The human brain closely monitors body signals essential for our survival, including our heartbeat, our breathing and even the temperature of our skin. This mostly unconscious process is called interoception. It helps people perceive potential or actual threats and helps them to respond appropriately. For example, a person charged by a wild animal will act instinctively to run, fight or freeze. Unlike most creatures, humans show an ability to counteract these survival instincts, and are capable of intentionally engaging in behaviors that result in physical harm. Recent increases in the rate of suicide have made it more urgent to try to understand what leads to this behavior in humans.

Now, DeVille et al. show that people with psychiatric disorders who have survived a suicide attempt have blunted interoception. In four experiments, people with a history of suicide attempts were compared to another group of individuals without a history of suicide attempts. The groups were carefully matched such that there were no significant differences in the demographic and clinical characteristics of the two groups, including in terms of their age, sex, body mass index and psychiatric symptoms.

Both groups completed uncomfortable tasks like holding their breath or keeping their hand in icy cold water. The participants also completed two tasks that required them to focus on their own heartbeat, one of which was paired with functional magnetic resonance imaging. Those with a history of suicide attempts held their breath and kept their hand in cold water for longer, and also were less in tune with their heart rate. This “interoceptive numbing” was associated with less activity in part of the brain called the insular cortex. These differences could not be explained by the individuals having a psychiatric disorder or a history of considering suicide, or by them taking psychiatric medications. Instead, the interoceptive numbing was most often seen in individuals who made an attempt on their own life.

The experiments identify physical characteristics that may differentiate people who attempt suicide from those who do not. This lays the groundwork for future research aimed at identifying biological indicators of suicide risk. More studies are needed to verify the results. If the results are verified, the next step would be prospective studies to determine whether measuring interoception can help clinicians predict who is at risk of a suicide attempt. If it does, it might give clinicians a new tool to try to prevent suicide by ensuring those at greatest risk receive appropriate care.

Suicide ranks among the leading causes of death worldwide (World Health Organization, 2014). In the US alone, suicide increased by nearly 30 percent between 2000 and 2016 (World Health Organization, 2014; Hedegaard et al., 2018). For every death by suicide, it is estimated that there are 25 additional suicide attempts (Hedegaard et al., 2018), each associated with significant social, emotional, and financial consequences. Experts have strived to understand and prevent death by suicide for decades, and yet, our current scientific grasp of the factors that contribute to suicidal behavior is lacking. Moreover, epidemiological data suggest that we are no better at preventing death by suicide than we were 100 years ago (Hedegaard et al., 2018; United States Department of Commerce Bureau of the Census, 1920; Kessler et al., 2005; Nock et al., 2008), with suicide rates rising despite the application of prevention and intervention efforts (Linehan, 2008; Nock, 2016; Paris, 2006).

Theoretical models of suicide have invoked the concept of ‘suicidal capacity’ to differentiate the small subset of individuals who attempt suicide from the much larger group of individuals who experience suicidal ideation but never resort to suicidal action (Ribeiro and Joiner, 2009; Smith and Cukrowicz, 2010). A basic tenet of this concept is the notion that most human beings are ‘hard-wired’ for survival and thus driven to avoid physical pain and threats to bodily homeostasis. Psychological theories of suicide suggest that in an individual with heightened suicidal capacity, certain dispositional (Klonsky and May, 2015) and acquired (Van Orden et al., 2010; Van Orden et al., 2008) traits result in a lower aversion to physical threat and a higher likelihood of transitioning from suicidal ideation to action. Consistent with this line of thinking, non-suicidal self-injury (Klonsky and May, 2014; Franklin et al., 2011) and high levels of fearlessness of the pain involved in dying (May et al., 2012) are behavioral and clinical factors that have been reported to predict suicide attempts. Thus it seems possible that suicidal behavior might be influenced by one’s ability to access and respond adaptively to homeostatic information regarding the internal state of the body, but few studies have directly investigated this topic.

Interoception describes the nervous system’s process of sensing, interpreting, and integrating signals originating from inside the body (Craig, 2002; Khalsa et al., 2018). Emerging evidence suggests that dysfunctions of interoception may contribute to certain mental illnesses (Khalsa et al., 2018; Khalsa and Lapidus, 2016), including mood and anxiety disorders (Paulus and Stein, 2010; Avery et al., 2014; Barrett et al., 2016; Wiebking et al., 2015; Harshaw, 2015; Domschke et al., 2010), substance use disorders (Paulus and Stewart, 2014; Verdejo-Garcia et al., 2012), eating disorders (Kerr et al., 2016; Berner et al., 2018; Khalsa et al., 2015), and nonsuicidal self-injury (Muehlenkamp, 2012), all of which are associated with an elevated risk of suicide (Nock et al., 2010; Harris and Barraclough, 1997; Smith et al., 2018a). Interoception is thought to be substantially supported by the insular cortex, with the primary representation of visceral sensations occurring in the mid-to-posterior insula, and the integration of interoceptive information with cognition, emotion, and other higher order processes occurring in more anterior regions (Barrett and Simmons, 2015; Critchley and Harrison, 2013; Hassanpour et al., 2018).

To test the hypothesis that abnormalities of interoception are associated with suicidal capacity in individuals with psychiatric disorders, we evaluated interoceptive processing in participants with a history of suicide attempts as compared to a matched psychiatric reference sample of participants with no history of suicide attempts. We measured aversive interoceptive processing across the respiratory and nociceptive domains, via an inspiratory breath-hold challenge and a cold-pressor challenge. We assessed cardiac interoception during a heartbeat perception task as well as during a functional magnetic resonance imaging (fMRI) task involving focused attention to heartbeat sensations. We predicted that relative to non-attempters, suicide attempters would 1) tolerate aversive interoceptive sensations to a greater extent, 2) demonstrate lower interoceptive accuracy, and 3) exhibit differences in brain activity in the insular cortex when attending to interoceptive sensations.

We investigated whether attenuated interoceptive processing is associated with self-reported suicide attempts in individuals with a range of psychiatric disorders including depression, anxiety, post-traumatic stress disorder, eating disorders, and/or substance use disorders. We found that suicide attempters show reduced responses to homeostatic threats to the body, including increased tolerance for sensations of air hunger and increased tolerance of cold pain relative to non-attempters. Additionally, we found that suicide attempters exhibit decreased heartbeat perception accuracy and decreased mid and posterior insula activity when attending to sensations from the heart, an interoceptive organ that is vital for maintaining survival. Taken together, these findings provide initial support for the hypothesis that an increased capacity to engage in self-destructive and life-threatening behaviors is associated, behaviorally and neurobiologically, with a blunted sensitivity to internal bodily signals.

Prior to the current investigation, only a few empirical studies have examined the relationship between suicidality and interoception. These studies also suggested a role for interoceptive deficits in suicidal behavior, although the measurement of interoception was restricted solely to self-report questionnaires (Forrest et al., 2015; Dodd et al., 2018; Smith et al., 2018b; Rogers et al., 2018). Here, using a battery of behavioral measures of interoceptive processing, we demonstrate that interoceptive dysfunction in suicide attempters extends beyond symptom measures and includes abnormalities across behavioral, physiological, and neural indices. Furthermore, we observed these differences in a well-characterized sample of participants with similar levels of psychiatric symptoms, increasing the likelihood that the history of suicidal behavior was the driving force behind the differences observed in interoception rather than a varying expression of psychopathology between groups.

Our use of the cold-pressor and inspiratory breath-hold challenges enabled us to examine interoceptive responses to homeostatic threat, revealing that suicide attempters were able to sustain both tasks for a longer period of time than non-attempters. There are several potential explanations for these findings. Suicide attempters may be less sensitive to the physiological cascade that typically follows a homeostatic threat to bodily integrity, allowing them to persist at the task for longer before noticing and responding to internal physiological cues to withdraw. Indeed, the suicide attempters demonstrated an ability to continue the breath-hold task for longer time periods than non-attempters, even in the face of greater reductions in O₂ and increases in CO₂ (i.e. physiological indicators of potential bodily harm). Suicide attempters may also appraise signals of homeostatic threat (e.g. pain, breathlessness) as less salient, resulting in slower withdrawal from aversive homeostatic perturbations. For example, while the suicide attempters sustained breath-hold and cold-pressor perturbations longer than to non-attempters, they did not demonstrate the expected corresponding increases in unpleasantness, stress, difficulty, and other ratings of task aversiveness. Rather, in the cold-pressor challenge, the suicide attempters’ ratings of stress and difficulty were marginally lower than that of non-attempters, even though they continued the cold-pressor challenge for an average of 18 s longer. Based on the current findings, we speculate that a disconnection between the physiological and affective experiences of pain may enable such individuals to engage in self-injurious behaviors and to fail to withdraw from painful stimuli, possibly to the point of inflicting physical harm.

Suicide attempters, relative to non-attempters, also displayed reduced interoceptive accuracy for heartbeat sensations and decreased hemodynamic responses in the right dorsal mid and posterior insula during interoceptive attention to naturally-occurring heart sensations. In terms of the importance of these regions, the mid/posterior insula is often presumed to be the primary recipient of thermal, nociceptive, tactile, and cardiovascular inputs from the ventromedial thalamic nuclei (Craig, 2002; Critchley and Harrison, 2013), supporting the subsequent representation of conscious awareness of internal bodily states. Prior research suggests that the dorsal mid-insula is sensitive to homeostatic signals (Simmons et al., 2013b) and critical to the brain’s representation of visceral stimulation (Hassanpour et al., 2018), visceral attention (Farb et al., 2013; Simmons et al., 2013a), and interoceptive memory (DeVille et al., 2018). From a clinical perspective, dysfunction in the dorsal mid-insula has been linked to somatic symptoms (Avery et al., 2014) and failure to meet energy needs (Simmons et al., 2016) among individuals with depression, suggesting that functional abnormalities within this region may be associated with the inability to effectively attend to and use homeostatically relevant information from the body. Additionally, the posterior insula has been conclusively linked to the detection of aversive sensory stimuli and associated shifts in behavioral response strategies in mice (Gehrlach et al., 2019), and it is a brain region that is closely tied to the amygdala and threat processing (Berret et al., 2019; Livneh et al., 2017). Meta-analytic functional neuroimaging findings in humans have specifically implicated the mid/posterior insula in cardiac interoception, with a particular role in attending to and accurately perceiving naturally occurring cardiac sensations (Schulz, 2016). Collectively, these studies illustrate that the observed blunting of insular cortex activation in suicide attempters is precisely located within subregions that are closely relevant for aversive threat processing, conscious awareness of the heartbeat signal, and homeostatic regulation.

