Is the ralationship between interoceptive sensitivity and alexithymia explained by mood m sc

General Scientific Summary Several attempts have been made to establish the links between alexithymia and body perception, especially interoceptive ability.. The problem might lead to m

UNITED KINGDOM SWANSEA UNIVERSITY Abnormal & Clinical Psychology

M.Sc PROJECT

Is the Relationship between Interoceptive Sensitivity

and Alexithymia Explained by Mood?

Student name: LUAN HUYNH NGUYEN

Student number: 921938 Supervisor: DR HAYLEY YOUNG

Year of submission: 2018 Word count: 7969

ACKNOWLEDGEMENT

First and foremost, I would like to express my deepest gratitude to my project supervisor, Dr Hayley Young for believing in my potential, and providing continuous support, patience, enthusiasm and valuable critiques She was not only my supervisor but also

my supporter and inspirer Her guidance shed light on all of my obstacles and motivated me

to work harder to complete this project She always comforts my anxiety and encourages me

to do the best that I can Working with her greatly enriched my experience as an international postgraduate student I appreciate all of her contribution

I also would like to expression my gratitude to all of my professors, lecturers and staff

at Department of Psychology, Swansea University for equipping me knowledge and skills for completing this project and for pursuing my career in this field My special thankfulness is to

Dr Rachael Hunter for being my special personal mentor and supporter

I am so thankful for always receiving care, support and encouragement from my friends and my colleagues at Faculty of Psychology, University of Social Sciences and Humanities, Vietnam National University I would like to express my gratitude to Dr Nga Minh To Hoang for your special support and belief on me

And last but not least, I am forever grateful to my family for their unparalleled love, unfailing emotional support throughout my project and continuous encouragement despite the distance between us Your sacrifice is always in my heart

DECLARATION

CONTENTS

ACKNOWLEDGEMENT i

DECLARATION ii

ABSTRACT iv

INTRODUCTION 1

METHOD 6

Participants 6

Procedures 8

Measures 8

RESULTS 10

Sample characteristics 10

Independent relations between components of interoceptive sensibility, confounds and alexithymia 11

Relative associations between components of interoceptive sensibility and alexithymia 13

Mediation analysis 15

DISCUSSION 18

CONCLUSION 29

REFERENCES 30

APPENDIX 1

APPENDIX A - Ethical Approval 1

APPENDIX B - Participant Information Sheet 2

APPENDIX C - Consent form 3

APPENDIX D - Additional tables 4

APPENDIX E - Distribution scores 9

self-‗experiencing body awareness‘ and ‗trusting body awareness‘ as fundamental factors of IS in relation to alexithymia Crucially, the present research is the first to claim the mediating effects of depression and anxiety on this relationship These findings provided the new pathway to understand the interaction between IS, mood and alexithymia, thus shed light on the influence of mindful attention style and trusting attitude in IS as well as the alexithymia subtypes

General Scientific Summary

Several attempts have been made to establish the links between alexithymia and body perception, especially interoceptive ability The findings were robust but still hindered due to the complexity of related concepts, measuring methods and covariates that need to be controlled In terms of interoception, the trend is to investigate more on sensibility aspect, especially the ability to interpret accurately and appreciate bodily signals The present research will explore the relationship between alexithymia and interoceptive sensibility by utilising the multidimensional approach and consider the mediating effects of potential covariates

INTRODUC TION

As being first described by Sifneos (1973), alexithymia is a personality construct

characterised by difficulties in describing and identifying emotions (DDF and DIF, respectively) and externally oriented thinking style (EOT) (Bagby, Taylor, & Parker, 1994) People with alexithymia typically struggle with describing and identifying different emotions caused by poor emotional vocabulary, then being unable to distinguish emotional reactions from bodily signals They also focus on external events rather than internal senses due to the lack of fantasy and imagination (Bagby et al., 1994) The prevalence of alexithymia in the general population is approximately 7 - 13% and even much higher in psychopathological groups (McGillivray, Becerra, & Harms, 2016; Salminen, Saarijärvi, Äärelä, Toikka, & Kauhanen, 1999) Regarding classification, the concept was classified into primary and secondary alexithymia The former reflects an enduring psychological trait while the latter refers to a dependently contemporary state after the psychological distress (Messina, Beadle,

