1. Neuroscience

Reduced discrimination between signals of danger and safety but not overgeneralization is linked to exposure to childhood adversity in healthy adults

  1. Maren Klingelhöfer-Jens  Is a corresponding author
  2. Katharina Hutterer
  3. Miriam A Schiele
  4. Elisabeth J Leehr
  5. Dirk Schümann
  6. Karoline Rosenkranz
  7. Joscha Böhnlein
  8. Jonathan Repple
  9. Jürgen Deckert
  10. Katharina Domschke
  11. Udo Dannlowski
  12. Ulrike Lueken
  13. Andreas Reif
  14. Marcel Romanos
  15. Peter Zwanzger
  16. Paul Pauli
  17. Matthias Gamer
  18. Tina B Lonsdorf
  1. Institute for Systems Neuroscience, University Medical Center Hamburg-Eppendorf, Germany
  2. Department of Psychiatry, Psychosomatics and Psychotherapy, Center of Mental Health, University Hospital of Würzburg, University of Würzburg, Germany
  3. Department of Psychiatry and Psychotherapy, Medical Center - University of Freiburg, Faculty of Medicine, University of Freiburg, Germany
  4. Institute for Translational Psychiatry, University of Münster, Germany
  5. Department of Psychiatry, Psychosomatic Medicine and Psychotherapy, University Hospital Frankfurt – Goethe University, Germany
  6. Department of Psychology, Humboldt-Universität zu Berlin, Germany
  7. German Center of Mental Health (DZPG), partner site Berlin-Potsdam, Germany
  8. Fraunhofer Institute for Translational Medicine and Pharmacology ITMP, Germany
  9. Department of Child and Adolescent Psychiatry, Psychosomatics and Psychotherapy, Center of Mental Health, University Hospital of Würzburg, University of Würzburg, Germany
  10. Kbo Inn Salzach Hospital Clinical Center for Psychiatry, Germany
  11. Department of Psychiatry, Ludwig-Maximilian-University Munich, Germany
  12. Department of Psychology and Center of Mental Health, Julius-Maximilians-University of Würzburg, Germany
  13. Department of Psychology, Biological Psychology and Cognitive Neuroscience, University of Bielefeld, Germany
https://doi.org/10.7554/eLife.91425.3
Share this article
Cite this article
  1. Maren Klingelhöfer-Jens
  2. Katharina Hutterer
  3. Miriam A Schiele
  4. Elisabeth J Leehr
  5. Dirk Schümann
  6. Karoline Rosenkranz
  7. Joscha Böhnlein
  8. Jonathan Repple
  9. Jürgen Deckert
  10. Katharina Domschke
  11. Udo Dannlowski
  12. Ulrike Lueken
  13. Andreas Reif
  14. Marcel Romanos
  15. Peter Zwanzger
  16. Paul Pauli
  17. Matthias Gamer
  18. Tina B Lonsdorf
(2025)
Reduced discrimination between signals of danger and safety but not overgeneralization is linked to exposure to childhood adversity in healthy adults
eLife 12:RP91425.
https://doi.org/10.7554/eLife.91425.3
  2. Download BibTeX
  3. Download .RIS
14 figures, 11 tables and 1 additional file

Figures

Figure 1
Download asset Open asset
Illustration of conditioned stimulus (CS) discrimination in skin conductance responses (SCRs) during acquisition training (A–B) and generalization phase (C–D) for individuals unexposed (gray) and exposed (pink) to childhood adversity.

Barplots (A and C) with error bars represent means and standard errors of the means (SEMs) including nunexposed = 1199 and nexposed = 203, respectively. The statistical parameters presented in A and C are derived from two-tailed independent-samples t-tests. The a priori significance level was set to α = 0.05. Distributions of the data are illustrated in the raincloud plots (B and D). Points next to the densities represent the CS discrimination of each participant averaged across phases. Boxes of boxplots represent the interquartile range (IQR) crossed by the median as a bold line, ends of whiskers represent the minimum/maximum value in the data within the range of 25th/75th percentiles ± 1.5 IQR. For trial-by-trial SCRs across all phases, see Appendix 1—figure 4. log = log-transformed, rc = range-corrected.

