (UMIN000019832) on 1st October 2018.
Participants recruitment and eligibility criteria
30 healthy young adults (15 females) will be recruited among students of Tohoku University (Sendai, Japan) via a specific online system. They will be reimbursed 1.000 yen per hour for their participation.
Participants must be females and males native Japanese speakers who are between 20 and 35 years old and self-report to be right-handed. In addition, they have to report no history of neurological, psychiatric or motor disorders and normal colour vision. To avoid eventual problems during the IVR sessions, we will exclude subjects that report to be extremely sensitive to motion sickness (e.g., nausea while driving a car).
Randomization of interventions
The two intervention conditions will be administered to all participants in two separate sessions (one week apart). Random assignment to the first intervention using an online program (http://www.graphpad.com/quickcalcs/index.cfm) will take place (47). Half of the participants (15 subjects) will perform fist the experimental condition, while the other half will perform first the control condition.
Interventions
Participants will visit the laboratory (located in the Smart-Aging Research Center) for two sessions, one week apart, corresponding to the two intervention conditions. In both sessions, the participants have to comfortably sit on a stool with their feet firmly resting on the ground and their arms relaxed on the body sides. They have to wear the Oculus visor (oculus.com), equipped with two lenses that show the virtual world. They will be instructed to stay still with their body but they can move and rotate their head, to explore the virtual body and the environment. The virtual environment, as well as the virtual bodies and the animations, is modelled in 3D Studio Max 2015 and implemented in Unity3D: it represents an open space with a green floor (simulating a meadow) and a naturally-like illuminated sky, with the horizon visible. Participants will see a gender-matched life-sized humanoid virtual body.
The intervention will be performed in two conditions: the experimental intervention consists in the display of the virtual body in first-person perspective (1PP), where the virtual body substitutes and it is spatially coincident with the real one, in other words, to observe the virtual body, the participant has to look towards him/herself; the control intervention consists in the display of the same virtual body in third-person perspective (3PP), where the virtual body is collocated around 1.5 meters to the left of the real participant’s body, otherwise speaking, to observe the virtual body, the participant has to rotate his head and look towards his left side.
In both conditions (1PP and 3PP), the intervention starts with a familiarization phase of 4 minutes (hereinafter baseline), where the gender-matched virtual body is displayed but there is no animation yet (i.e., the virtual body either in 1PP or 3PP is standing still). This phase is necessary to induce the illusory sensation of ownership over the virtual body (in 1PP) or to do not induce it (in 3PP) (see Measurements section for the online questionnaire on SoBO and SoA), for the baseline recording of the heart rate (see Measurements section) and to check for eventual sickness problems due to the virtual display.
After that, the actual intervention consists in a vHIE performed by the virtual body exclusively: while they’re sitting, participants will see the virtual body (either in 1PP or 3PP) alternating 30 seconds of running and 30 seconds of slow walking, for a total of 8 minutes (17,33). After that, we will ask the participant to stay still close his/her eyes, while we still record an extra 30 seconds the heart rate (see Measurements) (see Figure 2.B). While the slow walking animation will be the same for all participants, in order to choose the speed of the running animation, after the familiarization phase, we will show four different options on the virtual body and subjects verbally report which one is subjectively perceived as considerably fast but feasible. This is necessary to maximize the possibility to have an actual physiological activation but, at the same time, to display an animation that is subjectively plausible, in order to do not break the illusion of ownership in 1PP.
Measurements
Heart rate
During the interventions, in order to check if participants actually perceive the virtual body as their own (in 1PP and not in 3PP), we will record the heart rate (HR) with a Polar H10 heart rate monitor, controlled by a specific application via Bluetooth. Participants have to wear an elastic strip around the chest on which the HR monitor is attached, positioning it close to the heart, in order to record the HR for each intervention conditions (1PP and 3PP) during the baseline (4 minutes), the intervention (4 minutes running, 4 minutes slow walking, for a total of 8 minutes) plus extra 30 seconds (where the subject is still with closed eyes) to check for eventual delay in the recording and to wait for the synchronization of the HR.
Online and offline questionnaires on SoBO and SoA
We will administer an online questionnaire for the subjective experience of SoBO and SoA: to control for eventual changes in subjective feelings during the intervention at different timepoints, in the middle of the baseline phase (i.e., after 2 minutes from the beginning if the IVR session), during the intervention at minute 1 (i.e., while the avatar is running), at minute 2.30 (i.e., while the avatar in slowly walking), at minute 5 (i.e., while the avatar is running) and at minute 6.30 (i.e., while the avatar in slowly walking) we will ask participants to rate their level of agreement (on a 1-7 Likert scale where 1 means “totally disagree” and 7 means “totally agree”) respect to four statements, two of them (one is a “real statement” that check for the actual presence of the illusion, while the other is a “control statement”) about sense of body ownership and two of them about sense of agency (see Table 1) (see Figure 2.B).
