Figure 1. Trial overview
The project will be a parallel group trial using a pre-test/post-test control group design. There will be two data points (baseline and follow-up) with comparisons between two experimental groups (gastroception vs. cardioception; focussing on the lower abdomen or chest area, respectively) and a control group.
Since emotion research deals with constructs that are multiply realised (24), the project will employ multiple measures. There will be two primary, single-blinded studies, both lab-based, one utilising a spatial cueing task and another utilising a stop/signal task. In addition, physiological and self-report measurements will be taken (these are described in detail below).
The sample for the main data collection will be drawn from across the health professions and student healthcare cohorts in the Wellington area, New Zealand. Given the preliminary nature of the trial, only healthy individuals will be recruited at this stage.
Medical students have been reported as experiencing difficulties in emotion regulation when dealing with the many stressful situations faced during training (25). Further, a high prevalence of suicide ideation has been found in medical student populations (e.g. 11% incidence; 26; 15% incidence; 27). Qualified doctors also experience emotion regulation difficulties (28, 29), with a suicide rate that is 1.4 times higher among male doctors than in the general male population, and 2.3 times higher among female doctors than in the general female population (30).
Nurses may also be at greater risk from suicidality, with recent UK statistics (31) indicating this group to be the most at-risk of all health professions (32). Nursing students may, too, face similar challenges to their medical counterparts.
Finally, allied health professionals - particularly female, according to an Australian study - have also been identified as at risk (33) and hence will be included in the sample, along with allied health students given that these, too, may be likely to have similar experiences to their medical and nursing counterparts.
A power analysis calculation was based on the RT measures in Silverstein et al. (34), who reported a mean RT of 2134.1ms (SD = 856.6ms) pre-treatment and 1403.6ms (SD = 462.3ms) post-treatment for meditators, and 1865.1ms (SD = 800.4ms) pre-treatment and 1893ms (SD = 780.6ms) post-treatment for controls. This gave a treatment-by-time partial eta-squared of 0.15, with a p-value of 0.03. Using the same partial eta-squared for the three group, three time point studies, with alpha = 0.05 and 80% power, a minimum of 54 participants would be required in total (across all group conditions). This equates to a minimum of 18 participants per group to detect the main effect in the spatial cueing task. (regarding the stop/signal task, Verbruggen & De Houwer (35) detected effects in a single group with a sample size of 23 participants.) We will aim for 70 participants in total, assuming a drop-out buffer of approximately 5 – 10 participants. This will allow 20+ participants in each group.
Recruitment of participants
Potential participants will be recruited by several means. An advertisement (hard copy and electronic form) will be posted around the University of Otago Wellington (UOW), Massey University, Victoria University of Wellington, Wellington Hospital and the wider District Health Board. In addition, the New Zealand College of Clinical Psychologists and the New Zealand Psychological Society will be approached to attract the interest of clinical psychologists, either in-training or in-post.
The participant information sheet and consent form will be sent to those responding to the advertisement for further consideration. Prior to admission to the research studies, potential participants will also be screened to ensure they meet the inclusion/exclusion criteria.
Screening will be undertaken online (on the survey platform Qualtrics) by a brief bespoke questionnaire collecting information on all but two of the inclusion/exclusion criteria (see below) and using the following tools for the remaining criteria:
- The Somatosensory Amplification Scale (SSAS) (a brief, 10-item questionnaire: approximately two minutes to complete; 36);
- The Edinburgh Handedness Questionnaire (37) (a very brief version of this self-report tool will be used: approximately one minute to complete; 38).
Recruitment of potential participants began in November 2018 and will continue until April/May 2019. Random group allocation will not begin until the middle of May 2019.
There are the a number of requirements for participation. Participants must be:
- Currently enrolled on a full time medical, nursing or allied healthcare degree/diploma OR already qualified and working as a doctor/nurse/allied healthcare professional;
- Aged 18-35 years. This range is based on indications of declining interoception after approximately 40 years of age (39), and initial declines in reaction times have been identified as early as 24 years of age (40), hence age is capped at 35;
- BMI must be equal to or lower than 30 (i.e. non-obese); 32 for Maori/Pacific Islanders. Increased body fat percentage tends to make interoception more difficult (41);
- Regular levels of physical activity: at least 150 minutes of moderate-intensity aerobic physical activity throughout the week or least 75 minutes of vigorous-intensity aerobic physical activity or an equivalent combination of moderate- and vigorous-intensity activity (based on World Health Organisation guidelines). Regular exercise influences autonomic tone (42, 43), which is able to improve interoceptive accuracy as assessed by heartbeat perception (44-46);
- Normal or corrected vision;
- Normal or corrected hearing;
- Right-handed (left vs. right-handed people respond differently on reaction time tasks; see for example, 47; plus this criterion removes the need for bilateral counterbalancing);
- Availability for the entire study period, including attendance at two laboratory sessions (one in May/June 2019, one in July/August 2019, with an 8-week interval).