Prior theoretical work has suggested that a low signal-to-noise ratio of visceral input to the insula may influence interoception among individuals with certain psychiatric illnesses, who are thought to be less capable of discriminating random signal fluctuations from biologically relevant signals (Paulus and Stein, 2010; Barrett et al., 2016; Barrett and Simmons, 2015). As a result, it is thought that these individuals tend to erroneously evaluate benign signal changes as significant, and plan and act on these signals accordingly (Paulus et al., 2019). In this view, the suicide attempters’ reduced mid and posterior insula activation during heartbeat interoception could reflect a ‘noisier’ or weaker processing of interoceptive afferents, potentially interfering with the ability to accurately detect signals from their bodies. To extend this line of thought, it is also possible that difficulty distinguishing signal from noise manifests as a reduced ability to adaptively detect homeostatically-relevant signals (e.g. those related to pain and/or suffocation, as in the cold-pressor and breath-hold tasks), which could help to explain the group differences in homeostatic threat response observed in the current study. However, these notions remain tentative until further research can replicate and extend the findings across other measures of response to aversive and/or painful interoceptive stimuli.

We found that suicide attempters exhibited reduced activation within cortical midline structures during cardiac interoceptive attention (i.e. dorsomedial prefrontal cortex, posterior cingulate, and precuneus) that have been previously implicated as core brain regions underlying human self-referential processing (Northoff and Bermpohl, 2004; Northoff et al., 2006; Lemogne et al., 2011). Our finding is consistent with prior research linking abnormal self-referential processing to suicidal ideation and behavior (Marchand et al., 2012) and suggests that further exploration for a role of interoception in the relationship between interoception and self-referential processing in suicide may be worthwhile. Suicide attempters also exhibited reduced activation within the superior temporal gyrus during interoceptive attention. Prior studies have linked suicide attempts with structural and functional abnormalities within the superior temporal gyrus (Soloff et al., 2012; Pan et al., 2015; Aguilar et al., 2008; van Heeringen et al., 2014), arguing for a role in socio-emotional threat evaluation, but further investigation would be necessary to pinpoint the role of interoception in this relationship.

Our findings provide some key empirical support for modern theoretical models of suicide that are built upon desire-capability frameworks (Klonsky et al., 2017), such as the motivational-volitional model (O'Connor and Kirtley, 2018), the interpersonal-psychological theory of suicide (Ribeiro and Joiner, 2009; Van Orden et al., 2010), and the three-step theory (Klonsky and May, 2015). One assertion shared by each of these models is that, for suicide attempts and/or death by suicide to occur, the suicidal individual must express the capacity to approach (rather than avoid) bodily harm. According to these theories, individuals who engage in suicidal behavior demonstrate an ability to ignore—or override—the natural aversion to bodily harm that protects humans against physical injury. Our observations are consistent with these theories and provide early experimental evidence for the role of blunted interoception and heightened tolerance for homeostatic threat in suicidal attempts.

We are unable to publicly release our complete raw dataset due to the fact that our research engaged human participants to examine a very sensitive topic (history of suicide attempt), and the fact that our dataset contains information that could potentially identify individuals. However, should another investigator wish to obtain access to other aspects of our data for the purposes of re-analysis or verification of our findings, we would be willing to facilitate this process upon request, provided that the confidentiality of the participants can be protected. Individuals who are interested in obtaining access to this data should contact the corresponding author. De-identified processed source data and R code corresponding to each figure has been uploaded.

1. 1. Aguilar EJ
   2. García-Martí G
   3. Martí-Bonmatí L
   4. Lull JJ
   5. Moratal D
   6. Escartí MJ
   7. Robles M
   8. González JC
   9. Guillamón MI
   10. Sanjuán J

   (2008) Left orbitofrontal and superior temporal gyrus structural changes associated to suicidal behavior in patients with schizophrenia

   Progress in Neuro-Psychopharmacology and Biological Psychiatry 32:1673–1676.

   https://doi.org/10.1016/j.pnpbp.2008.06.016

   • PubMed
   • Google Scholar
2. 1. Aupperle RL
   2. Paulus MP
   3. Kuplicki R
   4. Touthang J
   5. Victor T
   6. Yeh HW
   7. Khalsa SS
   8. Tulsa 1000 Investigators

   (2020) Web-Based graphic representation of the life course of mental health: cross-sectional study across the spectrum of mood, anxiety, eating, and Substance use disorders

   JMIR Mental Health 7:e16919.

   https://doi.org/10.2196/16919

   • PubMed
   • Google Scholar
3. 1. Avery JA
   2. Drevets WC
   3. Moseman SE
   4. Bodurka J
   5. Barcalow JC
   6. Simmons WK

   (2014) Major depressive disorder is associated with abnormal interoceptive activity and functional connectivity in the insula

   Biological Psychiatry 76:258–266.

   https://doi.org/10.1016/j.biopsych.2013.11.027

   • PubMed
   • Google Scholar
4. 1. Avery JA
   2. Burrows K
   3. Kerr KL
   4. Bodurka J
   5. Khalsa SS
   6. Paulus MP
   7. Simmons WK

   (2017) How the brain wants what the body needs: the neural basis of positive alliesthesia

   Neuropsychopharmacology 42:822–830.

   https://doi.org/10.1038/npp.2016.128

   • PubMed
   • Google Scholar
5. 1. Bagby RM
   2. Parker JD
   3. Taylor GJ

   (1994) The twenty-item Toronto alexithymia scale--I. item selection and cross-validation of the factor structure

   Journal of Psychosomatic Research 38:23–32.

   https://doi.org/10.1016/0022-3999(94)90005-1

   • PubMed
   • Google Scholar
6. 1. Barrett LF
   2. Quigley KS
   3. Hamilton P

   (2016) An active inference theory of allostasis and interoception in depression

   Philosophical Transactions of the Royal Society B: Biological Sciences 371:20160011.

   https://doi.org/10.1098/rstb.2016.0011

   • Google Scholar
7. 1. Barrett LF
   2. Simmons WK

   (2015) Interoceptive predictions in the brain

   Nature Reviews Neuroscience 16:419–429.

   https://doi.org/10.1038/nrn3950

   • Google Scholar
8. 1. Berner LA
   2. Simmons AN
   3. Wierenga CE
   4. Bischoff-Grethe A
   5. Paulus MP
   6. Bailer UF
   7. Ely AV
   8. Kaye WH

   (2018) Altered interoceptive activation before, during, and after aversive breathing load in women remitted from anorexia nervosa

   Psychological Medicine 48:142–154.

   https://doi.org/10.1017/S0033291717001635

   • PubMed
   • Google Scholar
9. 1. Berret E
   2. Kintscher M
   3. Palchaudhuri S
   4. Tang W
   5. Osypenko D
   6. Kochubey O
   7. Schneggenburger R

   (2019) Insular cortex processes aversive somatosensory information and is crucial for threat learning

   Science 364:eaaw0474.

   https://doi.org/10.1126/science.aaw0474

   • Google Scholar
10. 1. Campbell-Sills L
    2. Norman SB
    3. Craske MG
    4. Sullivan G
    5. Lang AJ
    6. Chavira DA
    7. Bystritsky A
    8. Sherbourne C
    9. Roy-Byrne P
    10. Stein MB

    (2009) Validation of a brief measure of anxiety-related severity and impairment: the overall anxiety severity and impairment scale (OASIS)

    Journal of Affective Disorders 112:92–101.

    https://doi.org/10.1016/j.jad.2008.03.014

    • PubMed
    • Google Scholar
11. 1. Cox RW

    (1996) AFNI: software for analysis and visualization of functional magnetic resonance Neuroimages

    Computers and Biomedical Research 29:162–173.

    https://doi.org/10.1006/cbmr.1996.0014

    • PubMed
    • Google Scholar
12. 1. Cox RW
    2. Chen G
    3. Glen DR
    4. Reynolds RC
    5. Taylor PA

    (2017) FMRI clustering in AFNI: false-positive rates redux

    Brain Connectivity 7:152–171.

    https://doi.org/10.1089/brain.2016.0475

    • PubMed
    • Google Scholar
13. 1. Craig AD

    (2002) How do you feel? interoception: the sense of the physiological condition of the body

    Nature Reviews Neuroscience 3:655–666.

    https://doi.org/10.1038/nrn894

    • PubMed
    • Google Scholar
14. 1. Critchley HD
    2. Harrison NA

    (2013) Visceral influences on brain and behavior

    Neuron 77:624–638.

    https://doi.org/10.1016/j.neuron.2013.02.008

    • PubMed
    • Google Scholar
15. 1. De H
    2. Imai K
    3. King G
    4. Stuart EA

    (2011) MatchIt: nonparametric preprocessing for parametric causal inference

    Journal of Statistical Software 42:1–28.

    https://doi.org/10.18637/jss.v042.i08

    • Google Scholar
16. 1. Desmedt O
    2. Luminet O
    3. Corneille O

    (2018) The heartbeat counting task largely involves non-interoceptive processes: evidence from both the original and an adapted counting task

    Biological Psychology 138:185–188.

    https://doi.org/10.1016/j.biopsycho.2018.09.004

    • PubMed
    • Google Scholar
17. 1. DeVille DC
    2. Kerr KL
    3. Avery JA
    4. Burrows K
    5. Bodurka J
    6. Feinstein JS
    7. Khalsa SS
    8. Paulus MP
    9. Simmons WK

    (2018) The neural bases of interoceptive encoding and recall in healthy adults and adults with depression

    Biological Psychiatry: Cognitive Neuroscience and Neuroimaging 3:546–554.

    https://doi.org/10.1016/j.bpsc.2018.03.010

    • Google Scholar
18. 1. Dodd DR
    2. Smith AR
    3. Forrest LN
    4. Witte TK
    5. Bodell L
    6. Bartlett M
    7. Siegfried N
    8. Goodwin N

    (2018) Interoceptive deficits, nonsuicidal Self-Injury, and suicide attempts among women with eating disorders

    Suicide and Life-Threatening Behavior 48:438–448.

    https://doi.org/10.1111/sltb.12383

    • PubMed
    • Google Scholar
19. 1. Domschke K
    2. Stevens S
    3. Pfleiderer B
    4. Gerlach AL

    (2010) Interoceptive sensitivity in anxiety and anxiety disorders: an overview and integration of neurobiological findings

    Clinical Psychology Review 30:1–11.

    https://doi.org/10.1016/j.cpr.2009.08.008

    • PubMed
    • Google Scholar
20. 1. Edwards LJ
    2. Muller KE
    3. Wolfinger RD
    4. Qaqish BF
    5. Schabenberger O

    (2008) An R2 statistic for fixed effects in the linear mixed model

    Statistics in Medicine 27:6137–6157.

    https://doi.org/10.1002/sim.3429

    • PubMed
    • Google Scholar
21. 1. Farb NA
    2. Segal ZV
    3. Anderson AK

    (2013) Attentional modulation of primary interoceptive and exteroceptive cortices

    Cerebral Cortex 23:114–126.

    https://doi.org/10.1093/cercor/bhr385

    • PubMed
    • Google Scholar
22. 1. Forrest LN
    2. Smith AR
    3. White RD
    4. Joiner TE

    (2015) (Dis)connected: an examination of interoception in individuals with suicidality