Questionnaire verbally administered during the vHIE (at 5 timepoints, at 2 minutes of baseline, 1 minute, 2.30, 5 and 6.30 of vHIE) concerning the subjective participant’s feelings during the intervention. The statements s1 and s2 concern the SoBO, while the statements s3 and s4 concern the SoA. The statements s1 and s3 are “real statement” while statements s2
and s4 are “control ones”. Subjects will rate their level of agreement to the following statements on a 1-7 Likert scale (1 means “complete disagreement” and 7 means “complete
agreement”).
Another questionnaire will be administered right after every IVR session (both after 1PP and 3PP intervention conditions) to check in details for feelings of movements, motor control and physical effort (see Table 2) (see Figure 2.A). The statements are selected and adapted from previous studies (33,41).
Table 2. Offline questionnaire on SoBO and SoA.
Questionnaire self-administered after the vHIE interventions concerning the subjective participant’s feelings during the previous intervention. The underlying explored domain are listed in the first column (not shown to the participant) while the corresponding statements are listed in the second column. Subjects will rate their level of agreement to the following statements on a 1-7 Likert scale (1 means “complete disagreement” and 7 means “complete agreement”).
Stroop task
To measure the actual efficacy of the proposed intervention, we will record before and after every IVR session (either 1PP and 3PP) cortical hemodynamic changes in the participant’s PFC using the functional near-infrared spectroscopy (fNIRS) device during the execution of the colour-word matching Stroop task (see Figure 2.A).
We will adopt the colour-word matching Stroop task as previously used in several studies (17,22,48–50). The Stroop task we plan to use consists of 30 trials, including 10 neutrals, 10 congruent and 10 incongruent, presented in random order. For all trials, two words are displayed on the monitor one above the other: specifically for neutral trials, the upper row consists of XXXX printed in red, white, blue, pink or yellow, and the lower row shows the words ‘RED’, ‘WHITE’, ‘BLUE’, ‘PINK’ or ‘YELLOW’ printed in black. For congruent trials, the upper row contains the words ‘RED’, ‘WHITE’, ‘BLUE’, ‘PINK’ or ‘YELLOW’ printed in the congruent colour (e.g., RED was printed in red) and the lower row contains the same colour words printed in black. For incongruent trials, the colour word in the upper row is printed in an incongruent colour (e.g., RED was printed in yellow) to produce cognitive interference between the colour word and the colour name (i.e., Stroop interference). All words were written in Japanese (hiragana). The lower row is presented 100 ms later than the upper row, in order to achieve sequential visual attention. Between each trial a fixation cross is shown, as inter-stimulus interval for 9 to 13 seconds to avoid timing prediction (17,22,51). The stimulus remains on the screen for 2 seconds, independently from the subject’s answer. We will train participants to decide whether the colour of the upper word (or letters) corresponds to the colour name of the lower word by pressing button 1 on the keypad to give “yes” or button 2 “no” responses with their forefingers. 50% of the presented stimuli were correct (the correct answer is “yes”). We recorded response time (RT) and error rate (ER) as variable.
Functional near-infrared spectroscopy, fNIRS
We will use a wearable (without optical fibers) fNIRS optical topography system (WOT-HS, Hitachi Corporation & NeU Corporation, Japan). The NIRS headset (Figure 3.A) sends the signals to the Wearable Optical Topography High Sensitivity software Version 1.04 (Hitachi Solutions Inc.) through a Control Box.
This system is provided with 35 capsules, placed 3 cm of distance to each other, into which microprocessors, near infrared emitting or high-sensitivity receiving sensors are packaged: the top and the bottom lines of capsules alternate an emitting and a receiving sensor, while the central line comprises receivers only (Figure 3.B); this new multi-distance measurement mode, including a total of 12 emitting and 23 receiving (11 of them are short-distance receivers) sensors, significantly reduces the biological noise on the hardware side, resulting in 34 channels over lateral and anterior PFC. The device will be positioned on the forehead by centering the specific mark on bottom line of probes at the Fpz (10% of the distance between Nasion and Inion), according to the international 10–20 system (52).
The device detects the concentration of oxygenated haemoglobin (O2Hb), deoxygenated haemoglobin (HHb) and total haemoglobin, calculated in units of millimolar-millimeter (mM·mm) (53), by applying two short-distance wavelengths of near-infrared light (850 nm, 730nm) to monitor the above mentioned cortical hemodynamic changes in the PFC during the Stroop task (54).