- Poor physical health (including gastrointestinal disorders, heart disease, kidney disease, respiratory disease);
- Poor psychological, psychiatric or neurological health (e.g. alexithymia, schizophrenia, depression, anxiety, brain injury);
- Somatosensory amplification (score on the SSAS >35);
- In pregnancy;
- Formal dance or drama training (see 48);
- Formal musical training (see 49);
- Formal meditation or mindfulness training (>4 weeks consecutively over the past year) (see 50).
Besides the case of pregnancy, which will likely affect physiological measurement, gender will not be part of the screening, despite men tending to outperform women in interoceptive sensitivity tasks (e.g. cardiac activity detection), due to the likely explanation for this difference being varying BMI across the sexes (see 51).
Each participant will be assigned a number ID in order of recruitment into the study. Once the required sample size has been reached and scheduling completed for laboratory attendance, these numbers will be entered into an online random number generator (http://www.randomization.com/) to allocate these numbers into one of the three blocks (cardioception group; gastroception group; control group). The allocation ratio for the groups will be 1:1:1.
Other randomisation pertains to counterbalancing over trials and blocks of trials. This will be undertaken to control for ordering effects, and will be achieved using the standard functions of the stimulus presentation software program SuperLab (Version 6, Cedrus).
Participants will be told of their allocation immediately after the completion of baseline data collection. Concealment will be ensured using sealed notes provided during this first laboratory visit, within the process described below (see section: ‘Confidentiality and single blinding’).
The likelihood of drop-out from the project will be reduced by offering compensation for time (supermarket vouchers), the attraction of meditation training, maintaining regular contact, either by face-to-face (laboratory visits) or online monitoring (on Qualtrics), offering flexibility to participants on laboratory scheduling and limiting the sample to those who are likely to be already on-site (either the University campus or Hospital, which are interconnected buildings).
Given the need to undertake laboratory tasks and the use of physiological measures, coupled with the local sampling possibility, all in-person data collection will take place in the Centre for Translational Physiology, University of Otago Wellington (UOW).
For those allocated to the experimental groups, an intervention involving one of two interoceptive practices will begin immediately after completing baseline measurements in the laboratory. Data will be collected at baseline and 8 weeks later (a time-scale based on a study of body awareness and interoceptive accuracy, 52).
Over the 8-week period, participants from each of the experimental groups will be instructed to log on to Qualtrics six days per week to record their responses to several qualitative questions on their everyday emotional experiences, and then to attend to their assigned body area for 20 minutes (2 x 10 minutes) after responding to these same questions (see Appendix 1). The 20-minute period will be timed automatically with full task instructions provided. Participants will be requested to do this task at a set time each evening. All log-ins and time spent on the site will also be recorded and time stamped automatically, which will help with adherence, create a sense of connection for participants throughout the study period (in the absence of a bespoke class or programme to attend in person, participants will at least know their online attendance is monitored), and ensure that the task proceeds similarly for all participants. Adherence will also be encouraged by contacting all participants at the mid-point (i.e. four weeks after baseline) to check for problems or difficulties in adhering to the intervention instructions.
The instructions for cardioception will be to focus on the inside of one’s chest area without using non-interoceptive cues, throughout the daily 20-minute period. These participants will be provided with an image of a human body with a highlighted chest area, which displays the target area for interoceptive focus. The instructions for gastroception will be to focus on sensations in the lower abdomen (i.e. the dominant locus of gastrointestinal structures and activity). These participants will again be provided with an image of a human body, but with a highlighted lower abdominal area, displaying the target area for interoceptive focus. This does not, then, incorporate the mouth and - in particular - the oesophagus, which overlaps with the location of the heart and would, therefore, be confounding as a focal point. Given the dominance of the lower abdomen for gut volume/activity, there is arguably relatively little detectable activity throughout much of the mouth/oesophagus unless a participant is experiencing gastric reflux, vomiting, or other pain, or is eating/has just eaten, which the sample screening aims to preclude in any case.
All participants from the experimental groups will be required to not undertake any additional meditation outside of the intervention, during the 8 weeks.
Rationale for the control group
The control group will differ from the experimental groups only by not undertaking meditation/interoception training during the 8 weeks. They will otherwise be “active”; i.e. they will be asked to log on to Qualtrics for the same period to complete the self-reflection questionnaire. This will control for any effect of self-reflection or regular engagement in a task, measure any effect of interoceptive training in general, as well as any effect of each specific form of interoceptive training. The interoceptive groups together will also act as their own control for any general effect of interoception. If they choose to, controls will be provided with the online meditation instructions and task once their participation is complete.