& Paradiso, 2014; Freyberger, 1977) However, another approach by Moormann et al (2008) proposed five types of alexithymia based on differences in cognitive or affective deficits when using Bermond - Vorst Alexithymia Questionnaire (Vorst & Bermond, 2001) This five-factor model was evident to be more appropriate to examine this notion, compared to the above-mentioned higher-order structure (Bagby et al., 2009) Despite the differences, these classification models could contribute to explain the mixed results of the links between alexithymia and related concepts, especially body perception, among various groups of participants

Body perception was widely known as the significant antecedent of alexithymia Early theories have linked body perception to emotion and emotional processes Damasio (1994) proposed the ―somatic marker hypothesis‖, arguing that emotional processes could guide or bias behavioural outcomes James (1884) also stated that the effect of emotion on the body was bidirectional and the bodily change was a part of the emotional formation

Subjective feelings rely on the active interpretation of physiological changes on our bodies (Seth, Suzuki, & Critchley, 2011) On the contrary, the Cannon-Bard theory (Cannon, 1932; Bard, 1928) explained this link in a reverse direction, which stated that bodily reaction was not the cause but the result of emotions Emotions were first activated by the central nervous system then led to behavioural reactions Although the causal relationship has been controversial, emotion and body perception were evident to be inseparable These theories served as the cornerstone and provided theoretical themes to explain the mind-body connection However, due to the lack of conceptual clarity, empirical studies to further examine the topic were impeded

To date, current researchers have focused on the concept of interoception, the ability

to perceive and interpret internal bodily signals Recent findings have confirmed that interoception was significantly correlated with individual‘s emotional experience (Seth, 2013; Critchley, Wiens, Rotshetein, Öhman, & Dolan, 2004; Craig, 2003), which might increase our understanding of emotional anomalies and disorders However, these studies did not connect interoception with deficits in identifying and interpreting physiological cues (Forrest, Smith, White, & Joiner, 2015) In order to deepen the understanding of interoception and underlying mechanisms that associate this concept with related factors, Garfinkela, Seth, Barrett, Suzuki, and Critchley (2015) have defined three distinct types of interoception Interoceptive accuracy (or interoceptive sensitivity, Murphy et al., 2017) refers to the ability

to perform accurately on objective behavioural measures Interoceptive sensibility (IS) reflects the self-evaluation of subjective interoception by using self-report questionnaires or interviews Interoceptive awareness refers to the metacognitive level of awareness of interoceptive accuracy or our awareness of our objective performance Among these concepts, interoceptive accuracy has gained more attention to be explored with various objective assessment methods (Murphy, Catmur, & Bird, 2017) On the one hand, it could

provide robust evidence for the links between interoception and alexithymia On the other hand, the popularity of this concept could lead to the excessive use of objective measures regarding interoceptive accuracy Therefore, it is of notice to examine other facets of interoception to reflect the concept of interoception more fully

Several investigations have been made to initially explore the relationship between alexithymia and interoception For instance, by using the heartbeat tracking task, Herbert, Herbert, and Pollatos (2011) reported the negatively predictive effect of interoception on alexithymia in all three facets Depression also significantly predicted alexithymia in regression analysis Recent finding with the same method by Shah, Hall, Catmur, and Bird (2016) depicted the negative correlation between alexithymia and atypical interoception but not autistic traits These results emphasised the role of alexithymia as the marker of poor interoception (Murphy, Brewer, Catmur, & Bird, 2017) However, by using some experimental tasks, these studies only focused on the objective aspect, or interoceptive accuracy Therefore, they cannot adequately reflect all aspects of interoception Besides, cardiac-based measures of interoception were problematic and insensitive for the non-alexithymics with low ability levels They could also be distracted by exteroceptive ability by utilising touch receptors of the chest (Murphy et al., 2017) The problem might lead to mixed findings which reported no significant correlation between subjective interoceptive accuracy and alexithymia (i.e., Zamariola, Vlemincx, Corneille, & Lumineta, 2018) Consequently, the focus on subjective aspect of interoception has been applied as a solution to explore other significant factors of IS in relation to alexithymia