Figure 2
Download asset Open asset
Illustration of skin conductance responses (SCRs) during acquisition training (A–B) and the generalization phase (C–D) for individuals unexposed (gray) and exposed (pink) to childhood adversity separated by stimulus types (CS+: dark shades, CS-: light shades).

Barplots (A and C) with error bars represent means and standard errors of the means (SEMs) including nunexposed = 1199 and nexposed = 203, respectively. The presented statistical parameters are derived from a two-tailed independent-samples t-test (A) and a Yuen independent-samples t-test for trimmed means (C). The a priori significance level was set to α = 0.05. Distributions of the data are illustrated in the raincloud plots (B and D). Points next to the densities represent the SCRs of each participant as a function of stimulus type averaged across phases. Boxes of boxplots represent the interquartile range (IQR) crossed by the median as a bold line, ends of whiskers represent the minimum/maximum value in the data within the range of 25th/75th percentiles ± 1.5 IQR. CS = conditioned stimulus, log = log-transformed, rc = range-corrected.

Figure 3
Download asset Open asset
Illustration of skin conductance responses (SCRs) to the different stimulus types during the generalization phase (i.e. generalization gradients) for individuals unexposed (gray) and exposed (pink) to childhood adversity.

Ribbons represent standard errors of the means (SEMs) including nunexposed = 1199 and nexposed = 203, respectively. CS = conditioned stimulus, GS = generalization stimuli with gradual perceptual similarity to the CS+ and CS-, respectively. log = log-transformed, rc = range-corrected. *p<0.05.

Figure 4
Download asset Open asset
Illustration of general reactivity in skin conductance responses (SCRs) across all experimental phases for individuals unexposed (gray) and exposed (pink) to childhood adversity.

Barplots (A) with error bars represent means and standard errors of the means (SEMs) including nunexposed = 1199 and nexposed = 203, respectively. The statistical parameters presented in A are derived from a two-tailed independent-samples t-test. The a priori significance level was set to α = 0.05. Distributions of the data are illustrated in the raincloud plots (B). Points next to the densities represent the general reactivity of each participant averaged across all phases. Boxes of boxplots represent the interquartile range interquartile range (IQR) crossed by the median as a bold line, ends of whiskers represent the minimum/maximum value in the data within the range of 25th/75th percentiles ± 1.5 IQR.

Figure 5
Download asset Open asset
Means and standard errors of the mean (SEMs) of conditioned stimulus (CS) discrimination in skin conductance responses (SCRs) during acquisition training (A) and the generalization phase (B), Linear deviation score (LDS) (C), and general reactivity in SCRs (D) for the four Childhood Trauma Questionnaire (CTQ) severity groups, respectively.

The dashed line indicates the moderate CTQ cut-off frequently used in the literature and hence also employed in our main analyses: On a descriptive level, CS discrimination in SCRs during acquisition training and generalization test, as well as the strength of generalization (i.e. LDS) and the general reactivity are lower in all groups exposed to childhood adversity at an at least moderate level as compared to those with no or low exposure - which corresponds to the main analyses (see above). log = log-transformed, rc = range-corrected.

Figure 6
Download asset Open asset
Scatterplots with marginal densities illustrating the associations between the number of Childhood Trauma Questionnaire (CTQ) subscales exceeding a moderate or higher cut-off (Häuser et al., 2011) and conditioned stimulus (CS) discrimination in skin conductance responses (SCRs) (A) as well as SCRs to the CS+ (B) and CS- (C) during acquisition training.

log = log-transformed, rc = range-corrected.

Figure 7
Download asset Open asset
Scatterplots with marginal densities illustrating the associations between Childhood Trauma Questionnaire (CTQ) composite scores of abuse (left panel) and neglect (right panel) and conditioned stimulus (CS) discrimination in skin conductance responses (SCRs) (A and B) as well as SCRs to the CS+ (C and D) and CS- (E and F) during acquisition training.