Data collection procedures
Prior to lab attendance at baseline and follow-up, participants will provide online (via Qualtrics) self-report measures on emotion regulation and health. The self-report tools to be used are The Difficulties in Emotion Regulation Scale, The Emotion Reactivity Scale, and SF-36 sub-scales (outlined below in ‘Descriptions of self-report tools for outcome measures (emotion regulation and interoceptive sensibility)’). On an on-screen image of a human body, they will also click to indicate the location(s) where they most tend to detect emotional feeling. If they do not detect emotional feeling in any part of the body, then no response is required. This is to control for any pre-existing bias in body focus.
Other pre-task procedures will include providing instructions to participants stating that during the four hours prior to their arrival they should not consume a heavy meal (but are adequately sated by the time of arrival in order to avoid distracting or facilitating hunger pains) or items containing caffeine or nicotine, or a large quantity of water (i.e. just maintain usual drinking amounts to prevent differences in body awareness during the laboratory visit, such as awareness of the stomach or bladder); nor should they have consumed alcohol in the previous 24 hours. In advance of attending, each participant will be asked (using the SF-36) if they have been experiencing any pain over the previous four weeks (which may interfere with their performance or that may prove distracting).
During laboratory visits for each data point
All participants will have the chance to use the toilet upon arrival at the laboratory. All participants will also be offered a half-cup of water in order to attempt to equalise current levels of water intake in combination with the pre-attendance requirements (i.e. those who are less sated will be more likely to take up the offer). The pre-task procedures will be checked, participants will be rescheduled if necessary (such as if caffeine had been consumed). Mobile phones will be switched off prior to the start of experimentation.
Participants will undertake the spatial cueing task followed by the stop/signal task, then by a measure of interoception involving heart beat estimation, all on a laptop computer. This will be followed by further tests of interoception, including a water load test and a self-report tool. All tasks will take place in a physiology laboratory (with light attenuation on the first two tasks).
Physiological measurements will be taken throughout all computer tasks. Heart activity, measured by ECG, is commonly found in emotion studies as an objective indicator of affective changes. Heart Rate Variability (HRV) is measured by beat-to-beat changes in heart rate (53), where greater HRV is an indication of adaptability and has been associated with inhibitory processing (e.g. 54-56). Rapid heart rate deceleration has been established to occur following the presentation of external stimuli (57, 58), but which is more marked when viewing emotionally negative stimuli (see 59, 60). By taking continuous ECG measurements, this deceleration can be recorded and linked to the presentation of individual stimuli. In order to link the ECG with the stimulus presentation software, an audible tone will be the first event in each task, played through headphones with a microphone inserted into a speaker. The microphone will be plugged into the acquisition and analysis system Powerlab 16SP (ADinstruments). Superlab will be set up to track the cumulative time (in milliseconds) throughout trials, so that the initial tone will then be the reference point for all other measurements in both Superlab and Powerlab (using Labchart software, ADinstruments). Because the sound is played through headphones before each participant begins this will have no relevance to any other tone played during the stop/signal or the Schandry Task (the spatial cueing task involves no tones in any case). The magnitude of the heart rate deceleration (and “heart rate variability” more generally) can then be used as an objective physiological indicator of emotional responses.
In addition, constant measurement of heart rate will permit control of the influence of the systole heartbeat phase on responses to threat/fear stimuli (i.e. compared with diastole, responses to such stimuli during systole lead to greater arousal and are likely to be quicker; 61). This has the potential to create unmeasured variability in the data if stimuli are presented at varying heart phases.
Participants may attend with loose-fitting clothing or a scrubs top will be supplied; this is to allow easy application of the ECG electrodes (which will be self-applied according to given instructions). A 3-lead ECG will be recorded with surface Ag/AgCl electrodes on the chest in standard configuration (ADinstruments ECG, NZ). PowerLab will be set at a sampling rate of 1k/s, and with a range of 100mv.
Participants will move to a seated position in an enclosed booth to prevent visual distractions. They next rest their head in a chin rest to prevent overt orienting, facing a computer screen (AppleMacbook Pro 15-inch laptop screen) at a distance of 45cm from eye position and a mouse to the right-hand side. To ensure the comfort of each participant, an adjustable height desk will be used.
In order to be able to answer any questions and, insofar as possible, to monitor adherence to the task instructions, the researcher will be seated at the far side of the laboratory room throughout the tasks, unable to see the participant.
Spatial Cueing task
This will address RQ1 and RQ2. To address RQ1, an emotional cross-modal spatial cueing task will be used to record RTs to emotional stimuli (the International Affective Picture System, or IAPS; the Nencki Affective Picture System, or NAPS) when detecting activity at a location. The IAPS (62) is a well-established set of emotional stimuli, reliably inducing affect with short presentation duration (63), and without habituation during repeated exposure to same-valence images (64, 65). The NAPS is a more recently developed, less widely used, but still validated visual stimulus set (66) using the Self-Assessment Manikin (SAM), which was used for standardising ratings of arousal and valence in the IAPS.
