Behavioral inhibition system sensitivity is no longer associated with psychotic-like experiences after controlling for depression and anxiety symptoms

Abstract

Aim:

Elevated behavioral inhibition system sensitivity has been reported among schizophrenia patients. Yet, no study has investigated the relationship between behavioral inhibition system sensitivity and the occurrence of psychotic-like experiences (subthreshold psychotic symptoms considered to be less severe or impairing), despite evidence that behavioral inhibition system sensitivity is related to other forms of psychopathology known to co-occur with psychotic-like experiences, such as depression and anxiety. Thus, the aim of this study was to assess the relationship between behavioral inhibition system levels and psychotic-like experiences while controlling for depression and anxiety symptoms. We hypothesized that behavioral inhibition system sensitivity would be positively associated with the number of reported psychotic-like experiences, and that this association would be nonsignificant after accounting for depression and anxiety symptoms.

Methods:

Psychotic-like experiences, behavioral inhibition system sensitivity, depression symptoms, and anxiety symptoms were measured in 1162 young adults. Bivariate correlations were calculated and linear regressions performed to measure the relationship between variables of interest.

Results:

Psychotic-like experiences, behavioral inhibition system sensitivity, and symptoms of depression, and anxiety were all significantly and positively correlated with one another. Behavioral inhibition system sensitivity was no longer related to the number of psychotic-like experiences reported after controlling for anxiety symptoms, with and without controlling for depression symptoms.

Discussion:

These findings suggest that the hypersensitivity to threat observed among individuals reporting higher levels of psychotic-like experiences is likely related to co-occurring depression and anxiety symptoms. Thus, behavioral inhibition system sensitivity may be more reflective of a transdiagnostic phenotype of general psychopathology than specifically related to psychosis.

Introduction

Psychosis occurs on a continuum, with the symptoms that define formal psychotic disorders (e.g., schizophrenia) on one extreme and a non-clinical phenotype characterized by psychotic-like experiences (PLEs) on the other (Van Os et al., 2009). PLEs, subthreshold psychotic symptoms considered to be less severe or impairing share biological and experiential risk factors with clinical psychoses (Kelleher & Cannon, 2011) in addition to having common clinical correlates. For example, individuals with psychotic disorders are likely to have co-morbid mood and anxiety disorders (Buckley et al., 2008), and so do individuals experiencing PLEs (Unterrassner et al., 2017). Further, evidence suggests that PLEs confer risk for psychotic disorder onset, such as schizophrenia (Van Os et al., 2009). Thus, investigations that seek to further understand PLEs and their correlates may lend insight into clinical psychoses.

In order to further understand motivational, affective, and personality disturbances across the psychosis spectrum, recent work has investigated behavioral approach (i.e., behavior motivated by appetitive rewards) and avoidance (i.e., behavior motivated by the perception of threats) among individuals with varying degrees of psychotic experiences. Recent investigations have adopted Gray’s (1970, 1982) Reinforcement Sensitivity Theory (RST) as a framework with which to study approach and avoidance in schizophrenia patients (Barch et al., 2008; Horan et al. 2006, Scholten et al., 2011; Strauss et al., 2011) and individuals with subclinical symptomatology (Engel et al., 2013, Schlosser et al., 2014). Gray’s theory describes a behavioral inhibition system (BIS), which responds to conditioned aversive stimuli by fostering inhibitory behavior while increasing arousal and attention (Gray, 1970, 1982, 1994). The behavioral activation system (BAS), on the other hand, is sensitive to rewarding stimuli and fosters goal-directed behavior (Gray, 1970, 1982, 1994). The third system, the Fight / Flight / Freeze System (FFFS), proposed by Gray (1987) is responsible for initiating fight-or-flight responses to unconditioned, innate fearful stimuli. The role of the BIS was altered in Gray and McNaughton’s (2000) revision of the RST, where the BIS is responsible for goal conflict resolution, stimulating the BAS or FFFS depending on the perception of the reward and/or threat associated with a stimulus (Gray & McNaughton, 2000). Regardless, extreme levels of BIS and/or BAS sensitivity confer vulnerability for various forms of psychopathology, including unipolar depression, bipolar disorder, and anxiety disorders (Bijttebier et al., 2009).