Note that the different ranges of CTQ composite scores result from summing up two and three subscales for the neglect and abuse composite scores, respectively (see also Table 1 for more details). log = log-transformed, rc = range-corrected, R squ.=R squared.

Appendix 1—figure 1
Download asset Open asset
Illustration of zero-order correlations (Pearson’s correlation coefficient) between relevant sample characteristics: different scores of childhood adversity (CTQ sum, CTQ: Neglect, CTQ: Abuse), trait anxiety (STAI-T sum), and depression (ADS-K sum).

All depicted correlations were significant (all p’s<0.001). ADS-K=short version of the Center for Epidemiological Studies-Depression Scale (Hautzinger and Bailer, 1993), STAI-T=State-Trait Anxiety Inventory, Trait (Spielberger, 1983), Abuse and Neglect = composite scores built from childhood trauma questionnaire subscales (CTQ-SF, Bernstein et al., 2003; Wingenfeld et al., 2010).

Appendix 1—figure 2
Download asset Open asset
Illustration of the socioeconomic status information of individuals unexposed (left) and exposed (right) to childhood adversity (according to an at least moderate childhood adversity cut-off) inferred from questions about the school degree (A and B), and the current occupational status (C and D) including the type of employment (E and F).
Appendix 1—figure 3
Download asset Open asset
This illustration of the study design was created as part of the publication ‘Prediction of Changes in Negative Affect During the COVID-19 Pandemic by Experimental Fear Conditioning and Generalization Measures’ by Imholze et al., 2023; Figure 1; https://doi.org/10.1027/2151-2604/a000523, which was distributed as a Hogrefe OpenMind article under the license CC BY 4.0 (https://creativecommons.org/licenses/by/4.0).

Please note that the face stimuli used in the actual experiment are different from those shown here. For copyright reasons, we are unable to present the blonde women and have used an image of a redheaded woman as a substitute.

Appendix 1—figure 4
Download asset Open asset
Trial-by-trial skin conductance response (SCR) data across all experimental phases for the CS+ (red) and CS- (blue) for individuals exposed (dashed lines) and unexposed (solid lines) to childhood adversity separately.

Ribbons represent standard errors of the means (SEMs) including nunexposed = 1199 and nexposed = 203.

Appendix 1—figure 5
Download asset Open asset
Illustration of the distribution of the different numbers (i.e. 1–5) of exceeded subscales among the Childhood Trauma Questionnaire (CTQ) exposure types emotional abuse, physical abuse, emotional neglect, and physical neglect (A-E) and distribution of exposure types across all participants (F).
Appendix 1—figure 6
Download asset Open asset
Illustration of STAI-T (A–B) and ADS-K (C–D) sum scores for individuals unexposed (gray) and exposed (pink) to childhood adversity.

Barplots (A and C) with error bars represent means and standard errors of the means (SEMs) with nunexposed = 1199 and nexposed = 203, respectively. The statistical parameters presented in A and C are derived from Welch's tests. The a priori significance level was set to α = 0.05. Distributions of the data are illustrated in the raincloud plots (B and D). Points next to the densities represent the sum scores of each participant. Boxes of boxplots represent the interquartile range (IQR) crossed by the median as a bold line, ends of whiskers represent the minimum/maximum value in the data within the range of 25th/75th percentiles ± 1.5 IQR. STAI-T=Trait scale of the State-Trait Anxiety Inventory (Spielberger, 1983); ADS-K = Allgemeine Depressionsskala - Kurzform (short version of the Center for Epidemiological Studies-Depression Scale, CES-D; Hautzinger and Bailer, 1993).

Appendix 1—figure 7
Download asset Open asset
Illustration of associations of STAI-T (A, C, E, G) and ADS-K (B, D, F, H) with childhood adversity operationalized as cumulative risk models (A-D) and the specificity model (E-H).

STAI-T=Trait scale of the State-Trait Anxiety Inventory (Spielberger, 1983); ADS-K = Allgemeine Depressionsskala - Kurzform (short version of the Center for Epidemiological Studies-Depression Scale, CES-D; Hautzinger and Bailer, 1993).