The IAPS stimuli have been shown to elicit physiological responses, with one study (67) finding heart rate variability responses with images rated at arousal levels of =>4.5 (based on the scale for SAM, 1-5, where 5 is the highest level of arousal). If erotic imagery is excluded, negative emotional images that produce defensive responses, such as pictures of human or animal attack, tend to be the most highly arousing (68), eliciting startle blink responses, EMG responses, heart rate deceleration and skin conductance in both men and women (69). Based on findings such as these, and with the project’s overall focus on suicide prevention, it makes sense to focus on negative arousal.
A modified version of Thomas et al.’s (70) spatial cuing task will be used. The original study measured “interpersonal body representation” via responses to exogenous visual cues that shared a known location with tactile stimulation. Specifically, this used brief flashes of light on different locations of a model. Using a solenoid, a tactile stimulus was then applied to the same or different body part of the participant. Participants responded verbally as quickly (RT) and accurately (error rate) as possible. The findings from Thomas et al. indicated that valid cues resulted in faster RTs and fewer errors in contrast with invalid cues, suggesting that a cue in one mode can be effective for target identification in another mode. However, this was found only in the anatomical congruence condition (i.e. my right side is your right side), not the specular (i.e. my right side is your left side), which suggests a cross-modal relationship mediated by empathy (consistent with findings on “mirror neurons”; e.g. 71).
In the modified version of the task to be used here, the RT to cues will be treated as indicators of a previously unknown location of emotional feeling to which participants have responded. Participants will be requested to indicate emotional signal detection upon presentation of an emotional stimulus, which follows a visually cued location on a computer screen (a seated figure with a highlighted area of the body). Faster RTs will be more likely if there is in fact a detected bodily location that is accurately cued. Alternatively, slower RTs will be more likely if there is a detected bodily location that could be (cross-modally) distracted from by the presentation of a divergent visual cue.
Consideration needs to be given to possible ‘noise’ generated by these methods. Firstly, participants have been shown to have faster responses to a visual cue presented on the same side as the responding hand, which has been interpreted as a cross-modal cueing effect (see 72). However, this is not always considered in relation to the well-established “Simon Effect” (73), which is where RTs are faster due to a primed motor response to the target rather than an attentional cross-modal effect. For this reason, the “Interpersonal body representation” paradigm appears to be an advance on other spatial cueing approaches due to the anatomical versus specular differences found: the “Simon Effect” appeared to have been overridden by the social nature of the cue. Plus, the cues used in the current study will always be presented centrally, with no bilateral bias.
Other possible ‘noise’ includes:
The effect of overt bilateral orienting. This is essentially a cueing effect since it cuts down possible attentional foci by 50% at the point of stimulus onset. The stimulus, or key details within a stimulus, may obviously be on one side of a screen or the other. For this reason, central fixation in visual cueing tasks is necessary to prevent variation across a sample in this respect.
Lack of luminance. In all visual cues and stimuli, isoluminant colour changes (74), isoluminant onset cues (75) or sudden onset stimuli that do not involve a luminance change do not capture attention exogenously (76). For this reason, standardised presentations of visual stimuli with the validated ability to capture exogenous attention should be used.
The effect of sudden onset only. There can be ambiguity between specifically affective salience and the more general effect of the sudden onset of any physical stimulus (77). For this reason, blocks of trials must include neutral stimuli as well as emotional stimuli.
Creating expectations. In exogenous attention tasks, the cues may become task-relevant (i.e. predictive of a target stimulus and hence endogenous) (78). In exogenous tasks, attention is drawn to something, whereas in endogenous trials attention is directed at something with an expectation of an event (e.g. the difference between being woken in the night by a fire alarm and waiting for the school dinner bell to sound). To address this, the frequency of valid cues should, ideally, not depart from 50% of total trials (i.e. equal to chance), with suggestions in the literature (see 79) that >70% of overall trials may be predictive of a target (i.e. endogenous). In the emotional spatial cueing task, cues often predict the target location in 70% or more of the trials (see, for example, 80), which may facilitate task relevance.
Based on the foregoing, the spatial cueing task procedure will continue as follows:
Figure 2. Outline of the spatial cueing task.
- A practice block of 16 trials will be completed (using 8 neutral emotional stimuli, involving each of the cued locations, and neutral cues).
- In each trial, participants fixate on a word (“Ready”) at the centre of the screen for 500ms prior to cue onset.