It is well-demonstrated that individuals with schizophrenia have elevated BIS sensitivity, but no differences in overall BAS sensitivity, when compared to non-psychiatric controls (Barch et al., 2008; Horan et al. 2006, Scholten et al., 2011; Strauss et al., 2011). Such findings may imply that schizophrenia is hallmarked by a hypersensitivity to conditioned threatening or aversive stimuli (Scholten et al., 2006), although Reddy and colleagues (2014) describe the existence of several phenotypes of BIS activity among individuals with schizophrenia, including individuals with both extremely low and extremely high BIS sensitivities. Only two studies to our knowledge have reported examinations of BIS/BAS sensitivities among individuals reporting subthreshold psychotic symptomatology. Schlosser et al. (2014) reported elevated BIS sensitivity among individuals at clinical-high risk (CHR) for psychosis when compared to non-psychiatric controls. Engel et al. (2013) investigated BIS/BAS sensitivities and subthreshold negative symptoms in a non-clinical sample and discovered that negative symptoms were positively correlated with BIS sensitivity and negatively correlated with BAS sensitivity; however the association with BIS sensitivity was nonsignificant after controlling for depressive symptoms. Collectively, these studies point toward BIS hypersensitivity as an observable characteristic of individuals experiencing a range of psychotic symptomatology.

However, no study to our knowledge has investigated the relationship between positive PLEs and BIS/BAS sensitivity. It is worthwhile to determine if PLEs and BIS sensitivity are related, given the previous evidence that BIS hypersensitivity is observable among schizophrenia patients (Scholten et al., 2006) and CHR individuals (Schlosser et al. 2014), and that other contributors to motivation impairment have been observed among individuals reporting PLEs (Cooper et al., 2018). However, if such a relationship does exist, then the extent to which it would be explained by mood and anxiety symptoms is similarly unclear given that BIS hypersensitivity is related to both anxiety and depressive symptoms (Bijttebier et al., 2009). Thus, BIS hypersensitivity among people who report PLEs could very well be related to elevated levels of other psychopathology (e.g., anxiety and depression).

The aim of this study was to investigate the relationship between BIS sensitivity and PLEs in a non-clinical sample of young adults, accounting for the effects of depression and anxiety symptoms. We hypothesized that (1) BIS sensitivity would be positively correlated with PLEs, depression symptoms, and anxiety symptoms in our sample, and that (2) the relationship between BIS sensitivity and PLEs would be nonsignificant after accounting for the effects of depression and anxiety symptoms. In examining the relationship between BIS sensitivity and PLEs, we hope to provide more information about the non-clinical phenotype of psychosis, and therefore about the onset of a debilitating group of disorders.

Methods

Subjects

Participants (N=1162) were recruited via the online subject recruitment pool at a large, urban university. Participants had to be at least 18 years old and able to read written English. The sample was comprised of a diverse group of undergraduates who were largely representative of the university (see Table 1 for sample characteristics). Participants were recruited from courses with research participation requirements. Participation was voluntary; participants have other studies to choose from and may complete alternate assignments to fulfill requirements. Participants accessed a description of the protocol on the university’s recruitment website. Study measures were administered using a computer in the laboratory space, with lab staff present to provide instruction or respond to questions. Written consent was obtained at the time of participation.

Table 1.

Demographic characteristics and descriptive statistics of the sample

	Overall sample (n = 1162)
Male, n (%)	270 (23.2)
Age (years), mean (SD) [range]	20.09 (2.34) [18 – 34]
Race, n (%)
American Indian or Alaska Native	6 (0.5)
African American	224 (19.3)
Asian	190 (16.4)
Native Hawaiian or Other Pacific Islander	6 (0.5)
White	682 (58.7)
More than one race	54 (4.6)
Total number PLEs reported, mean (SD) [range]	9.75 (7.60) [0 – 44]
CES-D score, mean (SD) [range]	8.11 (5.43) [0 – 27]
STAI score, mean (SD) [range]	12.82 (4.95) [7–28]
BIS score, mean (SD) [range]	20.99 (3.93) [7 −28]

PLEs are the sum of reported psychotic-like experiences, CES-D score is the sum of item-level scores on the Center for Epidemiological Studies Depression Scale, STAI is the sum of the item-level scores on the State-Trait Anxiety Inventory, and BIS is the sum of item-level scores on the BIS subscale of the Behavioral Inhibition and Activation System scales.