Tables

Table 1
Operationalization of childhood adversity in different theoretical approaches and challenges of their statistical translation.
Approach name and referenceOperationalization of childhood adversityChallenges in translating theory into a statistical model
Main analyses
Moderate exposure based on CTQ (exposed vs. unexposed)Short description: At least one subscale met the published cut-off for at least moderate exposure (Bernstein and Fink, 1998; Häuser et al., 2011). The moderate cut-off was chosen, as it was recently identified as the most commonly used in the literature (for a review see Ruge et al., 2024)
Procedure: Dichotomization of the sample into exposed vs. unexposed individuals based on published cut-offs: emotional abuse ≥ 13, physical abuse ≥ 10, sexual abuse ≥ 8, emotional neglect ≥ 15, physical neglect ≥ 10. Such cut-off of moderate exposure was employed in previous work by our team (Koppold et al., 2023) and in the literature (Ruge et al., 2024)
Statistical test: See Materials and methods: Statistical analyses

  • Not based on an existing theory but on what is commonly used in the literature (Ruge et al., 2024)

  • Different cut-offs published (for a discussion, see Ruge et al., 2024)

  • (Statistical) Challenges linked to dichotomization of an inherently continuous variable

Exploratory analyses
Cumulative risk model
(Evans et al., 2013; McEwen, 2003)		Short description: Based on the assumed key role of cumulative exposure (exposure intensity and frequency)
Procedure (a): Classification into four severity groups (no, low, moderate, severe exposure) based on cut-offs published by Bernstein and Fink, 1998
Statistical test (a): Comparison of conditioned responding of the four severity groups by using one-way ANOVAs
Procedure (b): Number of subscales exceeding an at least moderate cut-off based on Bernstein and Fink, 1998 and Häuser et al., 2011
Statistical test (b): Number of sub-scales exceeding an at least moderate cut-off as a predictor and conditioned responding as the criterion in simple linear regression models

  • Problem with CTQ sum score: it assigns the same ‘value’ to all CM types (see also ‘General operationalizational challenges’ below)

  • Number of subscales exceeding cut-off: calculate ANOVA or regression?

  • Cumulative risk scores are based on the implicit assumption that different types of adverse events affect the same mechanisms and are of equal impact

Specificity model
(McMahon et al., 2003; Pollak et al., 2000; Pollak and Tolley-Schell, 2004)		Short description: Consideration of specific exposure types (abuse vs. neglect)
Procedure: Summing up the CTQ subscales of emotional abuse, physical abuse, and sexual abuse yielding a composite score for exposure to ‘abuse’ and summing up the subscales of emotional neglect and physical neglect to yield a composite score for ‘neglect’ (or threat vs. deprivation as done by Sheridan et al., 2017)
Statistical test: The abuse and neglect composite scores are tested for associations with conditioned responding in separate regression models.
In our sample, n=52 and n=96 individuals were exposed to abuse only and neglect only, respectively, while n=55 reported to have experienced both abuse and neglect. We included all participants in all analyses as done previously (Sheridan et al., 2017)

  • What qualifies as a specific exposure type? (i.e. subscales or composite scales for neglect vs. abuse?)

  • Which exposure subcategories are ‘too specific’ or ‘too broad’? (A heterogeneous category may obscure potentially relevant discrete associations)

  • Include only participants who experienced only one specific type but not any other types despite this being rather artificial due to high co-occurrences of different exposure types and requiring extremely large samples? Which cut-off should be used then to define exposure? We decided to include all participants in the analyses as done in previous studies (Sheridan et al., 2017)

  • Lack of specificity of exposure subtypes (e.g. sexual abuse also has an emotional component)

Dimensional model
(McLaughlin et al., 2016; McLaughlin et al., 2021)		Short description: Consideration of specific exposure types (i.e. abuse and neglect) that are assumed to co-occur and be controlled for the effect of one another (as opposed to the specificity model)
Procedure: See specificity model
Statistical test: Abuse and neglect scores are tested for associations with conditioned responding in a single linear regression model in which the influence of the other exposure type is controlled for
General operationalizational challenges

