Testing the theory of Differential Susceptibility to nightmares: The interaction of Sensory Processing Sensitivity with the relationship of low mental wellbeing to nightmare frequency and nightmare distress

Propensity to have nightmares has been theorised in terms of diathesis–stress models, with this propensity being seen as negative or even pathological. In contrast, a recent model proposes that nightmare propensity is due to Differential Susceptibility to stimuli, where high susceptibility can be beneficial in positive environments but detrimental in negative environments. This susceptibility to stimuli is assessed as the biobehavioural trait Sensory Processing Sensitivity, which refers to a greater responsivity to internal and external stimuli, and an increased depth of cognitive and emotional processing. To test the Differential Susceptibility Framework for nightmares, 137 participants (females = 104, males = 33; mean age = 33.66 years), recruited from a student population and social media sites, were divided into high (n = 39), medium (n = 59) and low (n = 39) Sensory Processing Sensitivity categories based on their score on the Highly Sensitive Person Scale. Low mental wellbeing and the presence of minor psychiatric problems, measured by the General Health Questionnaire, was found to be significantly correlated with nightmare frequency for the high and medium SPS groups (rs = .29 and .28, respectively), but not for the low Sensory Processing Sensitivity group (r = .19). General Health Questionnaire score was also significantly correlated with trait nightmare distress, for the high Sensory Processing Sensitivity group only (r = .32). These findings in favour of the Differential Susceptibility Framework have aetiology and treatment implications for nightmares that differ from diathesis–stress models.


| INTRODUC TI ON
Nightmares are a form of dream disturbance that involve extreme emotional manifestations, typically resulting in awakening.
Traditional models of nightmares usually focus on the diathesisstress framework, in which emotional dysregulation is implicated in the production of nightmares. For example, heightened emotional reactivity measured as neuroticism is associated with nightmare frequency (NMF), and also with nightmare distress (ND), a trait assessing negative reactions to having nightmares (Schredl & Goeritz, 2019). Such findings support the Affect Network Dysfunction Model (Levin & Nielsen, 2007). That model states that nightmares result from the inability to regulate emotions that follow from the cumulative toll of waking life negative emotional events (affect load) and the tendency to experience distress in response to negative emotional stimuli (affect distress), leading to a dysfunction in the neural networks responsible for the regulation of emotions during normal dreaming.
However, a more comprehensive theory that may account for nightmares was proposed by Carr and Nielsen (2017). Their Differential Susceptibility Framework (DSF) proposes that trait Sensory Processing Sensitivity (SPS) is a main contributing factor to nightmare production. SPS refers to a greater responsivity to internal and external stimuli, and an increased depth of cognitive and emotional processing (Aron & Aron, 1997). In general, high SPS individuals are exceptionally sensitive to and affected by their environment; this results in negative outcomes in response to adverse conditions, but beneficial outcomes in response to benign or positive conditions. There are thus many advantages of high SPS, such as the ability to thrive in positive environments (Aron, 2010), and exhibiting greater awareness of the social and physical environment (Greven et al., 2019). In humans and some other species, this behavioural and physiological trait results in different adaptive strategies for individuals regarding response to risk and danger, these intra-species strategies being caution and hesitation versus approach and engagement for high and low SPS individuals, respectively (Aron & Aron, 1997).
High SPS is associated with greater neural response to positive and negative emotional stimuli (Acevedo et al., 2017), which, together with greater perceived stress under negative environmental conditions, are factors highlighted by previous models of nightmare production. Therefore, according to the DSF, highly sensitive persons may demonstrate a particular vulnerability to nightmares, specifically when experiencing prolonged stress.
A similar personality trait, intrapsychic boundariness, has been previously associated with propensity to nightmares (Hartmann et al., 1991); however, the DSF and specifically SPS have not yet been tested empirically.
In order to test the DSF, healthy participants were recruited and tested for NMF, ND (a trait measure of negative reactivity to nightmares; Belicki, 1992) and SPS using the Highly Sensitive Person Scale (HSPS; Aron & Aron, 1997). The sample was split according to Lionetti et al.'s (2018) criteria that divide the population into three categories: high sensitives (approximately 31% highest scorers on the HSPS); medium sensitives (middle 40% scorers); and low sensitives (approximately 29% lowest scorers). The hypothesis was that the correlations between low mental wellbeing and NMF and ND would be largest for the high SPS group, smaller for the medium SPS group, and smallest for the low SPS group.