Due to the limitations mentioned above, more studies have been conducted to conceptualise IS and its related links Brewer, Cook, and Bird (2016) found that alexithymia was positively correlated with poor IS, measured by two newly designed but non-validated batteries (the Interoceptive Confusion Questionnaire and the State - Emotion Similarity

Questionnaire) The similar finding was also reported by Zamariola et al (2018), using the Multidimensional Assessment of Interoceptive Awareness (MAIA), Interoceptive Awareness Questionnaire and Body Awareness Questionnaire These results suggested that people with high alexithymia were less likely to focus on their body but oversensitive to pain and discomfort They also lacked self-confident with their bodily cues when not feeling their body as safe and trustworthy However, these findings were questionable by the random utilisation of self-report batteries related to body awareness without any particular definition

of IS On the contrary, Longarzo et al (2015) found that the correlation between IS (assessed

by Self-Awareness Questionnaire) and alexithymia was significant but negative They proposed a contradictory argument that the individuals with alexithymia had some cognitive impairment then focused on physiological aspects of emotions only Therefore, they may overrate their IS than their real ability presented in objective tests This contrast in correlational direction of Zamariola‘s and Longarzo‘s studies may result from the difference

in dimensions of IS or the difference in types of alexithymia that were assessed With this in mind, the current study continues to focus on IS and its link with alexithymia by using the multidimensional approach to deepen the understanding of this concept and reassure its predictive effect on alexithymia

Crucially, however, the ability to subjectively perceive our bodily signals cannot solely affect our emotions and behaviours Gross (2015) proposed the situation - attention - appraisal - response model, which emphasised the importance of individuals‘ interpretation and appraisal of stimuli in emotional regulation Burg, Probst, Heidenreich, and Michalak (2017) stated that the way we interpret and appreciate our body sensations would define how

we react to them Therefore, positive interpretation and appreciating attitude can lead to acceptance and prevention of difficult emotions Burg et al (2017) also suggested two facets

of body mindfulness, namely experiencing and appreciating body awareness, which

emphasised on the role of mindful awareness and appreciation in body perception On the contrary, anxiety and hypervigilance-driven thoughts may result in emotional discomfort and maladaptive strategies such as hypochondriasis, anxiety disorders and somatisation (Mehling, 2016) To measure IS in MAIA, Mehling et al (2012) also included two factors ‗body listening and ‗trusting‘ as attitudes towards bodily cues Therefore, it has been more obvious that the interpretation and appreciation aspects are vital components of IS, which the current research would further take into account

Additionally, as it was evident that alexithymia had strong links with different factors, these variables and their effects should be controlled and examined Various findings have reported the significant correlations between alexithymia and gender (Herbert et al., 2011), depression (Honkalampi, Hintikka, Tanskanen, Lehtonen, & Viinama, 2000; Marchesi, Brusamonti, & Maggini, 2000; Parker et al., 1991), anxiety (Karukivi et at., 2010; Marchesi

et al., 2000) and somatisation (Lipsanen, Saarijarvi, Lauerma, 2004) Alexithymia was also highly recorded in the clinical population with depression (Honkalampi et al., 2000; Marchesi

et al., 2000; Parker et al., 1991), anxiety (Karukivi et al., 2010; Marchesi et al., 2000) and those related to body dissatisfaction (Hamilton, 2008; de Berardis et al., 2007) Possibly, it was hypothesised that these factors could also predict alexithymia and mediate the relationship between alexithymia and interoception Previous studies have attempted to control their effects Shah et al (2016) measured all depression, anxiety and autistic quotient

as covariates of alexithymia and interoception Results indicated that alexithymia, not autism, can predict impaired interoceptive accuracy Anxiety traits and states were also found to be a significant predictor at the trend but no longer remained its effect when alexithymia was added Only taking depression into account, Herbert et al (2011) also reported that depression significantly predicted alexithymia total score, DDF and DIF as well However, there was no more mediated analysis with advanced statistics to further explore these effects,

which limited the interpretation and understanding of the associations Besides, mixed results also implied that the links between interoception and different domains of alexithymia could

be affected by different factors Bearing these limitations and inconsistency in mind, the present study will examine the mediating effects of potential covariates of the interaction between IS and alexithymia