Two-dimensional mood scale, TDMS
The Two-Dimensional Mood Scale (TDMS) is an effective measure to record changes in psychological mood states. TDMS was developed as a psychometric scale using eight mood-expressing items (energetic, lively, lethargic, listless, relaxed, calm, irritated, and nervous) that, combined, express mood states of pleasure and arousal (56). We will ask participants to rate their present psychological state using a six-point Likert scale from 0 = “Not at all” to 5 = “Extremely” (17). They will repeat the TDMS right before the pre Stroop task (before the vHIE), as well as right after the post Stroop task (after the vHIE) for each session (1PP and 3PP) (see Figure 2.A).
Outcomes
Primary outcomes
Considering that the main goal of this study is to determine that vHIE intervention performed with the own virtual body has a beneficial effect on cognitive executive functions, the primary outcome of this study is the cognitive task (i.e., the Stroop task), and specifically the two related measurements (RT primarily and ER secondarily).
Secondary outcomes
Secondary outcomes are the data extracted from the fNIRS recording. In addition, we will examine the relation between Stroop task and the induced cortical activation, by means of a correlation between behavioral data from Stroop task (RT and ER) and optical data from fNIRS.
Additional outcomes
Additional multiple outcomes are the results obtained during the intervention itself (i.e., HR, online and offline questionnaires), in addition to the TDMS scale results.
Sample size
Considering the main outcome of this study is the RT of the Stroop task (which is claimed by many to be one of the indicators of executive performance), that has typically a moderate effect, we calculated a sample size of 28 subjects (with α error probability set at 0.05 and power set at 0.8). We decided to recruit 30 participants to eventually control for drop off.
Data management
To every participant will be assigned a code, according to their arrival at the laboratory to attend the first session, independently from the intervention condition (S01, S02, etc). Personal information and data for this experiment are handled exclusively by the involved researchers (Tohoku University). According to the institutional policy (47), in order to ensure the security of all data and personal information is limited to the involved researchers (and to external parts only after their formal approval). After the end of the experimental period, any information directly linking data back to the participant will be discarded to guarantee anonymity. Personal data eventually collected will not be shared or disclosed in any way.
Statistical analysis
The current RCT is designed to determine if a virtual intervention has a beneficial effect on executive functions; this is possible only if some assumptions, during the intervention itself, are satisfied (e.g., embodiment over the virtual body in 1PP). Consequently, the data analysis will be organized in two phases: in the first phase, we need to check the results collected during the intervention itself (i.e., HR, online and offline questionnaires), in order to see whether the virtual intervention is actually effective. In a second phase, we’ll analyse the measurements in order to directly check our hypothesis (i.e., Stroop task and fNIRS data) and their correlations.
Heart rate data
For the HR, we will average the data of the three recording periods (baseline, running and slow walking) for 4 minutes each: the averaged baseline (HRb) results will be subtracted by the averaged running (HRr) and the averaged walking (HRw) results:
dHRr=(HRr-HRb); dHRw =(HRw-HRb)
Then, an ANOVA 2x2 with factor HR speed at two levels (dHRr that means fast, dHRw that means slow) and factor CONDITION at two levels corresponding to the intervention conditions (1PP, 3PP) will be run. We predict to find a significant increase in the dHRr in 1PP condition, respect to all the other measurements, in order to confirm the physiological counterpart of the embodiment of the virtual body in 1PP only.
Data of online and offline questionnaires on SoBO and SoA
As a subjective counterpart, we will analyse the data of the questionnaires. The online questionnaire will be repeated in five different timepoints (baseline, 1min, 2.30min, 5min, 6.30min): to exclude temporal effect, we will first compare with an ANOVA 5x4 the factors TIME (corresponding to the previously mentioned timepoints where the online questionnaire is administered) and QUESTION (corresponding to the 4 statements of the online questionnaire). We can eventually proceed to an average of the statements across timepoints, and then we will compare in an ANOVA 4x2 the four statements (specifically comparing results from “real statements” with “control ones”) and the CONDITION (1PP, 3PP). We predict to find higher levels of ownership and agency for the real statements in 1PP, respect to control statements and respect to 3PP.
Concerning the offline questionnaire (administered at the end of every IVR condition), we will analyse the data comparing statements among the two conditions (1PP and 3PP).
Stroop task’s RT and ER
The second phase of the data analysis concerns the data collected before and after every IVR intervention, that means the results necessary to confirm our main hypothesis (i.e., the efficacy of IVR training on cognition).
As previously mentioned, in the Stroop task we will include two measurements, RT and ER: the (incongruent – neutral) contrast, which is assumed to represent Stroop interference, will be calculated. Both (RT and ER) will be analysed by means of a repeated-measures ANOVA with TIME (before, after) and CONDITION (1PP, 3PP) as within-subject factors. As supplementary analysis, we’ll check for eventual differences between the two sessions considering only the Stroop task before the vHIE (predicting not differences). According to our hypothesis, we predict to find a shorter RT and lower ER after the vHIE in 1PP, compared to 3PP.
fNIRS data
The optical data from fNIRS will be analysed based on the modified Beer-Lambert law (57). We will set the sampling rate at 10 Hz and analysed the difference between O2Hb and HHb signals.