To avoid biasing our sample, our protocol description did not mention PLEs. Participants were told that they would complete questionnaires that assessed personal and cultural factors, as well as mental health experiences and symptoms.

Measures

The Prodromal Questionnaire.

The Prodromal Questionnaire (PQ; Loewy et al., 2005) was used to assess PLEs. The PQ is a 92-item self-report measure that instructs individuals to report the whether a PLE has occurred in the past month (Loewy et al., 2005). Individuals are asked to not include experiences that occurred while under the influence of alcohol, drugs, or other medications. The positive subscale of the PQ was used (45 items) and has been validated against interviews that assess for psychosis-risk, including the Structured Interview for Psychosis-Risk Syndromes and is psychometrically-sound, with a Cronbach’s alpha of 0.92 (Loewy et al., 2005; McGlashan et al., 2001). Individuals who endorsed an item were scored as having experienced that phenomenon and the total number of positive PLEs was used.

The BIS/BAS Scales.

The Behavioral Inhibition and Activation (BIS/BAS) Scales were used in order to capture individual differences in BIS sensitivity (Carver & White, 1994). The BIS/BAS scales are a 24-item self-report measure that instructs participants to determine how true a given statement is for them. The psychometric properties of the scales have been confirmed by studies using both nonclinical (Jorm et al., 1998) and clinical samples (Campbell-Sills et al., 2004); the BIS subscale utilized for this study has a Cronbach’s alpha of 0.74 (Carver & White, 1994). The scales have been demonstrated to represent a valid measure of BIS sensitivity, even though the measure was developed before the RST’s revision (Berkman, Lieberman, & Gable, 2009). The sum of the 7 BIS items on the scale was used. All but two of the items on the BIS/BAS Scales are reverse-scored; thus, a higher BIS score reflects higher BIS sensitivity.

The Center for Epidemiologic Studies Depression Scale.

Depression symptoms were measured using a brief version of the Center for Epidemiologic Studies Depression Scale (CES-D; Kohout et al., 1993; Sawyer-Radloff, 1977). The 10 items used in this version of the CES-D were identified as being highly predictive of scores on the full-length CES-D, with Cronbach’s alpha of 0.73 (Kohout et al., 1993) and thus an acceptable measure of depressive symptomatology. Individuals are instructed to report how frequently they experienced a given symptom in the past week. The total depression score was used.

The State-Trait Anxiety Inventory – Trait Version.

The State-Trait Anxiety Inventory, Trait Version (STAI; Spielberger et al., 1983) was used to measure trait anxiety symptoms. The current study utilized a 7-item version of the STAI that was comprised of items identified as representing a form of “pure” anxiety by a factor analysis; it is psychometrically-sound, with a Cronbach’s alpha of 0.78 (Beiling et al., 1998). Participants are instructed to indicate how often they experience a trait anxiety symptom. The total anxiety score was used.

Statistical analyses

Analyses were performed using Statistical Package for Social Sciences (IBM Version 25). First, the normality of the dependent variable (PLEs) was assessed visually and statistically by calculating and examining skewness and kurtosis values; no deviations from normality were observed (skewness and kurtosis values can be found in Table 2).

Table 2.

Skewness and Kurtosis Values for Study Variables

	Skewness	Kurtosis
PLEs	1.08	1.11
BIS	−0.41	−0.04
CES-D	0.81	−0.06
STAI	0.81	−0.19

PLEs are the sum of reported psychotic-like experiences, CES-D score is the sum of item-level scores on the Center for Epidemiological Studies Depression Scale, STAI is the sum of the item-level scores on the State-Trait Anxiety Inventory, and BIS is the sum of item-level scores on the BIS subscale of the Behavioral Inhibition and Activation System scales.