Altogether robust evidence has established the strong links between alexithymia and interoception However, these findings were controversial by the disproportionate use of interoceptive accuracy tasks and the lack of conceptual clarity and appropriate measurement for IS and awareness Besides, some possible covariates have not been fully controlled Therefore, it is essential to examine other aspects of interoception and clarify how they relate

to alexithymia under the effect of third variables Thus the present study aims to explore the link between IS and alexithymia, then to examine the influence of mediating factors on this relationship Based on the previous literature it is hypothesised that IS is correlated with alexithymia The correlation is expected to be negative and to vary among different components It is further hypothesised that depression and anxiety can be mediators of the link between IS and alexithymia

METHOD

Participants

Adults males and females who are eighteen and above were recruited to participate in

the online questionnaire (N = 161) Data were collected through Qualtrics online system However, only completed responses were further analysed (N = 132) The sample

characteristics are presented in the Table one

MAIA 132 21.64 5.16

Note N = number of participants; M = mean; SD = standard deviation; BMI =

Body Mass Index; TAS-20 = Toronto Alexithymia Scale - 20; BMQ = Body Mindfulness Questionnaire; MAIA = Multidimensional Assessment of Interoceptive Awareness

The mean age of this sample was M = 27.07 (N = 132, SD = 10.47) with a wide range from 18 to 59 The average BMI was 24.22 (N = 130, SD = 5.73) Participants included 21

males (15.9%) and 111 females (84.1%) Native English speakers were 81 (61.4%) and native speakers were 51 (38.6%) Most of the participants were not taking any medication at

non-the current time (N = 104, 78.8%) while medication users accounted for 21.2% (N = 28)

at any time without any penalty It took approximately 15 - 20 minutes to complete this questionnaire All responses were anonymous and were assessed at one-time point

Measures

Demographics Data related to age, gender, height, weight, currently-used medication

(as BMI and health status may serve as cofactors to body perception) and native language was collected at the beginning of the questionnaire

Toronto Alexithymia Scale (TAS - 20) Alexithymia in the population was measured

using the TAS-20 with three sub-scales, namely, ‗difficulty describing feelings‘ (DDF),

‗difficulty identifying feelings‘ (DIF) and ‗externally oriented thinking‘ (EOT) (Bagby, Parker, & Taylor, 1994) Each statement was rated from 1 to 5 when higher scores reflect more alexithymia In this sample, the Cronbach‘s alphas for the total scale, DDF, DIF and

EOT were 0.89, 0.87, 0.90, and 0.56 respectively Due to the low reliability, EOT was excluded from further analysis of the present study

Multidimensional Assessment of Interoceptive Awareness (MAIA) The 32-item

MAIA (Mehling et al., 2012) was used to assess eight domains of interoceptive body awareness: (1) ‗noticing‘ (awareness of uncomfortable, comfortable and neutral bodily sensations), (2) ‗not distracting‘ (tendency not to distract oneself from physical discomfort), (3) ‗not worrying‘ (tendency not to feel distressed when facing pain and uncomfortable sensations), (4) ‗attention regulation‘ (capacity to maintain attention to body sensations and ignore distractors), (5) ‗emotional awareness‘ (ability to recognise the connection between bodily symptoms and emotions), (6) ‗self-regulation‘ (ability to reduce and regulate distress

by focusing on body sensations), (7) ‗body listening‘ (tendency to actively listen to body for insight) and (8) ‗trusting‘ (capacity to experience one‘s body as secure and trustworthy) Higher scores in MAIA indicate better IS The Cronbach‘s alphas for the total scale and all subscales were 0.91, 0.68, 0.65, 0.67, 0.89, 0.86, 0.81, 0.87 and 0.87, respectively As

‗noticing‘, ‗not distracting‘ and ‗not worrying‘ subscales got low reliability scores, they were not further analysed

Body Mindfulness Questionnaire (BMQ) BMQ (Burg, Probst, Heidenreich, &

Michalak, 2017) was used to measure the concept of body mindfulness, consisting of

