Compassion-focused imagery (CFI, in which one imagines receiving or giving compassion) can be an effective emotion-regulation technique but some individuals respond as if it is a threat. However, these findings have been based on tasks involving receiving compassion from others. This study sought to examine whether CFI involving self-compassion is less threatening than relaxation and whether any threat-responses decrease with practice. This study will compare the effects of CFI, relaxation and a control task and will explore their effects on self-report symptoms and physiology. It is hypothesized that CFI involving self-compassion is less threatening than relaxation and that any threat-responses decrease with practice.

Introduction

Self-criticism represents a crucial phenomenon in a variety of mental disorders1. Highly self-critical individuals show a poorly regulated threat emotional system and underdeveloped capacities for emotional regulation2,3. Self-criticism is considered a vulnerability risk factor for most psychopathological conditions. Several interventions aiming at increasing compassion as an antidote to self-criticism have been developed4. One of the most empirically validated is Compassion Focused Therapy (CFT, 5), which aims to increase acceptance and compassion for one's own suffering in order to generate a self-soothing response6.

Compassion-focused imagery (CFI) is a key technique in CFT, which involves visualizing compassion towards others, or imagining people, places or objects directing compassion towards oneself7. Single trials of CFI have shown a reduction of negative affect, an increase in self-esteem, and physiological changes associated with the attenuation of threat-focused behaviors8-10. Regular CFI practice has increased self-compassion and reduced negative affect in clinical and non-clinical populations10-13. Despite these promising findings, CFI can create threat-focused responses in some individuals; however, these findings have been based on tasks involving receiving compassion from others.

This study will explore two empirical questions:

1. How do clinical participants initially respond to self-compassionate imagery in an initial trial? The investigators hypothesize that CFI and relaxation imagery would be associated with similar levels of HRV, and that both would be associated with higher HRV than during the control task (both in terms of group means and number of people showing clinically-significant changes). It is also hypothesized that some participants would show reliable increase in positive affect and reliable decrease in negative affect during CFI, but that others would show the opposite pattern (due to inhibitors of compassion).
2. Do any threat responses reduce following repeated trials of self-compassionate imagery? It is hypothesized that the number of individuals responding negatively to CFI would decrease over time.

Methods

Participants will complete an initial screening questionnaire online that consists of reading the study information and completing informed consent, demographics questions, ODSIS14, OASIS15 and FSCRS16. Eligible participants will be invited to attend in-person sessions, in which physiological and self-report responses will be collected. Participants will be asked to avoid exercising, drinking alcohol/caffeine, smoking nicotine and eating during the two hours prior to each session, since these are possible confounds to HRV measurements. Participants will also be asked each session to confirm that they have not made any medication changes, in line with inclusion criteria. HRV will be measured by a research assistant whilst the participant engages in three 4-minute activities, always in the following order: (i) reading a local city culture magazine (control task), (ii) engaging in relaxation imagery involving walking through a forest or on a beach, and (iii) engaging in self-compassion imagery.

Participants will be randomized to complete 3 or 4 trials using a 2:1 ratio. This was decided based upon the premise that more trials are helpful for answering the research questions regarding the impact of repeated trials, but also aiming to minimize dropout for a related study which explored the effects of psychotherapy which was offered following these 3-4 trials.

Although no study to date has explored the hypotheses of this study, the investigators have set a target sample of N=25 based upon other studies exploring effects of CFI on physiology, which have samples ranging from 22 to 25 participants and found significant between-group differences in physiological measures during CFI10,17.

Data analysis

The data will be uploaded to Open Science Framework. Paired sample t-tests will be used to explore differences in RMSSD HRV during the three tasks, and to test changes in positive and negative affect from pre- to post-CFI.

To complement group-level analyses, the investigators will use reliable or clinically-significant change analyses of individuals. A reliable change index (RCI,18) will be calculated for positive and negative affect. Each participant´s change scores (calculated by subtracting pre-CFI affect from post-CFI affect for each trial) will be compared against the RCIs. The investigators will also calculate how many individuals show a clinically-significant HRV response, defined as 5ms change in RMSSD during the experimental task (CFI or relaxation) compared to the control task. THis value was based upon literature on the differences in baseline levels of HRV between healthy controls and patients with depression19,20. The number of positive and negative clinically-significant responses will be compared (i) between relaxation and compassion at trial 1 in order to compare effects of both tasks, and (ii) over 3-4 trials for both relaxation and for compassion, to evaluate whether repeated trials improved response to each task.

Attrition and data loss

Participants will be included if they complete at least one trial of the three tasks, since they can still inform the research questions focused on the initial trial data. Participants will be excluded if HRV data at trial 1 is not interpretable, as this is the primary outcome measure. Individual HRV measurements will be excluded from analysis if the participant reports any of the external threats to HRV data validity mentioned above (e.g. medication changes, high caffeine use, exercise)

Inclusion Criteria:

• Being ≥18 years old
• A clinical level of depression or anxiety (defined as ≥ 8 on the ODSIS14 or OASIS15)
• High self-criticism or low self-reassurance, as measured by the FSCRS16 (≥24 on self-inadequacy, ≥ 8 on self-hatred, or ≤18 on self-reassurance). These cut-offs correspond to 0.5 SD above the mean in self-inadequacy/self-hatred, and 0.5 SD below the mean on self-reassurance, based on a validation of the FSCRS17 in Colombia28.

Exclusion Criteria (Note: these were added subsequent to seeking ethical approval but before study commencement):

• Positive screen on the Mood Disorder Questionnaire (MDQ), following the norms published by the authors
• Positive screen on the Standardised Assessment of Personality - Abbreviated Scale (SAPAS), defined as scoring 4+
• Severity of Dependence Scale (SDS): scoring 3+ for alcohol use or 5+ for other substances.
• Significant suicidal ideation. Participants were asked, "On a scale of 1 to 7, what is your intent to end your life right now?" from 1 (low) to 7 (high), and "Are you uncertain about being able to control suicidal impulses?" (Yes/No). Participants were excluded for responding ≥5 on item 1 or "yes" to item 2