The variables of interest (PLEs, BIS, CES-D, and STAI scores) were examined for associations with potential control variables. Gender differences were checked for study variables using point-biserial correlations given known gender differences in the rates of these symptoms. Relationships between the main study variables and participant age also were checked using bivariate Pearson correlations.

Bivariate correlations were calculated in order to test our hypothesis regarding the relationship between BIS sensitivity and PLEs, CES-D, and STAI scores. Following this, multiple linear regressions were conducted to determine the associations between BIS scores and PLEs (with PLEs as the dependent variable) with and without controlling for mood and anxiety symptoms with the following in separate models as independent variables (1) BIS scores, (2) BIS and CES-D scores, (3) BIS and STAI scores, and (4) BIS, CES-D, and STAI scores. Participants who had complete data for all measures of interest were included in analyses. Because all participants completed questionnaires in the lab, there were virtually no missing data.

Parametric statistical assumptions (e.g., normality, homogeneity of variance, linearity) were checked prior to performing t-tests and correlations. The assumptions of multiple regression (e.g., no multicollinearity) were checked at the time that they were performed. Significance was based on two-tailed p-values less than 0.05.

Results

Sample characteristics and bivariate correlations

Demographic characteristics and descriptive statistics of the sample can be found in Table 1. Results from point-biserial correlations between study variables and gender indicated that females had higher BIS and CES-D scores than males (see Table 3). No other significant gender differences were found. Given that males and females did not differ in the number of PLEs reported, we did not covary for gender in the following regression analyses. Pearson correlations between study variables and participant age are described in Table 3. There was no relationship between participant age and depression scores, but significant relationships between age and the other main study variables. Thus, to take a more conservative approach, we covaried for age in the following regression analyses.

Table 3.

Pearson correlations coefficients for gender, age, PLEs, depressive symptoms, anxiety symptoms, and BIS sensitivity

	1	2	3	4	5	6
1. Age	-
2. PLEs	−0.08**	-
3. Depressive symptoms	−0.03	0.54***	-
4. Anxiety symptoms	−0.08*	0.59***	0.73***	-
5. BIS	−0.11***	0.28***	0.40***	0.45***	-
6. Gender	−0.13***	−0.05	0.07*	0.03	0.19***	-

PLEs are the sum of reported psychotic-like experiences, CES-D score is the sum of item-level scores on the Center for Epidemiological Studies Depression Scale, STAI is the sum of the item-level scores on the State-Trait Anxiety Inventory, and BIS is the sum of item-level scores on the BIS subscale of the Behavioral Inhibition and Activation System scales.

^*p < .05.

^**p < .01.

^***p < .001.

Correlation results can be found in Table 3. Significant positive correlations were found between PLEs and BIS scores, CES-D scores, and STAI scores. In addition, BIS scores were positively correlated with CES-D scores and STAI scores.

Regression results

The results of the multiple regression analyses are presented in Table 4. In multiple regression model one, increases in BIS scores were significantly associated with increases in PLEs (β = .27, p < .001) while controlling for age. In multiple regression model two, the association between PLEs and BIS scores (β = .07, p = .016) remained significant when CES-D scores were included in the model. In model 3, BIS scores were not significantly associated with PLEs (β = .01, p = .607) after controlling for STAI scores. BIS scores were similarly not significantly associated with PLEs (β = −0.01, p = .731) in model four, after controlling for both CES-D scores and STAI scores.

Table 4.

Summary of multiple regression analyses predicting PLEs

	B	95% Confidence Intervals for B	SE	β	p	R	R2
Model 1						0.28	0.08
Intercept	1.15	[−3.35, 5.65]	2.29		0.616
Age	−0.12	[−0.30, 0.06]	0.09		0.199
BIS	0.52	[0.41, 0.63]	0.06	0.27	0.000
Model 2						0.54	0.30
Intercept	4.58	[0.65, 8.51]	2.00		0.023
Age	−0.17	[−0.32, −0.01]	0.08	−0.05	0.040
BIS	0.13	[0.03, 0.23]	0.05	0.02	0.016
CES-D	0.71	[0.64, 0.79]	0.04	0.51	0.000
Model 3						0.59	0.34
Intercept	−0.50	[−4.31, 3.31]	1.94		0.798
Age	−0.08	[−0.24, 0.70]	0.08	−0.03	0.289
BIS	0.03	[−0.08, 0.13]	0.05	0.02	0.607
STAI	0.89	[0.81, 0.97]	0.04	0.58	0.000
Model 4						0.61	0.37
Intercept	1.70	[−2.07, 5.47]	1.92		0.375
Age	−0.12	[−0.27, 0.03]	0.08	−0.04	0.127
BIS	−0.02	[−0.12, 0.08]	0.05	−0.01	0.731
CES-D	0.33	[0.24, 0.43]	0.05	0.24	0.000
STAI	0.63	[0.53, 0.74]	0.05	0.41	0.000