‗experiencing body awareness‘ (EBA) (being aware of the general state and compassionate with the body during its natural activities) and ‗appreciating body awareness‘ (ABA) (appreciating one‘s body experience with an openhearted and concerning attention to the body) This scale, together with MAIA, was used to assess individual differences in subjective IS BMQ used 6-point Likert-type scale ranging from 1 to 6 when the higher scores indicate better body mindfulness The reliability of the whole battery and two domains were α = 0.88, 0.91 and 0.91, respectively

Beck’s Depression Inventory (BDI) Given that depression, anxiety and somatisation

could be cofactors that affect alexithymia, they were all assessed and controlled in this study The level of depression was assessed by using BDI (Beck, Ward, Mendelson, Mock, & Erbaugh, 1961) Depression was taken into consideration as it could be a cofactor affecting body perception Twenty-one items regarding main depressive symptoms were scored from 0

to 3 Higher scores indicate more intensive symptoms of depression In the present study, the reliability for this scale was α = 0.94

Beck’s Anxiety Inventory (BAI) BAI (Beck, Epstein, Brown, & Steer, 1988) was

used to measure the severity of anxiety in participants Anxiety could also distort the body image and body awareness Twenty-one common symptoms of anxiety were self-assessed from 1 to 4 when higher scores reflect more severe anxiety The Cronbach‘s alpha for this battery was 0.93

Symptom Checklist 90 - SLC-90 - Somatisation subscale Twelve items of

somatisation were taken from SCL-90 (Derogatis, 1994) to measure distress arising from perceptions of bodily dysfunction Similar to depression and anxiety, this variable was included to examine its co-effect on body perception The severity of these somatic symptoms was evaluated from 0 = not at all to 4 = extremely disturbing during the past week The reliability for SCL - 90 - Somatisation subscale was α = 0.83

RESULTS

Sample characteristics

Initially, as the data were not normally distributed (see Appendix E), bootstrapping

was performed with 1000 samples at the level of 95% of confidence intervals Therefore, no assumption of normality was required for further analysis Next, to ensure that the considering the sample as a whole was appropriate, one-way ANOVA was used to explore

whether alexithymia and body perception scores significantly differed in groups of gender, language and current medication usage Besides, Pearson‘s r correlation was used to examine the association of these scales with age, BMI, depression, anxiety and somatisation (see Appendix D)

Previous studies reported the effect of gender on alexithymia score (Levant, Hall, Williams, & Hasan, 2009; Larsen, van Strien, Eisinga, & Engels, 2006) In this study, independent analysis indicated that there were no significant differences between male and

female groups in DDF (F(1,130) = 0.254, p = 0.615) and DIF (F(1,130) = 1.711, p = 0.193)

Therefore, despite the disproportion of gender frequency, we could combine all responses of

the two groups for further analysis Age was negatively correlated with DDF (r = -0.249, 95% CI [-0.415, -0.085]) but not DIF (r = -0.130, 95% CI [-0.319, 0.060]) BMI did not have any association with both DDF and DIF (r = 0.055 and r = 0.037, 95% CIs [-0.107, 0.208]

all interoceptive scales and subscales of BMQ and MAIA (see Table 3)

Independent relations between components of interoceptive sensibility, confounds and alexithymia

Pearson r correlational analysis was used to examine the independent correlations

between various components of alexithymia and confounds and IS through BMQ and MAIA Bootstrapping with 1000 samples at the 95% confidence level was also performed

Table 3

Independent Correlations between DDF/ DIF and Related Variables

MAIA - Self-

Regulation -.273 -.448 -.092 -.256 -.440 -.077 MAIA - Body

Listening -.185 -.346 -.013 -.040 -.223 .133 MAIA - Trusting -.344 -.491 -.184 -.312 -.489 -.129

Note CI = confidence interval; TAS = Toronto Alexithymia Scale; DDF = Difficulty Describing Feelings; DIF

= Difficulty Identifying Feelings; BMI = Body Mass Index; BMQ = Body Mindfulness Questionnaire; EBA = Experiencing Body Awareness; ABA = Appreciating Body Awareness; MAIA - Multidimensional Assessment

of Interoceptive Awareness; Bold = significant correlation

Confounding factors including depression, anxiety and somatisation had significant

positive correlations with DDF and DIF With DDF, correlation coefficients were r = 0.442,