    Journal of Abnormal Psychology 124:754–763.

    https://doi.org/10.1037/abn0000074

    • PubMed
    • Google Scholar
23. 1. Franklin JC
    2. Hessel ET
    3. Prinstein MJ

    (2011) Clarifying the role of pain tolerance in suicidal capability

    Psychiatry Research 189:362–367.

    https://doi.org/10.1016/j.psychres.2011.08.001

    • Google Scholar
24. 1. Gehrlach DA
    2. Dolensek N
    3. Klein AS
    4. Roy Chowdhury R
    5. Matthys A
    6. Junghänel M
    7. Gaitanos TN
    8. Podgornik A
    9. Black TD
    10. Reddy Vaka N
    11. Conzelmann KK
    12. Gogolla N

    (2019) Aversive state processing in the posterior insular cortex

    Nature Neuroscience 22:1424–1437.

    https://doi.org/10.1038/s41593-019-0469-1

    • PubMed
    • Google Scholar
25. 1. Harris EC
    2. Barraclough B

    (1997) Suicide as an outcome for mental disorders A meta-analysis

    The British Journal of Psychiatry : The Journal of Mental Science 170:205–228.

    https://doi.org/10.1192/bjp.170.3.205

    • PubMed
    • Google Scholar
26. 1. Harshaw C

    (2015) Interoceptive dysfunction: toward an integrated framework for understanding somatic and affective disturbance in depression

    Psychological Bulletin 141:311–363.

    https://doi.org/10.1037/a0038101

    • PubMed
    • Google Scholar
27. 1. Hassanpour MS
    2. Simmons WK
    3. Feinstein JS
    4. Luo Q
    5. Lapidus RC
    6. Bodurka J
    7. Paulus MP
    8. Khalsa SS

    (2018) The insular cortex dynamically maps changes in cardiorespiratory interoception

    Neuropsychopharmacology 43:426–434.

    https://doi.org/10.1038/npp.2017.154

    • PubMed
    • Google Scholar
28. 1. Hedegaard H
    2. Curtin SC
    3. Warner M

    (2018)

    Suicide rates in the united states continue to increase

    NCHS Data Brief pp. 1–8.

    • PubMed
    • Google Scholar
29. 1. Jaeger B

    (2017)

    Package ‘r2glmm’. R Found Stat Comput Vienna available CRAN R-project org/package = R2glmm

    Package ‘r2glmm’. R Found Stat Comput Vienna available CRAN R-project org/package = R2glmm.

    • Google Scholar
30. 1. Kerr KL
    2. Moseman SE
    3. Avery JA
    4. Bodurka J
    5. Zucker NL
    6. Simmons WK

    (2016) Altered insula activity during visceral interoception in Weight-Restored patients with anorexia nervosa

    Neuropsychopharmacology 41:521–528.

    https://doi.org/10.1038/npp.2015.174

    • PubMed
    • Google Scholar
31. 1. Kessler RC
    2. Berglund P
    3. Borges G
    4. Nock M
    5. Wang PS

    (2005) Trends in suicide ideation, plans, gestures, and attempts in the united states, 1990-1992 to 2001-2003

    Jama 293:2487–2495.

    https://doi.org/10.1001/jama.293.20.2487

    • PubMed
    • Google Scholar
32. 1. Khalsa SS
    2. Rudrauf D
    3. Sandesara C
    4. Olshansky B
    5. Tranel D

    (2009) Bolus isoproterenol infusions provide a reliable method for assessing interoceptive awareness

    International Journal of Psychophysiology 72:34–45.

    https://doi.org/10.1016/j.ijpsycho.2008.08.010

    • Google Scholar
33. 1. Khalsa SS
    2. Craske MG
    3. Li W
    4. Vangala S
    5. Strober M
    6. Feusner JD

    (2015) Altered interoceptive awareness in anorexia nervosa: effects of meal anticipation, consumption and bodily arousal

    International Journal of Eating Disorders 48:889–897.

    https://doi.org/10.1002/eat.22387

    • PubMed
    • Google Scholar
34. 1. Khalsa SS
    2. Adolphs R
    3. Cameron OG
    4. Critchley HD
    5. Davenport PW
    6. Feinstein JS
    7. Feusner JD
    8. Garfinkel SN
    9. Lane RD
    10. Mehling WE
    11. Meuret AE
    12. Nemeroff CB
    13. Oppenheimer S
    14. Petzschner FH
    15. Pollatos O
    16. Rhudy JL
    17. Schramm LP
    18. Simmons WK
    19. Stein MB
    20. Stephan KE
    21. Van den Bergh O
    22. Van Diest I
    23. von Leupoldt A
    24. Paulus MP
    25. Ainley V
    26. Al Zoubi O
    27. Aupperle R
    28. Avery J
    29. Baxter L
    30. Benke C
    31. Berner L
    32. Bodurka J
    33. Breese E
    34. Brown T
    35. Burrows K
    36. Cha Y-H
    37. Clausen A
    38. Cosgrove K
    39. Deville D
    40. Duncan L
    41. Duquette P
    42. Ekhtiari H
    43. Fine T
    44. Ford B
    45. Garcia Cordero I
    46. Gleghorn D
    47. Guereca Y
    48. Harrison NA
    49. Hassanpour M
    50. Hechler T
    51. Heller A
    52. Hellman N
    53. Herbert B
    54. Jarrahi B
    55. Kerr K
    56. Kirlic N
    57. Klabunde M
    58. Kraynak T
    59. Kriegsman M
    60. Kroll J
    61. Kuplicki R
    62. Lapidus R
    63. Le T
    64. Hagen KL
    65. Mayeli A
    66. Morris A
    67. Naqvi N
    68. Oldroyd K
    69. Pané-Farré C
    70. Phillips R
    71. Poppa T
    72. Potter W
    73. Puhl M
    74. Safron A
    75. Sala M
    76. Savitz J
    77. Saxon H
    78. Schoenhals W
    79. Stanwell-Smith C
    80. Teed A
    81. Terasawa Y
    82. Thompson K
    83. Toups M
    84. Umeda S
    85. Upshaw V
    86. Victor T
    87. Wierenga C
    88. Wohlrab C
    89. Yeh H
    90. Yoris A
    91. Zeidan F
    92. Zotev V
    93. Zucker N

    (2018) Interoception and mental health: a roadmap

    Biological Psychiatry: Cognitive Neuroscience and Neuroimaging 3:501–513.

    https://doi.org/10.1016/j.bpsc.2017.12.004

    • Google Scholar
35. 1. Khalsa SS
    2. Lapidus RC

    (2016) Can interoception improve the pragmatic search for biomarkers in psychiatry?

    Frontiers in Psychiatry 7:121.

    https://doi.org/10.3389/fpsyt.2016.00121

    • PubMed
    • Google Scholar
36. 1. Klonsky ED
    2. Qiu T
    3. Saffer BY

    (2017) Recent advances in differentiating suicide attempters from suicide ideators

    Current Opinion in Psychiatry 30:15–20.

    https://doi.org/10.1097/YCO.0000000000000294

    • PubMed
    • Google Scholar
37. 1. Klonsky ED
    2. May AM

    (2014) Differentiating suicide attempters from suicide ideators: a critical frontier for suicidology research

    Suicide and Life-Threatening Behavior 44:1–5.

    https://doi.org/10.1111/sltb.12068

    • PubMed
    • Google Scholar
38. 1. Klonsky ED
    2. May AM

    (2015) The Three-Step theory (3ST): A new theory of suicide rooted in the “Ideation-to-Action” Framework

    International Journal of Cognitive Therapy 8:114–129.

    https://doi.org/10.1521/ijct.2015.8.2.114

    • Google Scholar
39. 1. Kroenke K
    2. Spitzer RL
    3. Williams JB

    (2001) The PHQ-9: validity of a brief depression severity measure

    Journal of General Internal Medicine 16:606–613.

    https://doi.org/10.1046/j.1525-1497.2001.016009606.x

    • PubMed
    • Google Scholar
40. 1. Kuznetsova A
    2. Brockhoff PB
    3. Christensen RHB

    (2017) lmerTest Package: Tests in Linear Mixed Effects Models

    Journal of Statistical Software 82:82.

    https://doi.org/10.18637/jss.v082.i13

    • Google Scholar
41. 1. Lemogne C
    2. Gorwood P
    3. Bergouignan L
    4. Pélissolo A
    5. Lehéricy S
    6. Fossati P

    (2011) Negative affectivity, self-referential processing and the cortical midline structures

    Social Cognitive and Affective Neuroscience 6:426–433.

    https://doi.org/10.1093/scan/nsq049

    • PubMed
    • Google Scholar
42. 1. Linehan MM

    (2008) Suicide intervention research: a field in desperate need of development

    Suicide and Life-Threatening Behavior 38:483–485.

    https://doi.org/10.1521/suli.2008.38.5.483

    • PubMed
    • Google Scholar
43. 1. Livneh Y
    2. Ramesh RN
    3. Burgess CR
    4. Levandowski KM
    5. Madara JC
    6. Fenselau H
    7. Goldey GJ
    8. Diaz VE
    9. Jikomes N
    10. Resch JM
    11. Lowell BB
    12. Andermann ML

    (2017) Homeostatic circuits selectively gate food cue responses in insular cortex

    Nature 546:611–616.

    https://doi.org/10.1038/nature22375

    • PubMed
    • Google Scholar
44. Software

    1. Lüdecke D

    (2018)

    sjPlot: data visualization for statistics in social science

    R Package Version.
45. 1. Marchand WR
    2. Lee JN
    3. Johnson S
    4. Thatcher J
    5. Gale P
    6. Wood N
    7. Jeong EK

    (2012) Striatal and cortical midline circuits in major depression: implications for suicide and symptom expression

    Progress in Neuro-Psychopharmacology and Biological Psychiatry 36:290–299.

    https://doi.org/10.1016/j.pnpbp.2011.10.016

    • PubMed
    • Google Scholar
46. 1. May AM
    2. Klonsky ED
    3. Klein DN

    (2012) Predicting future suicide attempts among depressed suicide ideators: a 10-year longitudinal study

    Journal of Psychiatric Research 46:946–952.

    https://doi.org/10.1016/j.jpsychires.2012.04.009

    • PubMed
    • Google Scholar
47. 1. McCabe SE
    2. Boyd CJ
    3. Cranford JA
    4. Morales M
    5. Slayden J

    (2006) A modified version of the drug abuse screening test among undergraduate students

    Journal of Substance Abuse Treatment 31:297–303.

    https://doi.org/10.1016/j.jsat.2006.04.010

    • PubMed
    • Google Scholar
48. 1. Mehling WE
    2. Price C
    3. Daubenmier JJ
    4. Acree M
    5. Bartmess E
    6. Stewart A

    (2012) The multidimensional assessment of interoceptive awareness (MAIA)

    PLOS ONE 7:e48230.