- The cue (a seated male figure) will be presented for 2000 milliseconds in the middle of the screen, either with one of three locations highlighted at random using a red, highly luminous highlighted rectangular area (see 81, 82), or no location highlighted. The three locations will be the chest, lower abdomen or shins (pressed together in the image to form a single area). All highlighted areas will be the same size, colour and luminescence. The image will be large (345 x 715 pixels), and the highlighted area as luminescent as possible within the limitations of the package used to create it (Microsoft Office Word, version 16.23).
- The emotional stimuli (selected IAPS and NAPS images; see Appendix 2) will be either 600 x 800 pixels or 800 x 600 (i.e. landscape or portrait), all at a resolution of 72 pixels/inch. The stimuli will be delivered at an inter-stimulus interval (ISI) of 200ms, remaining for 200ms (changes in levels of arousal do not increase significantly beyond 150ms presentation duration; 83). With picture stimuli being presented to elicit exogenous attention, a mixture of neutral and negative emotional images will be presented to control for the effect of sudden onset of a stimulus. Further, since negative stimuli tend to reliably create the highest level of physiological arousal in men and women, this will be important for creating detectable changes in the body. Again, to ensure that the cues do not become task-relevant (i.e. predictive of a target stimulus and hence endogenous), the frequency of valid cues will be kept to below 70% of trials. Although it is not known in advance what a ‘valid’ cue is in this bespoke variant of the spatial cueing task, for the purposes of setting up the cues these are limited to the chest and lower abdomen areas (interoceptive/visceroceptive areas), each with 31% of ‘valid’ cues, which remains within the standard limits (i.e. 62% in total). A further 10% of trials will be ‘invalid’; i.e. the shins (a non-interoceptive/visceroceptive area).
- In 15% of trials, there will be no emotional stimulus following a cue (catch trials); plus, in a further 14% of trials there will be a neutral cue (no cue areas highlighted).
- Non-catch ‘valid’, ‘invalid’ and neutral trials will be further split into calculated proportions of negative and emotionally neutral target stimuli. In total, there will be 65% negative and 20% neutral stimuli (along with the 15% catch trials).
- Participants are required to give a speeded response: a mouse click upon detection of a bodily change - anywhere in the body - but they will be instructed to be sure of detection (i.e. guesses are discouraged; see 84). It is important to separate this initial response from the indication of a location in order to clearly disambiguate (i) no emotional feeling being detected, and (ii) an emotional feeling being detected but when the location was not known. If no response is registered within 6000ms, the next trial is presented.
- If a change was detected then participants indicate a location on an on-screen virtual body by mouse click, followed by an indication of the valence and intensity of the detected feeling (using the Self-Assessment Manikin; for a similar approach, see 85). If a location was detected but it was not known where, this can be indicated by clicking outside of the body area. The time taken to give these further responses will assist with a likely longer refractory period for an embodied emotional response compared with, say, visual attendance. (See, for instance, 86, where the mean cardiovascular recovery time following an emotional film stimulus - lasting up to 100 seconds in duration - was approximately 30 seconds. The estimated time per trial in the current spatial cueing task, including an inter-trial interval of 200ms, is around 15 seconds, but where a picture stimulus will be presented and for a far shorter duration of 200ms, which may then result in a shorter recovery period than a relatively lengthy film.) This process will be repeated throughout all blocks, where the cued response is exogenous.
- In total, there will be 6 main blocks, with a brief rest period between them, and 248 trials. Based on the percentages stated previously, this will result in 152 valid trials (76 gut; 76 heart), 26 invalid trials (shins), 37 catch trials, and 34 neutral cue trials (where no part of the body is highlighted). For each of the valid sets of trials, 58 emotional stimuli will be negative and 18 will be neutral. For invalid trials, 20 will be negative and 6 will be neutral. For neutral cue trials, 26 images will be negative and 8 will be neutral. For catch trials (no emotional stimulus), there will be 10 sets of each of the 3 bodily locations, with one set of 7 neutral locations (no areas cued on the image of a seated person).
- Finally, to control for order effects, blocks will be counterbalanced: one half of each of the groups (assigned at random) will complete blocks in the order 1-6. The other half will complete the blocks in reverse order.
The subsequent task addresses RQ3, and will use a second experimental paradigm involving the same experimental apparatus: a stop/signal task (a variant of the better-known go/no-go task). Whereas the latter requires participants to respond rapidly to a target stimulus (‘go’) and withhold a response to a non-target stimulus (‘no-go’), the stop-signal task usually requires infrequent withholding of a response when presented with a randomly presented ‘stop’ signal following a choice task between frequent presentations of two ‘go’ stimuli (87, 88). Also, whilst ‘go/no-go’ tasks have zero stimulus onset asynchrony (SOA), stop/signal tasks have variable SOA. Where the ‘stop signal delay’ (the interval between stimulus presentation and stop signal presentation; i.e. the SOA) is short, participants can easily inhibit a response. With longer delays, participants are more likely to execute a response and, therefore, fail to inhibit. The stop-signal task, then, represents a more difficult test of response inhibition than go/no-go tasks due to the increased likelihood of a pre-planned response being executed when there is more time for execution, in combination with the higher frequency of ‘go’ versus ‘stop’ signals.