0.387 and 0.253, 95% CIs [0.290, 0.593], [0.260, 0.518] and [0.080, 0.406], respectively

Similarly, correlations between DIF and these confounds had r = 0.553, 0.540 and 0.427,

95% CIs [0.410, 0.672], [0.420, 0.659] and [0.253, 0.579], respectively

DIF subscale of TAS-20 was negatively associated with the EBA subscale of BMQ

(r= -0.301, 95% CI [-0.456, -0.126]), ‗self-regulation‘ (r = -0.273, 95% CI [-0.448, -0.092]),

‗body listening‘ (r = -0.185, 95% CI [-0.346, -0.013]) and also with the ‗trusting‘ scale of MAIA (r = -0.344, 95% CI [-0.491, -0.184]) In addition, inverse correlations were also

observed between DIF and these interoceptive variables, which include EBA,

‗selfregulation‘ and ‗trusting‘ (r = 0.347, 0.256 and 0.312, 95% CIs [0.513, 0.163], [0.440,

-0.077] and [-0.489, -0.129], respectively) In general, although the above mentioned interoceptive variables were significantly associated with alexithymia scores, the effects were small

Relative associations between components of interoceptive sensibility and alexithymia

As alexithymia was significantly correlated with various interoceptive and confounding variables, it is of importance to assess their unique predictive utility by using linear regression for multiple variables To perform multiple linear regression, essential assumptions were fully examined First, to deal with non-normality and heteroscedasticity of variables, the robust method was applied with 1000 bootstrapped samples at the 95% level of confidence intervals There were no outliers or influential cases Additivity and linearity were also satisfied by checking the variance inflation factor (VIF) and tolerance statistics (see Appendix D) In separate hierarchical models, DDF and DIF were dependent variables while interoceptive factors were entered in the first block of independent variables and significant confounds including depression, anxiety and somatisation were inputted in the second block The interoceptive indices used as independent variables in the models were those that had significant independent relationships with alexithymia These were EBA, ‗self-regulation‘,

‗trusting‘ and ‗body listening‘ for DDF and the same without ‗body listening‘ for DIF

With DDF, the results revealed that both steps of the model could significantly predict

the outcomes The first step was significant with F(3, 126) = 8.932, p < 0.001 and accounted for 17.5% of the variance in DDF The second was also significant with F(6, 123) = 7.394, p

< 0.001, ΔR 2 = 0.090 At the first step, EBA and ‗trusting‘ predicted alexithymia with B =

-0.157 and -1.079, 95% CIs [-0.286, -0.033] and [-2.038, -0.249], respectively However, when depression, anxiety and somatisation were added at step two, neither of these predictors

were longer significant The only factor that predicted DDF was depression (B = 0.110, 95%

CI [0.004, 0.256]) This suggested that the effects of EBA and ‗trusting‘ on DDF scores were reduced under the influence of depression

Note R 2 = 0.183 for Step 1; ΔR 2 = 0.083 for Step 2 (ps = 0.003); EBA = Experiencing Body Awareness

Regression models in which DIF was the dependent variable also shared lots of similarities Both steps of the model were able to predict the difficulty in identifying

emotions with F(3,126) = 9.399, p < 0.001, R 2 = 0.183 for step one and F(6,123) = 13.989, p

< 0.001, ΔR 2 = 0.223 for step two At the first step, EBA was the only negative predictor of

alexithymia (B = -0.266, 95% CI [-0.439, -0.084]) Neither ‗self-regulation‘ nor ‗trusting‘

were significant When confounding factors including depression, anxiety and somatisation

were controlled in the second step, EBA still predicted alexithymia (B = -0.180, 95% CI 0.335, -0.028]) Noticeably, both depression (B = 0.165, 95% CI [0.043, 0.287]) and anxiety (B = 0.138, 95% CI [0.031, 0.252]) were positive predictors of DIF

was used to examine these relationships Hayes PROCESS (Hayes, 2013) model 4 was performed with 1000 bootstrapped samples at 95% confidence interval level