PLEs are the sum of reported psychotic-like experiences, CES-D score is the sum of item-level scores on the Center for Epidemiological Studies Depression Scale, STAI is the sum of the item-level scores on the State-Trait Anxiety Inventory, and BIS is the sum of item-level scores on the BIS subscale of the Behavioral Inhibition and Activation System scales.

In each of the models above, PLEs was the outcome variable and age was entered as a covariate. In model 1, the relationship between BIS and PLEs was measured. Models 2 and 3 measured the relationship between PLEs and BIS while controlling for CES-D and STAI scores respectively. Model 4 measured the relationship between PLEs and BIS while controlling for both STAI and CES-D scores. R² refers to the total variance explained by each model.

Discussion

Our study is the first to report that BIS sensitivity and PLEs are positively correlated in a non-clinical sample. Such a result extends previous findings among samples of individuals with schizophrenia (Barch et al., 2008; Horan et al. 2006, Scholten et al., 2006; Strauss et al., 2011) and among individuals at CHR for psychosis (Schlosser et al., 2014) to a non-clinical sample of individuals reporting PLEs. Nevertheless, our findings indicated that the association between BIS sensitivity and PLEs is largely explained by anxiety symptoms, suggesting that elevated BIS sensitivity in those experiencing PLEs likely reflects a heightened tendency to avoid conditioned stimuli that are perceived as being aversive or threatening comparable to what occurs in anxiety disorders (Zinbarg & Yoon, 2008).

Our study found that the relationship between BIS and PLEs is largely explained by anxiety symptoms. Depression symptoms could not account for all of the relationship between BIS levels and PLEs in our sample, as evidenced by the fact that BIS remained a significant predictor of PLEs in the regression model where only CES-D scores were included. Yet, the predictive value of BIS in this model was small, thus implying a substantial attenuation of the relationship between BIS and PLEs after controlling for depression symptoms. Such a finding indicates that much of the relationship between BIS and PLEs is likely explained by the depressive symptomatology known to be associated with the occurrence of PLEs (Jeppesen et al., 2015) and with increased BIS activity (Campbell-Sills et al., 2004).

Importantly, our study found that BIS was no longer related to PLEs after controlling for trait anxiety symptoms, both with and without controlling for depression symptoms. Such a result is perhaps unsurprising, given the heightened anxiety present among individuals who report PLEs (Nishida et al., 2008) and given how strongly BIS activity is related to anxiety symptoms (Gray, 1982). Scholten et al. (2006) conclude that heightened BIS levels among schizophrenia patients may represent a proneness to anxiety, which the authors argue may have etiological implications. Namely, a physiologically-based vulnerability to experiencing anxiety (BIS hypersensitivity) may play a role in the development of psychotic disorders. Such a mechanism is easy to conceptualize, given that Gray (1982) described BIS activation as related to anxiety, and that increased anxiety has been reported among samples of individuals reporting PLEs (Nishida et al., 2008). Schlosser and colleagues (2014) also report associations between levels of anxiety and BIS sensitivity among individuals with varying degrees of psychotic illness, including CHR individuals, recent-onset and chronic schizophrenia patients, suggesting that anxiety symptoms contribute to the hypersensitivity to threatening stimuli throughout the course of psychotic disorders. Thus, the heightened hypersensitivity to threat observed among individuals reporting PLEs may be contributed to, and indeed explained by, anxiety symptoms.