The “Horse Race Model” (89-91) is a theoretical explanation for performance on the stop-signal task, where go and stop processes are deemed independent and “race” each other to their respective thresholds and, thereby, one of two outcomes: a response or response inhibition. By varying the SOA (between the ‘go’ choice stimuli and ‘stop’ signal: termed “staircasing”), a time estimate for response inhibition can be calculated from the mean SOA required for a 50% success rate of response inhibition and the mean RT to Go stimuli. This is the ‘stop-signal reaction time’ (SSRT). The current study will be based on a key study of emotional inhibition undertaken by Verbruggen & De Houwer (92).
These authors found that when the target stimulus in a stop/signal task is emotional, reduced inhibition would result in slower RTs due to increased emotional activation (i.e. greater task interference from emotional stimuli). The current study will vary from their approach somewhat in that the Verbruggen & De Houwer also investigated differences in impact from varying valence and intensity of imagery (finding only an effect from the latter), whereas the current study seeks to compare levels of inhibition to arousing stimuli of either valence, and using neutral stimuli for similar reasons to the spatial cueing task (as a standard control for the sudden onset of a visual presentation). Using an array of valence scores aims to support a more general emotional effect (which the study on spatial cueing is not able to do given the need to maximise high arousal stimuli).
The full procedure will run as follows:
Figure 3. Outline of Stop/Signal task
- Instructions are provided on-screen for the practice block. This training block consists of 24 trials using 8 neutral IAPS/NAPS images, otherwise following the same process as for the experimental phase. For the experimental phase (8 blocks of 60 trials), 120 images will be used (40 positive, 40 negative, 40 neutral), with each image appearing four times. The set of 120 images (see Appendix 2) will be split into two halves, with half presented in even numbered trials, and the other half in odd numbered trials.
- For all blocks, participants first fixate on a cross at the centre of the screen for 500ms; IAPS/NAPS stimuli are then presented for another 500ms.
- On ‘go’ trials, participants must respond to the visual stimuli ‘‘#’’ or ‘‘@’’ presented at the centre of the screen, pressing a corresponding key as quickly and accurately as possible after presentation. These symbols will be reproduced as stickers on the keyboard to ensure participants make the required key press. Responses must be completed within 1250ms before the next trial begins. After incorrect trials, this ISI begins following a 250ms delay. The word “Incorrect” appears when there is an error, “Respond” if there is no response when there should be, and “Stop” where there is a response but should not be.
- On ‘stop’ trials (indicated by a tone on headphones; 750 Hz, 50 dB, 100 ms; presented randomly in 30% of trials = 48 Stop trials per valence set), participants must withhold a response to the visual stimulus. The stimulus onset asynchrony (SOA; the stop-signal delay) between the choice task stimulus and the auditory tone will begin at 250ms, then vary by staircasing: a response on a ‘stop’ trial results in a decrease in the SOA by 50ms; successfully withholding a response on a ‘stop’ trial results in an increase of 50ms.
- Participants will be informed of this auto-adjustment in staircasing, to discourage their anticipation of the ‘stop’ signal and to focus only on rapidly responding to the symbols (as emphasised in Verbruggen & De Houwer).
- If in the unlikely event a participant presses before the tone, a neutral message of “Next trial” appears on-screen in order to ensure that random, rapid key pressing is not encouraged whilst at the same time not penalising genuine attempts to respond rapidly. (This is a bespoke part of the design, not taken from Verbruggen & De Houwer.)
- The time between trials will be 1500ms.
- At the end of each block, performance feedback is provided to participants: the number of correct ‘go’ responses (i.e. pressing the correct symbol); number of incorrect ‘go’ responses (i.e. pressing the wrong symbol); number of incorrect ‘go’ non-responses (i.e. too slow); number of correct ‘stop’ non-responses (i.e. withholding a response when hearing a tone); number of incorrect ‘stop’ responses (i.e. responding in the presence of a tone). Trials where “Next trial” appears on-screen are not included in this tally.
- There will be a 30-second gap between blocks.
Counterbalancing: for half of the participants in each group (assigned at random), the blocks will be presented in the order 1-8; the order will be 8-1 in the other half.
Towards the end of each laboratory visit, measures of interoception will be taken. Interoceptive sensitivity will be assessed using two different methods.