In the previous section, it was reported that the significant effects of EBA and

‗trusting‘ on DDF were diminished once depression was added to the model Therefore, it was considered whether depression mediated the effect of these two interoceptive factors on DDF DDF was entered as the dependent variable (Y) while in turn either EBA or ‗trusting‘ was entered as the predictor (X) and the other was entered as the covariate, to control its effects, and vice versa Depression was tested as a mediator (M) The results revealed that when EBA was controlled as a covariate, there was a significant indirect effect of ‗trusting

body awareness‘ on DDF through depression, b = -0.424, 95% CI [-0.936, -0.084]

Contrastingly, when ‗trusting‘ served as a covariate, the indirect effect of depression on the

relationship between EBA and DDF was not significant, b = -0.044, 95% CI [-0.102, 0.001]

In other words, depression was the mediator of the relationship between DDF and ‗trusting body awareness‘ but not the relationship between DDF and EBA

In terms of DIF, it was reported that only EBA was the significant interoceptive variable Besides, the effect of EBA on DIF was reduced when depression and anxiety were added to the model Therefore, in the mediation analysis for DIF (Y), only EBA could serve

as a predictor (X) while both depression and anxiety were considered as possible mediators (M) In this case, the multi-mediation analysis was applied instead The results indicated that both depression and anxiety significantly mediated the relationship between EBA and DIF,

with the total coefficient b = -0.145, 95% CI [-0.249, -0.053] (b = -0.080 and -0.063, 95% CIs

[-0.159, -0.020] and [-0.133, -0.015] for depression and anxiety, respectively)

Figure 1 Mediation Model of Experiencing, Trusting Body Awareness, Depression and

Difficulty Describing Emotions

Difficulty describing emotions

Experiencing body awareness

Trusting body

awareness

Experiencing body awareness

Depression

Difficulty describing emotions

Indirect effect on Trusting and DDF: b = -0.424, 95% CI [-0.936; -0.084]

Indirect effect on Experiencing and DDF: b = -0.044, 95% CI [-0.102; 0.001]

Figure 2 Mediation Model of Experiencing Body Awareness, Depression, Anxiety and

Difficulty Identifying Emotions

DISCUSSION

The present study aimed to examine the link between alexithymia and IS and the effect of covariates on this relationship We hypothesised that interoceptive sensibility (IS) was inversely correlated with alexithymia and mood could serve as the mediator of this association To confirm these hypotheses, we proposed several main results to further discuss Firstly, difficulties in describing and identifying feelings (DDF and DIF, respectively) were negatively correlated with different aspects of IS such as ‗experiencing body awareness‘, ‗self-regulation‘, ‗body listening‘ and ‗trusting‘ Secondly, when mindfulness was taken into account to measure the observance, accompanying and

Depression

Difficulty identifying emotions

Experiencing body awareness identifying emotions Difficulty

Anxiety

Total indirect effect: b = -0.145, 95% CI [-0.249; -0.053]

Indirect effect of depression: b = -0.080, 95% CI [-0.159; -0.020]

Indirect effect of anxiety: b = -0.063, 95% CI [-0.133; -0.015]

appreciation aspect of body perception, ‗experiencing body awareness‖ (EBA) together with

‗trusting body awareness‘ was evident to be one of the most powerful factors in regression models of alexithymia Finally, as being expected, the effect of EBA on DIF was multi-mediated by depression and anxiety whereas the effect of ‗trusting‘ on DDF was mediated by depression when EBA was controlled

Various IS aspects were evident to be correlated with alexithymia in the present study Difficulties in describing emotions were related to EBA, ‗self-regulation‘, ‗body listening‘ and ‗trusting‘ while difficulties in identifying emotions were related to EBA, ‗self-regulation‘ and ‗trusting‘ All of these correlations were negative In other words, alexithymia gets higher when individuals are not able to observe and accompany with their bodies, regulate their distress by body attention, listen to and feel their body as safe and trustworthy Interestingly, this result replicated the finding of Zamariola et al (2018) which reported the same correlations between alexithymia and MAIA subscales, particularly ‗self-regulation‘, ‗body listening‘ and ‗trusting‘ This consistency supported our first hypothesis that alexithymia, DDF and DIF in particular, were inversely linked with various aspects of IS Thus those with high alexithymia score struggle with bringing back their body awareness to regulate their emotions and get insight Moreover, they tend to hold negative attitudes toward their bodily signals Nevertheless, these findings supported only one out of three dimensions of interoception, which are interoceptive accuracy, interoceptive sensibility and interoceptive awareness (Garfinkela et al., 2015) In terms of objective interoception, Shah et al (2016) and Herbert et al (2011) also proposed a negative correlation between alexithymia and interoceptive accuracy, measured by the heartbeat counting task Therefore, although interoceptive accuracy and awareness were not assessed in this study, our finding also contributed to supporting the argument considering alexithymia as a general deficit of interoception (Murphy et al., 2017; Brewer et al., 2016)