    https://doi.org/10.1371/journal.pone.0048230

    • PubMed
    • Google Scholar
49. 1. Millner AJ
    2. Lee MD
    3. Nock MK

    (2017) Describing and measuring the pathway to suicide attempts: a preliminary study

    Suicide and Life-Threatening Behavior 47:353–369.

    https://doi.org/10.1111/sltb.12284

    • PubMed
    • Google Scholar
50. 1. Morgan JF
    2. Reid F
    3. Lacey JH

    (2000) The SCOFF questionnaire: a new screening tool for eating disorders

    Western Journal of Medicine 172:164–165.

    https://doi.org/10.1136/ewjm.172.3.164

    • PubMed
    • Google Scholar
51. 1. Muehlenkamp JJ

    (2012) Body regard in nonsuicidal Self-Injury: theoretical explanations and treatment directions

    Journal of Cognitive Psychotherapy 26:331–347.

    https://doi.org/10.1891/0889-8391.26.4.331

    • Google Scholar
52. 1. Murphy J
    2. Millgate E
    3. Geary H
    4. Ichijo E
    5. Coll MP
    6. Brewer R
    7. Catmur C
    8. Bird G

    (2018) Knowledge of resting heart rate mediates the relationship between intelligence and the heartbeat counting task

    Biological Psychology 133:1–3.

    https://doi.org/10.1016/j.biopsycho.2018.01.012

    • PubMed
    • Google Scholar
53. 1. Nock MK
    2. Borges G
    3. Bromet EJ
    4. Cha CB
    5. Kessler RC
    6. Lee S

    (2008) Suicide and suicidal behavior

    Epidemiologic Reviews 30:133–154.

    https://doi.org/10.1093/epirev/mxn002

    • PubMed
    • Google Scholar
54. 1. Nock MK
    2. Hwang I
    3. Sampson NA
    4. Kessler RC

    (2010) Mental disorders, comorbidity and suicidal behavior: results from the national comorbidity survey replication

    Molecular Psychiatry 15:868–876.

    https://doi.org/10.1038/mp.2009.29

    • PubMed
    • Google Scholar
55. 1. Nock MK

    (2016) Recent and needed advances in the understanding, prediction, and prevention of suicidal behavior

    Depression and Anxiety 33:460–463.

    https://doi.org/10.1002/da.22528

    • PubMed
    • Google Scholar
56. 1. Northoff G
    2. Heinzel A
    3. de Greck M
    4. Bermpohl F
    5. Dobrowolny H
    6. Panksepp J

    (2006) Self-referential processing in our brain--a meta-analysis of imaging studies on the self

    NeuroImage 31:440–457.

    https://doi.org/10.1016/j.neuroimage.2005.12.002

    • PubMed
    • Google Scholar
57. 1. Northoff G
    2. Bermpohl F

    (2004) Cortical midline structures and the self

    Trends in Cognitive Sciences 8:102–107.

    https://doi.org/10.1016/j.tics.2004.01.004

    • PubMed
    • Google Scholar
58. 1. O'Connor RC
    2. Kirtley OJ

    (2018) The integrated motivational-volitional model of suicidal behaviour

    Philosophical Transactions of the Royal Society B: Biological Sciences 373:20170268.

    https://doi.org/10.1098/rstb.2017.0268

    • PubMed
    • Google Scholar
59. 1. Pan LA
    2. Ramos L
    3. Segreti A
    4. Brent DA
    5. Phillips ML

    (2015) Right superior temporal gyrus volume in adolescents with a history of suicide attempt

    British Journal of Psychiatry 206:339–340.

    https://doi.org/10.1192/bjp.bp.114.151316

    • PubMed
    • Google Scholar
60. 1. Paris J

    (2006) Predicting and preventing suicide: do we know enough to do either?

    Harvard Review of Psychiatry 14:233–240.

    https://doi.org/10.1080/10673220600968662

    • PubMed
    • Google Scholar
61. 1. Paulus MP
    2. Feinstein JS
    3. Khalsa SS

    (2019) An active inference approach to interoceptive psychopathology

    Annual Review of Clinical Psychology 15:97–122.

    https://doi.org/10.1146/annurev-clinpsy-050718-095617

    • PubMed
    • Google Scholar
62. 1. Paulus MP
    2. Stein MB

    (2010) Interoception in anxiety and depression

    Brain Structure and Function 214:451–463.

    https://doi.org/10.1007/s00429-010-0258-9

    • PubMed
    • Google Scholar
63. 1. Paulus MP
    2. Stewart JL

    (2014) Interoception and drug addiction

    Neuropharmacology 76:342–350.

    https://doi.org/10.1016/j.neuropharm.2013.07.002

    • PubMed
    • Google Scholar
64. 1. Pollatos O
    2. Füstös J
    3. Critchley HD

    (2012) On the generalised embodiment of pain: how interoceptive sensitivity modulates cutaneous pain perception

    Pain 153:1680–1686.

    https://doi.org/10.1016/j.pain.2012.04.030

    • PubMed
    • Google Scholar
65. 1. Posner K
    2. Brown GK
    3. Stanley B
    4. Brent DA
    5. Yershova KV
    6. Oquendo MA
    7. Currier GW
    8. Melvin GA
    9. Greenhill L
    10. Shen S
    11. Mann JJ

    (2011) The Columbia-Suicide severity rating scale: initial validity and internal consistency findings from three multisite studies with adolescents and adults

    American Journal of Psychiatry 168:1266–1277.

    https://doi.org/10.1176/appi.ajp.2011.10111704

    • PubMed
    • Google Scholar
66. 1. R Development Core Team

    (2013) R: A language and environment for statistical computing

    R: A language and environment for statistical computing, Vienna, Austria, http://www.r-project.org.

    http://www.r-project.org

    • Google Scholar
67. 1. Ribeiro JD
    2. Joiner TE

    (2009) The interpersonal-psychological theory of suicidal behavior: current status and future directions

    Journal of Clinical Psychology 65:1291–1299.

    https://doi.org/10.1002/jclp.20621

    • Google Scholar
68. 1. Rogers ML
    2. Hagan CR
    3. Joiner TE

    (2018) Examination of interoception along the suicidality continuum

    Journal of Clinical Psychology 74:1004–1016.

    https://doi.org/10.1002/jclp.22564

    • PubMed
    • Google Scholar
69. 1. Schandry R

    (1981) Heart beat perception and emotional experience

    Psychophysiology 18:483–488.

    https://doi.org/10.1111/j.1469-8986.1981.tb02486.x

    • PubMed
    • Google Scholar
70. 1. Schulz SM

    (2016) Neural correlates of heart-focused interoception: a functional magnetic resonance imaging meta-analysis

    Philosophical Transactions of the Royal Society B: Biological Sciences 371:20160018.

    https://doi.org/10.1098/rstb.2016.0018

    • Google Scholar
71. 1. Sheehan DV
    2. Lecrubier Y
    3. Sheehan KH
    4. Amorim P
    5. Janavs J
    6. Weiller E
    7. Hergueta T
    8. Baker R
    9. Dunbar GC

    (1998)

    The Mini-International neuropsychiatric interview (M.I.N.I): The development and validation of a structured diagnostic psychiatric interview for DSM-IV and ICD-10

    The Journal of Clinical Psychiatry 59:22–33.

    • PubMed
    • Google Scholar
72. 1. Sikandar S
    2. Dickenson AH

    (2012) Visceral pain: the ins and outs, the ups and downs

    Current Opinion in Supportive and Palliative Care 6:17.

    https://doi.org/10.1097/SPC.0b013e32834f6ec9

    • PubMed
    • Google Scholar
73. 1. Simmons WK
    2. Avery JA
    3. Barcalow JC
    4. Bodurka J
    5. Drevets WC
    6. Bellgowan P

    (2013a) Keeping the body in mind: insula functional organization and functional connectivity integrate interoceptive, Exteroceptive, and emotional awareness

    Human Brain Mapping 34:2944–2958.

    https://doi.org/10.1002/hbm.22113

    • PubMed
    • Google Scholar
74. 1. Simmons WK
    2. Rapuano KM
    3. Kallman SJ
    4. Ingeholm JE
    5. Miller B
    6. Gotts SJ
    7. Avery JA
    8. Hall KD
    9. Martin A

    (2013b) Category-specific integration of homeostatic signals in caudal but not rostral human insula

    Nature Neuroscience 16:1551–1552.

    https://doi.org/10.1038/nn.3535

    • PubMed
    • Google Scholar
75. 1. Simmons WK
    2. Burrows K
    3. Avery JA
    4. Kerr KL
    5. Bodurka J
    6. Savage CR
    7. Drevets WC

    (2016) Depression-Related increases and decreases in appetite: dissociable patterns of aberrant activity in reward and interoceptive neurocircuitry

    American Journal of Psychiatry 173:418–428.

    https://doi.org/10.1176/appi.ajp.2015.15020162

    • PubMed
    • Google Scholar
76. 1. Smith AR
    2. Zuromski KL
    3. Dodd DR

    (2018a) Eating disorders and suicidality: what we know, what we don't know, and suggestions for future research

    Current Opinion in Psychology 22:63–67.

    https://doi.org/10.1016/j.copsyc.2017.08.023

    • PubMed
    • Google Scholar
77. 1. Smith A
    2. Forrest L
    3. Velkoff E

    (2018b) Out of touch: interoceptive deficits are elevated in suicide attempters with eating disorders

    Eating Disorders 26:52–65.

    https://doi.org/10.1080/10640266.2018.1418243

    • PubMed
    • Google Scholar
78. 1. Smith PN
    2. Cukrowicz KC

    (2010) Capable of suicide: a functional model of the acquired capability component of the Interpersonal-Psychological theory of suicide

    Suicide and Life-Threatening Behavior 40:266–275.

    https://doi.org/10.1521/suli.2010.40.3.266

    • PubMed
    • Google Scholar
79. 1. Soloff PH
    2. Pruitt P
    3. Sharma M
    4. Radwan J
    5. White R
    6. Diwadkar VA

    (2012) Structural brain abnormalities and suicidal behavior in borderline personality disorder

    Journal of Psychiatric Research 46:516–525.

    https://doi.org/10.1016/j.jpsychires.2012.01.003

    • PubMed
    • Google Scholar
80. 1. Stewart JL
    2. Khalsa SS
    3. Kuplicki R
    4. Puhl M
    5. T1000 Investigators
    6. Paulus MP

    (2019) Interoceptive attention in opioid and stimulant use disorder

    Addiction Biology e12831.

    https://doi.org/10.1111/adb.12831

    • PubMed
    • Google Scholar
81. 1. Taylor S
    2. Zvolensky MJ
    3. Cox BJ
    4. Deacon B
    5. Heimberg RG
    6. Ledley DR
    7. Abramowitz JS
    8. Holaway RM
    9. Sandin B
    10. Stewart SH
    11. Coles M
    12. Eng W
    13. Daly ES
    14. Arrindell WA
    15. Bouvard M
    16. Cardenas SJ

    (2007) Robust dimensions of anxiety sensitivity: development and initial validation of the anxiety sensitivity Index-3

    Psychological Assessment 19:176–188.

    https://doi.org/10.1037/1040-3590.19.2.176

    • PubMed
    • Google Scholar
82. Report

    1. United States Department of Commerce Bureau of the Census

    (1920)

    Mortality Statistics 1918

    Nineteenth Annual Report.