| Participants
One-hundred and thirty-seven participants (females = 104, males = 33; mean age = 33.66 years, standard deviation [SD] = 16.90) were recruited from social media sites and from the student population at Swansea University. Potential recruits were informed that the study concerned "Sleep, dreaming and lifestyle", so as not to refer to nightmares.

| Measures
Sensory Processing Sensitivity was measured using the HSPS (Aron & Aron, 1997). This has 27 items, responded to on a scale from 1 = Not at all to 7 = Extremely. 3 = about once a month; 2 = less than once a month; 1 = never.
Low mental wellbeing and the presence of minor psychiatric problems was measured by the General Health Questionnaire (GHQ; 12 items, each scored 1-4, scores range from 12 to 48; Goldberg & Williams, 1988). The GHQ assesses mental wellbeing (Jackson, 2007) and "inability to carry out one's normal 'healthy' functions, and the appearance of new phenomena of a distressing nature" (Goldberg & Williams, 1988). Items include: "Have you recently felt constantly under strain?". Trait ND was measured by the Belicki (1992) Nightmare Distress Questionnaire (13 items, each scored 1-5, scores range 13-65).
Items include: "Are you ever afraid to fall asleep for fear of having nightmares?".

| Procedure
Ethical approval for the study was obtained from the Research Ethics Committee of the Department of Psychology, Swansea University.
Participants gave written informed consent online to take part, and then completed questionnaires presented online by Qualtrics (Qualtrics).

| Power calculation
The only prior research that gives an expected r-value for the study is Blagrove et al. (2004), in which GHQ and trait ND correlate with r = .28. A sample size of 39 is needed for this to be significant on a one-tail test with two variables (sex and age) partialled out. From Lionetti et al. (2018), the high and low SPS groups are intended to be 31% and 29%, respectively, of the total sample, with the medium group 40%. We thus aimed for a total sample size of at least 39 × 100/29 = 135, so as to ensure that the intended smallest SPS group was sufficiently powered at n = 39. There is no literature on MADRE NMF correlated with GHQ, and so sample size was determined by the ND calculation.

| Inferential statistics
The variables SPS, NMF, ND, DRF and GHQ were all normally distributed with |kurtosis| and |skewness| < 1 and no outliers. ANOVA was used to compare the high, medium and low SPS groups on GHQ, NMF, ND and DRF. Pearson partial correlations were then used to correlate GHQ with NMF and ND, for the high, medium and low SPS groups separately. Sex and age were partialled out of the correlations as the literature in general shows associations between sex and age and both NMF and ND.

| RE SULTS
For the entire sample (N = 137), mean SPS score =109.92, SD = 22.87, min = 46, max = 160. We aimed to divide the participants into the low, medium and high SPS groups defined by Lionetti et al. (2018) of 29%, 40% and 31% of the sample, respectively. The most appropriate cut-offs for this were 95.0 and 122.5. This resulted in three groups, comprising 28.5%, 43.1% and 28.5% of the sample, with ns = 39, 59 and 39, respectively.
Table 1 also shows that, as hypothesised, GHQ was significantly correlated with NMF for the high and medium SPS groups, but not for the low SPS group. Also, GHQ was significantly correlated with ND for the high SPS group only. Partialling out DRF made negligible difference to the correlations between NMF and GHQ.

| D ISCUSS I ON
The current study demonstrates that high and medium SPS individuals, compared with low SPS individuals, show larger positive correlations between GHQ score and NMF. Furthermore, only the high SPS individuals showed a significant relationship of GHQ score with trait ND. The results accord with the DSF, which has implications for the aetiology of nightmares, and also for the treatment of nightmare sufferers. Aron (2010) describes how there can be benefits to patients when their predispositions can be explained as resulting from a "for better or for worse" trait. She describes high SPS people as being more conscientious, with greater empathy towards others, as well as having a greater awareness of and response to their own emotions. Thus, specifically when considering therapeutic interventions, for Aron (2010) Tukey test significant differences between following superscripts: b p < .05, c p < .01, d p < .001, e p < .05.
TA B L E 1 Mean (and SDs) of NMF, ND, DRF and GHQ for the three SPS groups, and Pearson partial correlations between GHQ and NMF, and between GHQ and ND, for the three SPS groups separately, with age and sex partialled out should be placed upon the positive implications of SPS, along with