Schandry heart tracking task (93)
This will be undertaken by all participants, again using Superlab and Labchart with Powerlab. This entails a participant counting their heart beats within set time periods (a training interval of 25 seconds, followed by four experimental intervals of 25, 35, 45, and 55 seconds presented in random order) without taking their own pulse or using any other non-visceroceptive cues (e.g. hearing their heart beat, watching a part of their body pulse in tandem with heart beat) at the same time as the actual heart rate is measured using ECG. The onset of each time period is signalled by a single tone played through headphones, with the offset of each period signalled by a double tone (200ms gap between two single tones). Participants will then type in their estimated number of heart beats. Each counting period will be separated by a 30 second interval. For ECG recordings, the single and double tones will be used as reference points to permit the calculation of difference between actual heart rate and perceived heart rate, as a measure of interoceptive sensitivity (accuracy).
Water Load test
After disconnecting from the ECG and computer, the water load test will be undertaken by all participants (94). This has two stages: the first stage entails participants’ self-assessment of having reached a state of satiation following drinking water from a 5L glass container through a 4.8mm clear PVC 1 metre tube. The glass containers will be covered to prevent participants from knowing the volume of water in each, to give a sense of a practically unlimited supply of water. Unknown to participants, each flask contains no more than 1.5L of water. This limitation on the amount of water prevents the risk of water intoxication, in combination with the screening measures in place regarding kidney health (see below). Further, in addition to the previous use of this task in published, peer-reviewed research, this component was deemed safe following consultation with an endocrinologist and during the University ethics process. Further, this was confirmed by a second physician, who will provide oversight of the project in general).
The second stage entails drinking water from an identical set of apparatus until reaching a sense of fullness. In the original paradigm, each participant was not informed about this subsequent stage until after the first stage was complete. However, given that this measure will be repeated over two data points, participants will be informed about the second stage prior to the onset of the task to ensure consistency across data points. Each stage is limited to 5 minutes. The amount of water consumed at each stage - measured by weight differences of the containers between pre- and post-consumption of water - is then used to calculate the participant’s interoceptive sensitivity to gastric sensations. Due to the fact that having a full stomach of water would likely confound measurement during tasks (not least due to relative discomfort), this measure of interoception will be taken towards the end of each laboratory visit. Also, running the water load at the end (approximately 2 hours in to their visit) also helps to ensure greater parity between participants in terms of prior water intake.
Finally, each participant will complete a brief self-report questionnaire to measure interoceptive sensibility (The Multidimensional Assessment of Interoceptive Awareness or MAIA; 95). This is outlined in the next section.
In undertaking measures of sensitivity and sensibility, this also allows the calculation of metacognitive awareness; in this way, all three forms of interoception can be measured (96).
After the 8-week intervention period, participants attend the follow-up, involving repetition of all procedures outlined above. Participants will be reminded at the start of the second visit that all procedures will be identical and to perform them as at baseline.
Descriptions of self-report tools for outcome measures (emotion regulation and interoceptive sensibility)
The Difficulties in Emotion Regulation Scale (DERS) consists of 36 items (reduced to 18 in a validated short version; 97) in six subscales (98): (i) non-acceptance of emotional responses; (ii) difficulties engaging in goal-directed behaviour; (iii) impulse control difficulties; (iv) lack of emotional awareness; (v) limited access to emotion regulation strategies once an emotion is in process; (vi) lack of emotional clarity. (See, for example, 99 for information on the scale’s psychometric properties.) A key rationale for using a tool such as this is to assess how the measures taken in the laboratory (i.e. at an early stage during an emotional reaction) may relate to later stage emotion regulation. For instance, decreased RTs in the spatial cueing task may indicate increased access to emotional sensations and which may relate to emotion regulation.
The Emotion Reactivity Scale (ERS) consists of 21 items in three subscales (i) the sensitivity to emotion (e.g., “Other people tell me I’m overreacting”); (ii) intensity (e.g., “When I experience emotions, I feel them very strongly/intensely”); and (iii) persistence (e.g., “When something happens that upsets me, it’s all I can think about for a long time”). (For information on the scale’s psychometric properties, see 100). Regarding this measure, according to some theorists the primary reason given for the decision to attempt suicide is to escape from aversive and intolerable emotion (101-104), which may result from increased emotional reactivity associated with some psychological disorders (105). The rationale for the use of the ERS in the current trial is the same as for the DERS.
SF-36. The SF-36 is a short-form health questionnaire. For the purposes of the current project, several sub-scales have been chosen to track physical and mental wellbeing over time: Physical Functioning; Emotional Well-being; Pain; General Health (see https://www.rand.org/health-care/surveys_tools/mos/36-item-short-form.html; RAND corporation).