    • Google Scholar
83. 1. van Heeringen K
    2. Bijttebier S
    3. Desmyter S
    4. Vervaet M
    5. Baeken C

    (2014) Is there a neuroanatomical basis of the vulnerability to suicidal behavior? A coordinate-based meta-analysis of structural and functional MRI studies

    Frontiers in Human Neuroscience 8:824.

    https://doi.org/10.3389/fnhum.2014.00824

    • PubMed
    • Google Scholar
84. 1. Van Orden KA
    2. Witte TK
    3. Gordon KH
    4. Bender TW
    5. Joiner TE

    (2008) Suicidal desire and the capability for suicide: tests of the interpersonal-psychological theory of suicidal behavior among adults

    Journal of Consulting and Clinical Psychology 76:72–83.

    https://doi.org/10.1037/0022-006X.76.1.72

    • PubMed
    • Google Scholar
85. 1. Van Orden KA
    2. Witte TK
    3. Cukrowicz KC
    4. Braithwaite SR
    5. Selby EA
    6. Joiner TE

    (2010) The interpersonal theory of suicide

    Psychological Review 117:575–600.

    https://doi.org/10.1037/a0018697

    • PubMed
    • Google Scholar
86. 1. Verdejo-Garcia A
    2. Clark L
    3. Dunn BD

    (2012) The role of interoception in addiction: a critical review

    Neuroscience & Biobehavioral Reviews 36:1857–1869.

    https://doi.org/10.1016/j.neubiorev.2012.05.007

    • PubMed
    • Google Scholar
87. 1. Victor TA
    2. Khalsa SS
    3. Simmons WK
    4. Feinstein JS
    5. Savitz J
    6. Aupperle RL
    7. Yeh HW
    8. Bodurka J
    9. Paulus MP

    (2018) Tulsa 1000: a naturalistic study protocol for multilevel assessment and outcome prediction in a large psychiatric sample

    BMJ Open 8:e016620.

    https://doi.org/10.1136/bmjopen-2017-016620

    • PubMed
    • Google Scholar
88. 1. Wickham H

    (2011) ggplot2

    Wiley Interdisciplinary Reviews: Computational Statistics 3:180–185.

    https://doi.org/10.1002/wics.147

    • Google Scholar
89. 1. Wiebking C
    2. de Greck M
    3. Duncan NW
    4. Tempelmann C
    5. Bajbouj M
    6. Northoff G

    (2015) Interoception in insula subregions as a possible state marker for depression-an exploratory fMRI study investigating healthy, depressed and remitted participants

    Frontiers in Behavioral Neuroscience 9:82.

    https://doi.org/10.3389/fnbeh.2015.00082

    • PubMed
    • Google Scholar
90. 1. Willem Van der Does AJ

    (1997) Voluntary breath holding: not a suitable probe of the suffocation alarm in panic disorder

    Behaviour Research and Therapy 35:779–784.

    https://doi.org/10.1016/S0005-7967(97)00031-4

    • PubMed
    • Google Scholar
91. Website

    1. World Health Organization

    (2014) Preventing suicide: a global imperative

    Accessed November 21, 2018.

    https://www.who.int/mental_health/suicide-prevention/world_report_2014/en/
92. Software

    1. Yoshida K
    2. Bohn J
    3. Yoshida MK

    (2019)

    Package ‘tableone’

    R Foundation for Statistical Computing.

Acceptance summary:

The authors examine whether individuals with a history of suicide attempt differ in interoceptive processing, relative to a matched psychiatric reference sample. The study includes a battery of interoceptive tasks (breath-hold, cold-pressor, heartbeat perception, and fMRI). The results indicate that suicide attempters show increased tolerance for aversive respiratory/nociceptive signals, reduced cardiac interoceptive accuracy, and a decreased insula signal during interoception. The findings suggest that blunted awareness of interoceptive sensations and attenuated responses to bodily threat may help to distinguish individuals at risk of suicide. The reviewers and I are enthusiastic about the work.

Decision letter after peer review:

Thank you for submitting your article "Diminished responses to bodily threat and blunted cardiac interoception in suicide attempters" for consideration by eLife. Your article has been reviewed by two peer reviewers, and the evaluation has been overseen by a Reviewing Editor (Alexander Shackman) and a Senior Editor (Christian Büchel). The following individual involved in review of your submission has agreed to reveal their identity: Nils Kroemer (Reviewer #2).

The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare a revised submission.

The two reviewers were enthusiastic about the work:

• I think this is a very strong and important paper. Its key strength is the convergence of multiple methods. This helps address critiques that might be levelled at individual measures. I am impressed by how well the groups are matched in terms of key variables. Some would view differences in psychotropic medication between the groups to be a major flaw, but I think the approach the authors have taken is sensible. Overall, this paper is well guided by theory, methodologically strong, and potentially important clinically.

• This is a very important topic that is notoriously difficult to investigate well so I would like to commend the authors. This is arguably one of the best studies on this topic that I am aware of.

Essential Revisions:

1) Stats.

a) The description of the statistical analysis in the manuscript is currently not sufficient. I was happy to look at the code, but it should not be necessary to know R to understand what exactly the authors did. Therefore, more details on the models should be provided in the Materials and methods or the supporting information.

b) Relatedly, it was not obvious to me why the authors used the lme function for some analyses but not others (although it was even passed to a _gls variable) and sometimes the summary function or the anova function was used, not both.

c) In the manuscript, it was not clear to me which results were reported. Thus, it would be preferable to be a bit more precise in terms of the statistical descriptions and to include the output of the estimated models in the supporting information.

2) Cold pressor challenge.

a) The authors write: "Examination of fixed effects revealed that, while there were no significant group differences in the amount of time elapsed prior to reaching mild, moderate, and peak pain, suicide attempters kept their hands submerged in the cold water for significantly longer (18 seconds on average, SD = 25) than non-attempters after reaching peak pain (t(276)=2.8, p = 0.006, Cohen's d = .34)."

b) First, does the average refer to the group difference in duration and the SD to the pooled SD across groups?

c) Second, if it only refers to this category, why are the degrees of freedom so high that it appears as if a fixed effect across all repeated measurements was taken for the comparison?

d) Third, how was Cohen's d calculated because the study would not be large enough to find significant group differences of this magnitude if the degrees of freedom are calculated correctly (1-β = .36 for d = .34, n1 = 34, n2 = 68). This indicates that a formula was used to estimate d from t, which is only valid for non-nested t-contrasts ("simple" two-sample t-tests).

3) Relations amongst measures.

a) At the moment, DeVille et al. are not fully capitalizing on the strength of the design provided by the complementary measures of interoceptive awareness because the correlations between the interoceptive measures are not reported.

b) The authors report associations between percent signal change in the insula and their behavioral measures of interoceptive awareness, but not between the behavioral measures only.

c) However, the associations between insular signal changes and interoceptive measures are weak and inconsistent. Why would the signal related to the heartbeat detection task show a stronger correlation with the breath-hold duration than with behavioral measures on the same task?

d) The selective report of the significant difference in the manuscript is misleading and should be rephrased. Without a conclusive indication of behavioral loadings on a latent awareness factor, I would be very cautious to interpret such correlations at all. Moreover, the absence of loadings on a shared factor should also be reflected in the Discussion which emphasizes the relevance of interoceptive awareness for the risk of suicide attempts.

Essential Revisions:

1) Stats.

a) The description of the statistical analysis in the manuscript is currently not sufficient. I was happy to look at the code, but it should not be necessary to know R to understand what exactly the authors did. Therefore, more details on the models should be provided in the Materials and methods or the supporting information.

We agree that more information describing each statistical analysis would improve the manuscript. In response to the reviewers’ questions we sought additional consultation with a statistician regarding the best way to clearly present our analyses and results. During this process, we opted to make two additional changes to our analysis approach that should increase the accuracy and methodological rigor of our study. First, we realized that by not specifying “type = marginal” in our ANOVA, R defaulted to a sequential approach, which is not the best approach for our models and can result in inconsistency between ANOVA output and LME fixed effects summary tables. We have made the appropriate correction. All ANOVA findings reported in this revision are calculated using marginal sums of squares. Additionally, we have adopted the Kenward-Roger approximation of degrees of freedom for all tests conducted using the LME. Despite having made this revision to our analysis approach, our findings are consistent with those described in our initial submission, with the minor exception of the heartbeat perception task.

We have rewritten the Materials and methods and Results sections to reflect these changes and to include more details regarding our model specifications, each analysis conducted, and the associated results. We feel that these revisions should substantially increase the clarity for readers regarding our statistical approach.

b) Relatedly, it was not obvious to me why the authors used the lme function for some analyses but not others (although it was even passed to a _gls variable) and sometimes the summary function or the anova function was used, not both.

We had initially made the decision to use a GLS for the breath-hold analyses. Although our breath-hold task consists of two trials, when we planned our analyses we decided the most parsimonious approach would be to take the mean duration of both trials and test for overall group differences, since the two trials were identical and we were not necessarily interested in performance differences from trial to trial. However, after consideration of the reviewers’ comments, we agreed that switching between analysis approaches could cause unnecessary confusion. Additionally, while a between-groups comparison may be a simpler test overall, the examination of each trial is likely more appropriate given the task design. Therefore, we have re-analyzed data from the breath-hold task using a linear mixed effects model. Our updated findings are consistent with the initial results. The linear mixed effects analyses for breath hold duration, O₂ concentration, and CO₂ concentration are reported in Results subsection “Breath-Hold Challenge”, and displayed in Supplementary file 1.

The reviewers note that sometimes the summary function was used without the anova function. The ANOVA function in R provides F tests for each coefficient, while “summary(model)” displays post-hoc tests describing contrasts between all levels of each factor. When only two groups are being compared (i.e., in the absence of potential interactions), the output of the fixed effects model summary and the output of the ANOVA should be identical. Additionally, when the only significant effect is a main effect of a factor with < 2 levels (e.g., a main effect of group, or a main effect of trial where there are 2 trials), the estimates of fixed effects will not provide additional information. Thus in our case it would be redundant to report both the ANOVA F-statistics and the fixed effects t-statistics. In our revised manuscript, we report overall F-statistics in the text, except for cases in which the fixed effects needed to be examined to clarify interactions or main effects involving three or more levels. To provide more detail on the results obtained with the models, we have also added the output of the LME summary tables to the revised manuscript supporting information section (Supplementary Files 1-3).