Additionally, BIS sensitivity may operate transdiagnostically, simultaneously conferring vulnerability for multiple psychopathologies, manifesting as a result of these psychopathologies, or both. This is plausible given the BIS’s role in motivating individuals away from aversive stimuli (Gray, 1970, 1982, 1994), and BIS-related motivational impairments in psychosis (Schlosser et al., 2014), depression, and anxiety disorders (Struijs et al., 2018). That is, BIS hypersensitivity may provide a shared mechanism through which individuals develop symptoms of depression and anxiety, and PLEs.

This study was limited by its use of cross-sectional data. Such data limits our ability to comment definitively on the temporal and causal characteristics of the relationships between BIS, PLEs, depression symptoms, and anxiety symptoms. The use of such data with our nonclinical sample also renders it difficult to definitively discuss the relevance of these findings to psychotic disorders. It is also possible that our sample contains an overrepresentation of individuals with low levels of PLEs, which would limit our ability to generalize these findings to samples of individuals with more clinically-significant PLEs. However, these findings are reasonably generalizable given the wide range of PLEs reported in this sample.

We cannot rule out the possibility that bias associated with online data collection could have influenced our findings (e.g., those with subclinical paranoid beliefs may be less inclined to respond to an online invitation). However, the rates of PLEs were relatively high in our sample and a large number of our participants endorsed at least one PLE, indicating that we likely had a large enough distribution of PLEs to examine our primary study aims. Even though, our sample was comprised of undergraduates who may not be generalizable to other populations. Future studies should attempt to replicate these findings using alternative recruitment strategies.

Unmeasured clinical variables, such as psychiatric and other medical conditions, are a possible confounder. Of note is the effect that individuals with psychotic disorders such as schizophrenia would have on these results: given the findings of elevated BIS sensitivity in individuals with schizophrenia (Barch et al., 2008; Horan et al. 2006, Scholten et al., 2011; Strauss et al., 2011). While we do not have information on the presence of such conditions, the effect of such conditions should be mitigated by our large, nonclinical sample, and the low base-rate of these conditions in the population.

The generalizability of these findings may be hindered by the gender distribution of our sample, in which females outnumbered males. Nevertheless, there were still a large number of males in this sample (n=270), and no gender differences in PLEs were found, reducing the possibility the gender influenced study results. Further, our study is limited by its measurement of BIS sensitivity with the BIS/BAS scales, which were created before the revision of the RST, though they still likely represent a reliable BIS sensitivity (Berkman, Lieberman, & Gable, 2009). The strengths of this study include its large, racially diverse sample and its use of a non-clinical, non-help-seeking population.

Our sample size and demographic characteristics make our findings reasonably generalizable to the population of young adults who report positive PLEs, though our use of a university sample may limit generalizability given that members of our sample are likely higher functioning on average. Similarly, our use of a non-clinical sample is a strength in that it amplifies our ability to generalize these findings to the population at large, where it has become apparent that PLEs occur at a higher prevalence than that of clinical psychoses (Van Os et al., 2009).

Given that all of the studies that have investigated the link between BIS activity and psychosis have adopted cross-sectional designs, future studies would benefit from using longitudinal designs to examine the temporal nature of the relationships between BIS, PLEs, as well as depression and anxiety symptoms. It would also be worthwhile to utilize neuroimaging techniques in these longitudinal designs to determine if biological changes in the brain areas known to be responsible for BIS activity, such as septohippocampal structures (Depue & Iacono, 1989; Gray, 1982, 1994), precipitate PLEs and other psychopathological experiences, or emerge as a result of these experiences.

The results of our study suggest that individuals who report more PLEs are characterized by a hypersensitivity to conditioned threatening or aversive stimuli, a hypersensitivity that appears to be explained by anxiety symptoms and potentially depression symptoms. Such a finding is novel and may indicate that BIS hypersensitivity is a specific contributor to motivation impairment that is observable before the commencement of clinically significant psychotic symptoms. Additionally, the inclusion of anxiety symptoms and symptoms of depression in our models led to a nonsignificant relationship between BIS and PLEs, indicating that the BIS-PLE relationship may be better accounted for by these symptoms of other psychopathologies. As such, BIS hypersensitivity may represent a transdiagnostic phenotype unique to PLEs, rather than psychotic disorders overall.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.