Multidimensional Assessment of Interoceptive Awareness (MAIA; 106). The MAIA is a 32-item questionnaire involving 8 scales: (i) Noticing: awareness of uncomfortable, comfortable, and neutral body sensations; (ii) Not-Distracting: tendency not to ignore or distract oneself from sensations of pain or discomfort; (iii) Not-Worrying: tendency not to worry or experience emotional distress with sensations of pain or discomfort; (iv) Attention Regulation: ability to sustain and control attention to body sensations; (v) Emotional Awareness: awareness of the connection between body sensations and emotional states; (vi) Self-Regulation: ability to regulate distress by attention to body sensations; (vii) Body Listening: active listening to the body for insight; (viii) Trusting: experience of one’s body as safe and trustworthy. One advantage of using the MAIA is that it is able to differentiate between interoceptive sensibility that is linked to ‘somatization’ (a problematic hyper-awareness of the body) and interoceptive sensibility linked to well-being.
Predictor variables (both tasks)
- Group membership ([interoception groups] vs. [control];
- Data point/time (baseline vs. follow-up);
- Interoceptive accuracy/sensitivity (continuous measure);
- Interoceptive sensibility (continuous measure);
- Interoceptive metacognitive awareness (continuous measure);
- Heart phase ([systole] vs. [diastole]);
- Valence of emotional stimuli ([positive: stop/signal task only] vs. [negative] vs. [neutral]).
Predictor variables (spatial cueing task only):
- Presence/absence of emotional stimulus (i.e. ‘catch’ trials);
- Cued position on body in the spatial cueing task ([shins] vs. [chest] vs. [lower abdomen]; or [a part of the body cued] vs. [no part of the body cued]);
Dependent variables (both tasks):
- Primary outcome: RTs for spatial cueing task, and SSRT for stop-signal task;
- Scores on self-report measures;
- Heart rate variability, heart rate deceleration.
Dependent variables (spatial cueing task only):
- Frequency of response to emotional stimuli;
- Intensity of experience (continuous);
- Valence reported ([positive] vs. [negative] vs. [neutral]);
- Location of felt experience (point indicated on a human model; to be coded into a categorical measure);
Based on the foregoing detail, the hypotheses for the studies are as follows:
Spatial cueing task
- One of the interoception groups will show significantly faster reaction times (RTs) and greater response frequency to emotional stimuli regardless of cue type compared with each of the other groups after baseline, plus there will be a greater change between baseline and follow-up. Each of the interoception groups separately and combined will show significant differences from the control group.
- One set of cues for the same group (i.e. the interoception group in (1)) will facilitate significantly faster RTs and greater response frequency. This set will match the target area of the body for that group (i.e. a chest cue for the cardioception group; a lower abdomen cue for the gastroception group). The ‘matching’ cue for the other interoception group will not facilitate faster RTs or greater response frequency. If any cue facilitates faster RTs or greater response frequency for the other groups it will be the same cue that facilitates for the interoception group in (1).
- This same group will display greater within-group coherence in the location of the body indicated to be active during trials compared with the other groups after baseline. It will also show the greatest change between baseline and follow-up compared with the other groups.
- This same group will report greater intensity compared with the other groups after baseline. It will also show the greatest change between baseline and follow-up compared with the other groups. The interoception groups combined will report greater intensity than the control group.
- This same group will show greater heart rate deceleration after baseline following the presentation of emotional stimuli compared with the other groups; both interoceptive groups will show greater heart rate deceleration than the control group.
- This same group will show significantly slower RTs on the stop-signal task after baseline and a greater change in RTs between baseline and follow-up compared with the other groups, but with each of the interoceptive groups separately and combined showing significantly slower RTs compared with the control group.
Hypotheses on interoception and emotion regulation:
- Both interoceptive groups will show increases in interoceptive sensitivity (measured by the Schandry task and water load test), sensibility (measured by the MAIA) and metacognitive awareness over the study period, but there will be no significant differences between each other either at baseline or between baseline and follow-up. They will each be significantly different from the control group on these measures. Further, the cardioception group will show the greatest improvement on the Schandry task, and the gastroception group will show the greatest improvement on the water load test.
- The two experimental groups will each show significant improvements over the control group on the DERS, the ERS and the SF-36 (emotional well-being sub-scale) scores between baseline and follow-up, with the same group as in (1) showing greater improvements compared with both of the other groups.
Generalised Estimating Equations (GEE) will be used to analyse the two data points, including variable interactions. GEE can account for repeated measures as well as the potential problem that the primary outcome variable (RT) may not be normally distributed (although it will be continuous). Using GEE, the effects of the predictors on RT can be modelled under different circumstances, and the interrelationships between predictors can also be investigated. Significance testing will be undertaken at the p<0.05 level.
 Across different modes of perception: the task involves a cue in one mode or modality (i.e. visual) but the response relates to another modality (i.e. a tactile perception), hence ‘cross-modally’.
 N.B. Total >100% due to rounding.
 Following communication with the lead author of the publication on which our use of the water load is based.