We have rewritten our Materials and methods and Results section with clearer language describing our analyses. Please note that in the Results section of the updated manuscript we have followed eLife’s guidelines to use an active voice and more engaging style. Additionally, we provide the following statement that describes our overall statistical approach (in addition to more specific details provided throughout the Materials and methods and Results section for each model):

“A marginal analysis of variance (ANOVA) was used on each LME model to examine F-tests for interactions and main effects. In the event of significant interactions, the summaries of LME fixed effects were examined to clarify which factors were driving the effect. The Kenward-Roger approximation of degrees of freedom was used for all LME analyses.”

We have submitted the revised code and statistical output in the form of an R Markdown HTML file.

c) In the manuscript, it was not clear to me which results were reported. Thus, it would be preferable to be a bit more precise in terms of the statistical descriptions and to include the output of the estimated models in the supporting information.

We agreed with the reviewers that the manuscript could be improved by providing more precision in our Materials and methods and Results sections. In making these changes in the revised manuscript, we have rewritten both sections. We have now added a paragraph in the Materials and methods section providing a description of our overall statistical approach (under the subheading “Analysis of Demographic, Behavioral, and Physiological Data”).Additionally, we have revised the language throughout the Materials and methods and Results section for each task to provide more specific details about the statistical modeling approach. To further enhance clarity, we have also added Tables displaying the exact model specification (e.g., Duration ~ Group x Trial + 1|id) and model outputs (i.e., fixed effects estimates with p-values, random effects estimates, and model fit indices). Please see Supplementary files 1-3 in our revised submission for these details.

2) Cold pressor challenge.

a) The authors write: "Examination of fixed effects revealed that, while there were no significant group differences in the amount of time elapsed prior to reaching mild, moderate, and peak pain, suicide attempters kept their hands submerged in the cold water for significantly longer (18 seconds on average, SD = 25) than non-attempters after reaching peak pain (t(276)=2.8, p = 0.006, Cohen's d = .34)."

b) First, does the average refer to the group difference in duration and the SD to the pooled SD across groups?

The average refers to the difference between the mean of attempters and non-attempters, and the SD is the pooled SD across groups. After giving this more consideration, we have decided that it does not make sense to report the SD here. The decision to report the group difference in duration in number of seconds is to provide a metric that is straightforward and accessible to all readers, and reporting the pooled SD here adds unnecessary confusion. Additionally, we report clearer measures of variability elsewhere (i.e., as illustrated by the error bars in Figure 2).

c) Second, if it only refers to this category, why are the degrees of freedom so high that it appears as if a fixed effect across all repeated measurements was taken for the comparison?

After receiving these reviews, we consulted with a statistician regarding our overall statistical approach, which revealed there is a debate surrounding the best approach for calculating degrees of freedom for the fixed effects in linear mixed effects models. In the revised submission, we use the Kenward-Roger approximation. This approach provides a more conservative estimate of statistical significance of fixed effects. We have clarified this in the manuscript in the paragraph that follows the subheading “Analysis of Demographic, Behavioral, and Physiological Data”.

d) Third, how was Cohen's d calculated because the study would not be large enough to find significant group differences of this magnitude if the degrees of freedom are calculated correctly (1-β = .36 for d = .34, n1 = 34, n2 = 68). This indicates that a formula was used to estimate d from t, which is only valid for non-nested t-contrasts ("simple" two-sample t-tests).

Cohen’s d was calculated using the “lme.dscore” function from the R package “EMAtools.” This function does use a formula to estimate Cohen’s d from the t-statistic [i.e., (2*t)/sqrt(df)]. In this revised submission, in lieu of Cohen’s d, we have decided to take a simpler approach and report the standardized β coefficients (i.e., slope differences in units of standard deviation of our dependent variable) in our LME summary tables. Additionally, we report Marginal R2 and Conditional R2 for the overall model. However, if the reviewers or Editors prefer that we use an alternative metric for reporting effect size, we would be happy to do so.

3) Relations amongst measures.

a) At the moment, DeVille et al. are not fully capitalizing on the strength of the design provided by the complementary measures of interoceptive awareness because the correlations between the interoceptive measures are not reported.

b) The authors report associations between percent signal change in the insula and their behavioral measures of interoceptive awareness, but not between the behavioral measures only.

c) However, the associations between insular signal changes and interoceptive measures are weak and inconsistent. Why would the signal related to the heartbeat detection task show a stronger correlation with the breath-hold duration than with behavioral measures on the same task?

d) The selective report of the significant difference in the manuscript is misleading and should be rephrased. Without a conclusive indication of behavioral loadings on a latent awareness factor, I would be very cautious to interpret such correlations at all. Moreover, the absence of loadings on a shared factor should also be reflected in the Discussion which emphasizes the relevance of interoceptive awareness for the risk of suicide attempts.

The reviewers’ concerns related to the selective reporting of the correlations in our initial submission of the manuscript was well-received. We did not intend to mislead our audience, and we agree that the practice of reporting noteworthy correlations in the main manuscript while relegating other correlations to the supplemental materials can be problematic. Therefore, we removed all text that selectively described some correlations but not others from the manuscript. Although by removing discussion of the correlations throughout the manuscript addresses the concerns described by the reviewers in point (d), it does not address the initial concern described in (a)-(c). Thus, we have added a Figure to Appendix 1 that displays the correlations across all measures from the behavioral and neuroimaging tasks, and include a brief comment at the end of the Results section indicating that “Exploratory correlations across behavioral and neuroimaging variables are reported in Appendix 1—figure 1.” Although this approach may not fully capitalize on the strength of the design, given the length of the manuscript, the breadth of the findings we are already reporting, and our lack of specific hypotheses regarding the strength and direction of the correlations across tasks, we believe that any additional focus on the correlations would detract from rather than enhance our manuscript. Additionally, given that our data is a subsample of a larger study, it may be better to analyze the relationships across interoception measures using the larger dataset that includes healthy subjects. We have added the following comment to the “Limitations and Future Directions” section:

“Although not the primary intent of our investigation, we observed several interrelationships within and across levels of analysis raising the possibility of a latent interoceptive awareness trait factor. However, these relationships were inconsistent and were not pre-specified in our hypotheses. Identification of such a latent factor would likely require additional investigation using larger samples and inclusion of individuals not meeting criteria for psychiatric disorders.”

Regarding the comment on the weak relationship between insular signal change and interoceptive measures described in (c), the reviewers raise an excellent question about the unexpected differences between the within-task and across-task correlations. It is possible that we were better able to capture variability (and therefore stronger correlations) in the measure of intensity obtained during the heartbeat detection task due to the fact a) the heartbeat perception task involved modulation of the heartbeat, and b) required subjects to provide ratings on a 0-100 scale instead of 0-6 scale (that is, more variability in the dependent variable). The relationships between these two tasks may also be worth exploring in an analysis that includes the full sample of 500 individuals, rather than a subsample of suicide attempters and matched non-attempters.

[Editors' note: we also include below the reviews that the authors received from another journal prior to submission to eLife, along with the authors’ responses.]

Reviewer #1:

Comments to authors:

The manuscript “Diminished responses to bodily threat and blunted cardiac interoception in suicide Attempters" is an excellent, novel, and well-written contribution that reports physiological and self-report data that is in line with theories suggesting that suicide may root in blunted interoceptive processing or in other words increased tolerance to (impending) physiological damage. This is currently the most comprehensive study on this topic. Being a sub-study of a larger study has allowed recruiting a particularly well described sample and has offered the opportunity for individual matching of patients between suicide-attempters and non-attempters. The findings have important implications for research and clinical practice.

The manuscript is well structured; language and grammar are clear and precise. The Introduction is sound, methods, sample selection as well as sample size appear appropriate and the Discussion is well balanced.

We thank the reviewer for this kind assessment of our study and manuscript.

Major issues:

The only major point I did not really understand was, why the authors have chosen to match 34 suicide-attempters with 68 non-attempters, which is a sample exactly twice the size and creates un-even cell-sizes (which is associated with statistical problems).

We chose to increase the number of individuals in our reference group by using a 1:2 matching ratio in order to increase our confidence in the between-groups comparisons due to the relatively low number of suicide attempters available to us. Although unequal cell sizes can be problematic for certain statistical analyses, the analysis methods we selected are tolerant to the issues associated with unequal cell sizes (e.g., heterogeneous variances between groups).

In the sample description, I missed an overview of diagnoses and comorbidities. In addition, a bit more information on type of and reason for medication would be helpful (maybe in the appendix).

Information about medication usage and diagnostic comorbidities has been added to the Supplement. See Supplemental Table S1 for an overview of the diagnostics between groups and Supplemental Table S2 for an overview of the types of medications participants were taking.

In the main manuscript, a little more information regarding missing data would make it easier for the reader to appreciate what is reported in detail in the appendix. I would suggest shifting the sentence "non-attempter subjects (n=239) were only included for matching to suicide attempters if their data had been manually checked and they had complete observations on the behavioral and psychophysiological variables examined" to the Materials and methods section.

The relevant information has been moved to the main text of the manuscript.

I also missed a-priori power considerations.

We did not conduct or report an a-priori power analyses in the current study, due to the fact that we conducted all analyses retrospectively using a dataset that had previously been collected as part of a longitudinal naturalistic study. Additionally, given the lack of literature on behavioral and neuroimaging assessments of interoception and suicide, expected effect sizes would have been difficult to estimate beforehand. A comment on this has been added to the manuscript (subsection “Participants”) to clarify that our findings are from a retrospective analysis of a pre-existing dataset.

In the Discussion, maybe a brief remark regarding the problem of relatively low reliability of physiological measures might be useful. In particular, it should be made clear that none of the tasks applied in the study could serve for (individual) diagnosis or prognosis.

This is an important point and we have added this comment to the Discussion (“it should be noted that none of the interoceptive tasks applied in the current study have demonstrated sufficient reliability to be considered appropriate for implementation in prognostic assessments of suicidality in clinical settings.”).

Minor issues:

Please add version/build no. of R package 'nlme'.

In the sentence "Suicide attempters held their breath longer (10 seconds on average)" I'd suggest adding SD.

In the figure captions, I would suggest to explain the meaning of the asterisk and report the associated significance level.

We thank the reviewer for pointing these issues out. These changes have been made in the text and in the figure captions.

Reviewer #2:

Comments to authors: The paper by Khalsa and colleagues reports on the findings of an interesting study examining the topic of diminished interoception to orthostatic threats in suicide attempters. With rising numbers of suicide deaths, identifying indices specific to attempters versus non-attempters is highly relevant and timely. The paper is well written and presents strong and interesting findings.

We appreciate the reviewer for their kind assessment of our manuscript.

I only have a few suggestions:

Did the authors assess for current suicidal ideation in their sample? It seems important not only to ensure that there were no group differences but whether there is a difference between non-ideators versus ideators versus past attempters. My understanding is that research on indices between ideators versus attempters is greatly needed to better predict and target the group at greatest risk. What about current or past NSSI, particularly given that NSSI is thought to decrease pain perceptions.