We will employ the general linear model to identify O2Hb and HHb hemodynamic brain responses with reference to experimental factors. If necessary, we will combine few adjacent channels in order to create ROI (58) according to the LBPA40 anatomical labelling system (59).
Then, the changes in the concentration of O2Hb and HHb for each channel will be treated according to the following steps:
1) Excluding skin blood flow (i.e., heartbeat pulsations) from raw data by using the specific software provided;
2) Pre-processing each channel using 0.01 – 0.5 Hz bound pass filter to account for the effects of Mayer waves, high-frequency fluctuations and baseline drift (60).
2) Performing a 3-second moving average to smooth the raw O2Hb concentrations (61,62).
3) We will use the colour-word matching Stroop task in an event-related design which presents each neutral, congruent and incongruent conditions in random order, so we will pick up the changes related to each condition in the concentration of O2Hb from each channel (or the combined ROIs). Especially, we will calculate the mean of the changes in the concentration of O2Hb 2 seconds before onset task, as a “baseline”, and 10 seconds after the onset of the inter-stimulus interval as “vascular response” in each Stroop task condition (48). That will be necessary because NIRS signals are delayed respect to participants response (48,63).
4) Translating the mean of O2Hb concentrations of each channel to normalized values using linear transformations, so that the mean±standard deviation of O2Hb level in the 2 seconds of the baseline period is 0±1(AU). This method will be useful because of the avoidance of the influence of differential pathlength factors among the subjects and that of cortical regions (60,64,65).
4) Lastly, the channels over target ROI areas, eventually, will be averaged respectively in each Stroop task condition.
TDMS data
We will calculate levels of arousal and pleasure from TDMS scores. As first, we will confirm that the sample represents the normal population, i.e., shows a normal distribution (all p-values for arousal and pleasure level in both intervention conditions have to be over 0.05 at Shapiro-Wilk test). As well, we’ll check if levels of arousal and pleasure are not different between sessions (comparing results before 1PP and before 3PP) so that all participants start from the same baseline level. That means we can proceed to subtract the results after the intervention from the results before the intervention (after-before).
By performing an ANOVA comparing 1PP and 3PP, we predict to find an increased level of arousal (but not necessarily of pleasure) after the vHIE, especially after the intervention in 1PP, as described in previous studies (17).
Correlations
The crucial correlation for our hypothesis concerns the relationship between Stroop task and fNIRS data: we’ll first investigate the cortical regions mainly activated during the Stroop task before the vHIE in both intervention conditions, as a baseline reference. Then, the (incongruent – neutral) contrasts for the two intervention conditions will be averaged as substrates for ROI analysis. The (incongruent – neutral) contrast for RT and ER with O2Hb changes in all ROIs will be treated with a repeated-measures ANOVA considering as factors intervention condition (1PP and 3PP) and time (before and after the vHIE).
In addition to that, as described in previous studies, we will examine the relation between Stroop data and cortical activation or TDMS results in a binominal manner (22,51): for each variable the following contrast will be calculated, {[(incongruent – neutral) of after vHIE] – [(incongruent – neutral) of before vHIE] in 1PP condition} – {[(incongruent – neutral) of after vHIE] – [(incongruent – neutral) of before vHIE] in 3PP condition}. Both values will be subjected to the McNemar test to examine the correspondence between the two incidences (66).
Data monitoring and auditing
According to the recommended guidelines for clinical research in Japan (https://www.mhlw.go.jp/topics/bukyoku/seisaku/kojin/dl/161228rinsyou.pdf) and the institution regulations, data monitoring by a third party is not applicable for the RCT here proposed, since we are not providing participants with any medications nor surgery (47).
Risks and benefits to participants
Participants are unlikely to encounter any serious risks or burdens.
Participants will possibly experience fatigue and discomfort during the Stroop task and the fNIRS recording. The participant will be informed in advance that, should they feel any discomfort, the test can be interrupted at any time.
The interventions in IVR are not particularly difficult and should not cause the participants any pain since they just have to stay still. Possibly, participants may experience a sense of dizziness or nausea due to the IVR display; to avoid that, exclusion criteria mention that people who are very sensitive to motion sickness are excluded. Anyway, if for some reason the participant experiences any discomfort, they will have been previously informed that they may immediately interrupt the training. Any adverse event will be formally reported.
In accordance with the regulations of the institution, the participants will be given a monetary reward, based on the number of hours invested performing the experiment. Therefore, in case of the participant decides to leave the ongoing experiment, he/she won’t receive any reward (i.e., only participants who complete the entire experiment will be rewarded).