However, it is also worth underlying that significant relationships were observed mostly in more developed factors of IS in MAIA scale These comprising ‗self-regulation‘,

‗body listening‘ and ‗trusting‘ were classified into higher levels of interoception (Mehling et al., 2012) These skills require awareness of not only body sensations but also the integration

of mind and body Mehling et al (2016) also constructed these factors as the more beneficial and mindful ways to focus on and self-regulate bodily symptoms The authors also suggested attention style and regulatory aspects as the main criteria to distinguish two types of IS, namely beneficial and maladaptive style of attention Moreover, Zamariola et al (2018) also reported no significant relationships between alexithymia and ‗distracting‘ and ‗worrying‘ subscales, which indicated a more anxious and hypervigilance-driven style of attention Therefore, it could be suggested that only adaptive attention style of IS negatively linked with alexithymia Difficulties in describing and identifying emotions reduced when individuals developed beneficial and mindful interoceptive attention

Equally important, for the first time body mindfulness was examined and proved to strongly correlate with alexithymia EBA was found to correlate negatively with DDF and DIF and was one of the most powerful factors in regression models Especially, its predicting power for DIF remained despite the mediating effect of depression and anxiety In other words, the ability to observe and ‗be with‘ the body in general status without being distracted was less prevalent in those who suffered difficulties in describing and identifying their emotions This finding was in line with previous studies about the links between general mindfulness and alexithymia For instance, Teixeira and Pereira (2015) indicated the negative correlation between alexithymia and all aspects of mindfulness, comprising awareness, acceptance and quality of mindfulness Cooper, Yap, & Batalha (2018) also reported that alexithymia score significantly diminished after a long-term mindfulness practice Accordingly, neuroimaging evidence has shown the mutual activation in insula while

interoception tasks and mindfulness practice have been conducted (Farb, Segal, & Anderson, 2013; Craig, 2009), suggesting the linked mechanism between IS and awareness of present-moment experience Although mindfulness was differently conceptualised and measured, its connection with alexithymia may be reassured by these consistent results However, no previous studies remarkably focused on body mindfulness as a component of interoception Thus the present study contributes to establishing the link between body mindfulness and difficulties in describing and identifying emotions Besides, as EBA was one of the most powerful factors in predicting DDF and DIF, it could suggest that experiencing our body awareness (observing, being aware of and accompanying) was the fundamental component of

IS In relation to alexithymia, EBA could contain the effects of overlapped concepts and should be accomplished before acquiring other abilities of IS such as body appreciation

Furthermore, the finding of the relationship between EBA and alexithymia is of clinical significance, especially in mindfulness - based practice Greater body awareness gained by attentive observance and accompaniment with bodily signals was related to the decrease of DDF and DIF This empirical finding also supported the mindful emotion regulation model, especially the bottom - up process (Guendelman, Medeiros, & Rampes, 2017) Bottom-up mindfulness-based emotion regulation strategies emphasise the role of body sensations in implicitly affecting emotion regulation It also differed from top-down strategies, which correspond to cognition and thought process Thus in mindfulness-based interventions, explorations of bodily sensations are essential to form any type of emotion and mental content Therefore, this potentially could suggest that interventions for people with alexithymia should not solely address emotional difficulties but need to enhance body awareness and the link between body, mind and emotion For instance, in the Mindfulness-based Cognitive Training programme, the ―opening the door of the body‖ strategy is used to enhance individuals‘ awareness of body sensations related to emotional reactions, step back