Suicidal ideation was assessed in two ways, using 1) the Patient Health Questionnaire (PHQ-9) during screening, and 2) using the Columbia Suicide Severity Rating Scale (CSSR) upon entry to the Tulsa 1000 longitudinal study. Participants in the study were excluded from the study if they reported active suicidal ideation on the PHQ-9. During the study, administration of the CSSR focused on assessing lifetime suicidal ideation, not current suicidal ideation at that time. As a result, we were unable to examine current and/or recent suicidal ideation in our sample, and we recognize that this is a limitation in the Discussion. We agree that research on indices that distinguish ideators from attempters is greatly needed. To assess this limitation to the best of our ability, we have re-run our analyses after a) covarying for lifetime ideation intensity and b) dividing our non-attempter group on the basis of presence or absence of lifetime ideation. These analyses are reported in the supplement. We acknowledge that this is not a perfect way of addressing this and that further research will be needed to effectively parse out the contribution of ideation to the differences in interoception between ideators and non- ideators.

Regarding the relationship between non-suicidal self-injury (NSSI) and pain perception:

In our sample, 47% (n=16) of suicide attempters reported a history of NSSI, and 50% (n=17) of suicide attempters denied a history of NSSI. We re-analyzed the four primary interoceptive domains examined in the current study with the inclusion of an interaction term between attempt history and NSSI. We found no significant interaction between attempt history and NSSI affecting overall heartbeat perception accuracy (t(90) = 1.6, p = .12), breath hold duration (t(90) = 1.15, p = .26), cold pressor duration (t(85) = .16, p = .87), or signal in the right posterior insula during cardiac attention (t(90) = 1.9, p = .06). We have added this analysis of the role of non- suicidal self-injury and some interpretive comments to the supplement.

It would be helpful to provide a bit more support for the chosen tasks. For example, how is the sensation elicited by breath holding (suffocation) or pain using the CPT thought to relate to suicide behaviors? Why would people who can tolerate shortness of breath better be more willing to take their life? It seems a plausible pathway in NSSI were repeated cutting leads to habituation and this in terms of less pain avoidance, but the importance of the sensory perception is less clear in attempters.

Our decision to examine interoception and suicide attempt history was motivated primarily by existing theories of suicide (Klonsky and May, 2015; Van Orden et al., 2010) and the small body of experimental work on suicide attempt history and interoception (Dodd et al., 2018; Forrest, Smith, White, and Joiner, 2015; Smith, Forrest, and Velkoff, 2018). We selected available tests of interoception from our study database that had a clear relationship to vital organ systems (i.e., cardiac, respiratory) or threats to bodily integrity (i.e., pain). The ability to tolerate aversive respiratory sensations is particularly relevant, as methods involving asphyxiation account for a large and increasing proportion of death by suicide (Yau and Paschall, 2018). Although the cold pressor task may not readily resemble a specific method of suicidal behavior, it is a strong assessor of pain tolerance and has been shown to correlate well with assessments of interoceptive sensitivity (Pollatos, Füstös, and Critchley, 2012) and other modalities of pain processing (Harris and Rollman, 1983).

Regarding the comment that “It seems a plausible pathway in NSSI were repeated cutting leads to habituation and this in terms of less pain avoidance, but the importance of the sensory perception is less clear in attempters,” theoretical work on suicide has previously argued that there is are two plausible pathways between pain perception and suicide attempts, one in which pain tolerance increases over time due to repeated exposure to pain (e.g., NSSI), and one in which the individual, regardless of pain exposure, has a heightened pain threshold and/or pain tolerance (i.e., “dispositional” contributors to suicidal capability as described in the model proposed by Klonsky and May (Klonsky and May, 2015)). While it is true that repeated cutting may lead to pain habituation which, in turn, might increase one’s suicidal capacity (as described in Joiner’s theory of suicide (Van Orden et al., 2010)), a trait-level dispositional degree of pain insensitivity could also predispose individuals to engage in life-threatening and/or physically harmful behaviors (Klonsky and May, 2015). In the current study we do not attempt to dissociate whether the interoceptive differences observed in our suicide attempters is due to a dispositional or acquired factors. Our primary aim was to establish whether evidence for such interoceptive differences exists, which would lay a foundation for future studies to distinguishing the role of dispositional or acquired factors.

Since one cannot rule out that the attempt itself (e.g., via hanging) caused the diminished interoception, additionally comparing non-attempters/non-ideators to non- attempters/ideators could further elucidate interoceptive differences in suicidal versus non-suicidal individuals.

As the reviewer points out, there are two possible interpretations to our findings. The first interpretation is that individuals who possess heightened tolerance for aversive interoceptive sensations are predisposed to engage in suicidal behavior in the presence of suicidal ideation, whereas individuals who lack tolerance for aversive interoceptive sensations will be less likely to engage in suicidal behavior, even in the presence of severe ideation. Another interpretation is that there are no interoceptive differences distinguishing those with the capacity to engage in suicidal behaviors from those who lack this capacity, but that suicide attempts themselves result in diminished interoception. While we believe that the former interpretation makes the most sense given the theoretical literature on suicidal capacity (Klonsky and May, 2015; Van Orden et al., 2010), we agree with the reviewer that we cannot rule out the possibility that the act of having previously engaged in suicidal behavior caused the diminished interoception in our sample. Additionally, it will be interesting to examine whether diminished interoception at baseline predicts the later emergence of suicidal behavior from a longitudinal perspective. Since the findings we are reporting are from an ongoing naturalistic longitudinal study we plan to explore this question in the future.

The reviewer also raises the issue of comparing of attempters, ideating non-attempters, and non-ideators. We agree that the ability to distinguish between these three groups is important. We have conducted additional analyses within our current sample in an attempt to provide clarification on this issue. In these additional analyses, we re-examined our four primary interoception variables after covarying for ideation intensity, and we examined group differences after splitting our sample into three groups: suicide attempters, ideators without a history of attempts, and non-ideators (See supplement). We have also added to the Discussion of this limitation in the main body of the manuscript.

Why was the breath-holding test repeated?

The breath holding test was repeated and averaged in order to increase our confidence in the measure through multiple observations, matching the approach of previous studies utilizing this test (Asmundson and Stein, 1994).

"greater expired O2 following breath hold" – should this say "lesser"?

This typo was corrected.

It is not clear why the authors report the findings on CO2 and O2. Of course, one would expect that breath holding time goes along with higher CO2 and (a small but non-clinical) decrease in O2 – in my mind that is not a result but just a manipulation check.

This is more than a manipulation check. Measurement of CO2 levels provided physiological evidence of suicide attempters’ greater ability to tolerate acute hypercapnic body states. Of course, based on gas exchange laws the O2 and CO2 levels were expected to be reciprocally related, and in that regard, including both could be viewed as a form of manipulation check. Leaving both results allows readers to assess both possibilities.

Reviewer #3:

Comments to authors: This interesting and potential important paper further strengthens previous studies suggesting a link between interoceptive processes and suicidality. Its strengths are twofold, firstly replicating previous findings of an association between self report interoceptive measures and recent (last 5 years) suicide plans, but more importantly substantiating these with more objective experimental measures of interception and their linked brain regions i.e. insula.

We thank the reviewer for this assessment.

I only have minor comments:

1) Many of the interception measures have either an implicit or explicit visceral pain component – it would be nice to be able to dissociate visceral pain (e.g. cold water immersion) from more somatic types of pain (e.g. Von Frey hairs) or alternately somatic sensory sensitivity to try and clarify whether suicidality is linked to impaired pain or somatic perception per se (nociception) or more specifically to interoceptive processes. It would be useful to at least raise this issue in the Discussion. e.g. [1]

[1] Pollatos O, Füstös J, and Critchley HD (2012). On the generalised embodiment of pain: how interoceptive sensitivity modulates cutaneous pain perception. Pain, 153 (8), 1680-6

We agree with the reviewer and have added a comment to the Discussion and is replicated below:

“The current study employed an imprecise, albeit commonly employed, measure of pain perception. It would be advantageous to be able to clarify whether suicidal action is differentially linked to impaired processing of visceral pain signals, which tend to be poorly localized, often referred to somatic structures, and produce strong autonomic and affective responses, as opposed to somatic pain signals, which tend to be discretely localized to somatic structures and produce more variable autonomic and affective responses (Sikandar and Dickenson, 2012). Beyond stimulating visceral and somatic pain processing via different neuroanatomical pathways, it would be helpful to evaluate the degree to which altered pain responding was directly driven by differences in nociception per se as opposed to indirectly modulated by differences in interoceptive processing (Pollatos et al., 2012).”

2) Previous authors have shown/ not shown an association between interception and personality features particularly emotionally unstable PD [2]. Do personality features mediate moderate link between interoception and suicidality in this cohort where personality measures were recorded?

[2] Hart N, McGowan J, Minati L, Critchley HD. Emotional regulation and bodily sensation: interoceptive awareness is intact in borderline personality disorder. J Personal Disord. 2013;27(4):506-18.

[3] Müller LE, Schulz A, Andermann M, Gäbel A, Gescher DM, Spohn A, et al. Cortical representation of afferent bodily signals in borderline personality disorder: neural correlates and relationship to emotional dysregulation. JAMA psychiatry. 2015;72(11):1077-86.

We unfortunately do not have personality measures related to emotional stability/instability assessed in the current study. However, this raises an interesting question, and we have discussed the possibility further in the Supplement.

Reviewer #4:

This study examines several indices of interoceptive awareness in psychiatric patients with histories of suicide attempts (n=34) and a psychiatric comparison group with similar levels of psychopathology (n=68). The topic of the study is timely, as the field knows little about what factors facilitate transition from suicidal thoughts to acts, and interoceptive awareness may be one of them. However, problems with sample size and design I think greatly limit what information can be gleaned from this study's results.

Sample size.

It seems sample size is insufficient for accurate estimation of population parameters, which not only means that real effects can be missed, but that seemingly 'significant' effects can be false positives. I worry this is a substantial rather than small problem when it comes to interpreting results: there are so many different statistical comparisons reported between the two groups, but by my calculation power for even a *single* statistical test is at most around 70%.

The reviewer raises an important point regarding the concern of false positives when conducting analyses within smaller samples. In the current study, we assessed our subjects across multiple distinct, yet related domains (i.e., heartbeat tapping, insula activation during cardiac interoceptive attention, breath holding, cold pressor performance, and self-report indices of interoception). While we did not produce statistically significant results for every analysis conducted (nonsignificant results are reported in the main text of the manuscript and the supplement), we observed a significant difference between groups on cold pressor and breath hold task performance as well as behavioral and neuroimaging assessments of cardiac interoception. We argue that erroneously receiving false positives across multiple measures of interoception is improbable, particularly given that 1) differences between groups were consistent across tasks and in the same direction (i.e., suicide attempters with blunted interoception relative to non-attempters), 2) we observed similar effect sizes across tasks that were in the moderate range (i.e., reducing suspicion for inflated effect sizes associated with small samples), and 3) our findings are supported by pre-existing theoretical work.

It should be noted that our sample size (N=102) is larger than what has been typically used in neuroimaging studies. The median sample size in a published neuroimaging study has been reported as 28 (Poldrack et al., 2017) and 33 subjects (Yeung, 2018), with nearly 95% of studies reporting a sample size of <100 subjects (Yeung, 2018). Additionally, with our data coming from a larger longitudinal study, we will have the opportunity to replicate our findings as more data are collected.

Design.

The potential role of suicide capacity highlighted in the manuscript is to facilitate “transitioning from suicidal ideation to action”. Thus, the group with attempts should have been matched to a group with ideation but no attempts. In contrast, matching on psychopathology is an inappropriate control for this research question, and at best is a rough and unsatisfactory proxy for ideation. Table 1 does report differences between the two groups on a single self-report item assessing suicidal ideation showing a moderate elevation in ideation among the attempter group (d is approximately.3). But the manuscript indicates the Columbia Suicide Severity Rating Scale was administered, and this provides an assessment of suicidal ideation that could be the basis for matching – why would this suicide ideation assessment be ignored when the difference between ideation and attempts is the focus of the manuscript? Matching the groups on ideation is critically important because all attempters have ideation; thus, in a design comparing attempters to non-attempters, a variable that correlates with ideation can masquerade as a correlate of attempts. We only know if a variable correlates with attempts above and beyond its relationship to ideation if ideation is explicitly and carefully controlled for. This design feature is necessary to address the manuscript’s focus on suicide capacity, since the whole point of capacity is that it may contribute to the difference between ideation alone and ideation that progresses to attempts.

We agree that matching on psychopathology alone would be an inappropriate and unsatisfactory proxy for ideation. However, we did not intend to use psychiatric symptom severity as a proxy for ideation. Rather, our goal was to match for illness markers that could independently impact our interoception variables. Anxiety, depression, eating disorders, and substance use disorders have all been previously associated with differences in interoception [e.g., response to cardiac and respiratory perturbation in anxiety disorders (Pohl et al., 1988; Van den Hout et al., 1987); differences in the insula BOLD signal during interoceptive attention in depression (Avery et al., 2014) and eating disorders (Kerr et al., 2016); cold pressor performance differences in individuals with depression (Schwier, Kliem, Boettger, and Bär, 2010) and substance use disorders (Pud, Cohen, Lawental, and Eisenberg, 2006); aberrant insula activation during an inspiratory breathing load task in individuals with a substance use disorder (Stewart et al., 2014); heightened sensitivity to heartbeat sensations (Khalsa et al., 2015) and aberrant insula response to respiratory perturbation in anorexia nervosa (Berner et al., 2018)]. By matching on psychopathology, we increased the likelihood that any differences observed on our measures of interoception were related to suicide attempt history rather than differences in psychopathology between groups. To clarify this in the manuscript, we have added text to the Discussion section (“We demonstrated these differences in a well-characterized sample of participants with similar levels of psychiatric symptoms, increasing the likelihood that the participants’ history of suicidal behavior was the driving force behind the differences observed in interoception, rather than varying features of psychopathology between groups”).

To address concerns about distinguishing ideators and attempters, we report on additional analyses in the Supplement showing that the relationship between interoception and suicide attempts is maintained even after covarying for suicidal ideation. We have also expounded on the limitations that we face by broadly comparing suicide attempters and non- attempters in the Discussion section of our manuscript, and call for additional research that will better enable us to examine the role of interoception in attempters, ideators, and non-ideators. We do believe, however, that the present examination of the role of interoception in suicide represents the strongest starting point for this emerging body of research to date.

I think there’s a mistake in Table 1.% Female is listed as 70% but the n is listed as just 4, which is well under 10%

We thank the reviewer for pointing out this error. The number of females in this cell should have been 47, not 4. Table 1 has been revised accordingly.

References:

Asmundson, G. J., and Stein, M. B. (1994). Triggering the false suffocation alarm in panic disorder patients by using a voluntary breath-holding procedure. The American journal of psychiatry.

Avery, J. A., Drevets, W. C., Moseman, S. E., Bodurka, J., Barcalow, J. C., and Simmons, W. K. (2014). Major depressive disorder is associated with abnormal interoceptive activity and functional connectivity in the insula. Biol Psychiatry, 76(3), 258-266. doi:10.1016/j.biopsych.2013.11.027

Berner, L. A., Simmons, A. N., Wierenga, C. E., Bischoff-Grethe, A., Paulus, M. P., Bailer, U.,. Kaye, W. H. (2018). Altered interoceptive activation before, during, and after aversive breathing load in women remitted from anorexia nervosa. Psychological medicine, 48(1), 142-154.

Dodd, D. R., Smith, A. R., Forrest, L. N., Witte, T. K., Bodell, L., Bartlett, M.,... Goodwin, N. (2018). Interoceptive deficits, nonsuicidal self-injury, and suicide attempts among women with eating disorders. Suicide Life Threat Behav, 48(4), 438-444.

Forrest, L. N., Smith, A. R., White, R. D., and Joiner, T. E. (2015). (Dis)connected: An examination of interoception in individuals with suicidality. J Abnorm Psychol, 124(3), 754-763.

Harris, G., and Rollman, G. B. (1983). The validity of experimental pain measures. Pain, 17(4), 369-376.

Kerr, K. L., Moseman, S. E., Avery, J. A., Bodurka, J., Zucker, N. L., and Simmons, W. K. (2016). Altered Insula Activity during Visceral Interoception in Weight-Restored Patients with Anorexia Nervosa. Neuropsychopharmacology, 41(2), 521-528. doi:10.1038/npp.2015.174

Khalsa, S. S., Craske, M. G., Li, W., Vangala, S., Strober, M., and Feusner, J. D. (2015). Altered interoceptive awareness in anorexia nervosa: effects of meal anticipation, consumption and bodily arousal. International Journal of Eating Disorders, 48(7), 889-897.

Klonsky, E. D., and May, A. M. (2015). The three-step theory (3ST): A new theory of suicide rooted in the “ideation-to-action” framework. International Journal of Cognitive Therapy, 8, 114-129. doi:https://doi.org/10.1521/ijct.2015.8.2.114

Pohl, R., Yeragani, V. K., Balon, R., Rainey, J. M., Lycaki, H., Ortiz, A.,... Weinberg, P. (1988). Isoproterenol-induced panic attacks. Biological psychiatry, 24(8), 891-902.

Poldrack, R. A., Baker, C. I., Durnez, J., Gorgolewski, K. J., Matthews, P. M., Munafò, M. R.,... Yarkoni, T. (2017). Scanning the horizon: towards transparent and reproducible neuroimaging research. Nature Reviews Neuroscience, 18(2), 115.

Pollatos, O., Füstös, J., and Critchley, H. D. (2012). On the generalised embodiment of pain: how interoceptive sensitivity modulates cutaneous pain perception. Pain, 153(8), 1680-1686.

Pud, D., Cohen, D., Lawental, E., and Eisenberg, E. (2006). Opioids and abnormal pain perception: New evidence from a study of chronic opioid addicts and healthy subjects. Drug and alcohol dependence, 82(3), 218-223.

Schwier, C., Kliem, A., Boettger, M. K., and Bär, K.-J. (2010). Increased cold-pain thresholds in major depression. The Journal of Pain, 11(3), 287-290.

Sikandar, S., and Dickenson, A. H. (2012). Visceral pain–the ins and outs, the ups and downs.

Current opinion in supportive and palliative care, 6(1), 17.

Smith, A. R., Forrest, L. N., and Velkoff, E. (2018). Out of touch: Interoceptive deficits are elevated in suicide attempters with eating disorders. Eat Disord, 26(1), 52-65.

Stewart, J. L., May, A. C., Poppa, T., Davenport, P. W., Tapert, S. F., and Paulus, M. P. (2014). You are the danger: attenuated insula response in methamphetamine users during aversive interoceptive decision-making. Drug and alcohol dependence, 142, 110-119.

Van den Hout, M. A., der Molen, V., Margo, G., Griez, E., Lousberg, H., and Nansen, A. (1987). Reduction of CO₂-induced anxiety in patients with panic attacks after repeated CO₂ exposure. The American journal of psychiatry.

Van Orden, K. A., Witte, T. K., Cukrowicz, K. C., Braithwait, S., Selby, E. A., and Joiner, T. E. (2010). The interpersonal theory of suicide. Psychol Rev, 117(2), 575-600. doi:doi:10.1037/a0018697

Yau, R. K., and Paschall, M. J. (2018). Epidemiology of asphyxiation suicides in the United States, 2005–2014. Injury epidemiology, 5(1), 1.

Yeung, A. W. (2018). An updated survey on statistical thresholding and sample size of fMRI studies. Frontiers in human neuroscience, 12, 16.

Author details

1. Danielle C DeVille

   1. Laureate Institute for Brain Research, Tulsa, United States
   2. Department of Psychology, The University of Tulsa, Tulsa, United States

   Contribution

   Conceptualization, Formal analysis, Investigation, Methodology

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0001-8208-2705

2. Rayus Kuplicki

   Laureate Institute for Brain Research, Tulsa, United States

   Contribution

   Formal analysis, Methodology

   Competing interests

   No competing interests declared

3. Jennifer L Stewart

   1. Laureate Institute for Brain Research, Tulsa, United States
   2. Oxley College of Health Sciences, The University of Tulsa, Tulsa, United States

   Contribution

   Writing - review and editing, Conceptualization, Supervision, Funding acquisition, Methodology

   Competing interests

   No competing interests declared

4. Tulsa 1000 Investigators

   Laureate Institute for Brain Research, Tulsa, United States

   Contribution

   Writing - review and editing, Conceptualization, Supervision, Funding acquisition, Methodology

   1. Robin L Aupperle, Laureate Institute for Brain Research, Tulsa, United States
   2. Jerzy Bodurka, Laureate Institute for Brain Research, Tulsa, United States
   3. Yoon-Hee Cha, Laureate Institute for Brain Research, Tulsa, United States
   4. Justin Feinstein, Laureate Institute for Brain Research, Tulsa, United States
   5. Jonathan B Savitz, Laureate Institute for Brain Research, Tulsa, United States
   6. Teresa A Victor, Laureate Institute for Brain Research, Tulsa, United States

5. Martin P Paulus

   1. Laureate Institute for Brain Research, Tulsa, United States
   2. Oxley College of Health Sciences, The University of Tulsa, Tulsa, United States

   Contribution

   Conceptualization, Supervision, Funding acquisition, Methodology

   Competing interests

   No competing interests declared

6. Sahib S Khalsa

   1. Laureate Institute for Brain Research, Tulsa, United States
   2. Oxley College of Health Sciences, The University of Tulsa, Tulsa, United States

   Contribution

   Conceptualization, Supervision, Funding acquisition, Methodology

   For correspondence

   skhalsa@laureateinstitute.org

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0003-2124-8585

• 2,257

  Page views

• 245

  Downloads

• 27

  Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.