  • Non-comparability of dimensional and categorical approaches: CTQ sum score assumes an equal contribution of all items which contradicts different thresholds for being considered as exposed for different subscales (e.g. lower cut-off for sexual abuse as compared to emotional neglect)

  • Associations in a full sample may differ from associations in the group of exposed individuals only which is a challenge for the interpretation of data

  • Multiple cut-offs published (Bernstein et al., 1997; Bernstein and Fink, 1998)

  • Specific challenges relating to abuse and neglect: They

  • Heterogeneity in the assessment of childhood adversity across studies - both with respect to the assessment tools (e.g. questionnaires, interviews) as well as with respect to the operationalization of adversity (i.e. definition)

  • Different response formats (yes/no vs. specification of duration and frequency) and the number of trauma types/events included in assessment tools impact on prevalence rates and potentially also associations between the number of adverse experiences and symptom severity (e.g. Contractor et al., 2018)

  • Distinction between stressful events and trauma is often unclear (Richter-Levin and Sandi, 2021)

Table 2
Results of t-tests comparing conditioned stimulus (CS) discrimination, the linear deviation score (i.e. strength of generalization), and general reactivity between exposed and unexposed participants.
OutcomePhaseMeasuretdfpCohen’s dLL (95% CI)UL (95% CI)
CS discriminationACQSCR2.331,4000.020–0.18–0.33–0.03
Arousal ratings–1.521,4000.1280.12–0.030.26
Valence ratings0.201,4000.845–0.01–0.160.13
Contigency ratings0.701,4000.484–0.05–0.200.10
GENSCR2.341,4000.020–0.18–0.33–0.03
Arousal ratings–0.281,4000.7770.02–0.130.17
Valence ratings0.061,4000.9530.00–0.150.14
Contigency ratings0.581,4000.560–0.04–0.190.10
LDSGENSCR1.412950.158–0.10–0.250.05
Arousal ratings–0.621,4000.5380.05–0.100.20
Valence ratings0.301,4000.765–0.02–0.170.13
Contigency ratings–0.951,4000.3440.07–0.080.22
General reactivityALLSCR2.061,4000.040–0.16–0.31–0.01
Arousal ratings–0.101,4000.9200.01–0.140.16
Valence ratings0.831,4000.408–0.06–0.210.09
Contigency ratings–0.972500.3340.07–0.090.24

  1. Note. ACQ = acquisition training, GEN = generalization phase, LDS = linear deviation score. Bold numbers indicate significant results (p<0.05).

Table 3
Descriptive information on the subsamples being exposed or unexposed to childhood adversity.
VariableExposedUnexposedStatistics
N203 (14%)1199 (86%)Χ2(1)=707.57, p<0.001
Female/Male124 (61%) / 79 (39%)721 (61%) / 478 (39%)Χ2(1)=0.03, p=0.858
Age (M/SD)26.80 (6.99)25.14 (5.50)t(246.1)=–3.21, p<0.001, d=0.29
STAI-T sum (M/SD)38.73 (9.52)34.04 (7.83)t(250.4)=–6.65, p<0.001, d=0.58
ADS-K sum (M/SD)8.71 (6.31)6.69 (5.70)t(261)=–4.28, p<0.001, d=0.35

  1. Note. STAI-T = State-Trait Anxiety Inventory, Trait scale (Spielberger, 1983), ADS-K = Allgemeine Depressionsskala - Kurzform (short version of the Center for Epidemiological Studies-Depression Scale, CES-D; Hautzinger and Bailer, 1993). Individuals were classified as exposed to childhood adversity if at least one subscale met the published cut-off (Bernstein and Fink, 1998; Häuser et al., 2011) for an at least moderate exposure (i.e. emotional abuse ≥13, physical abuse ≥10, sexual abuse ≥8, emotional neglect ≥15, physical neglect ≥10).

Appendix 1—table 1
Split-half reliability for skin conductance responses (SCRs).
PhaseStimulus typerr lower CIr upper CI
AcquisitionCS+0.8020.7830.820
CS-0.7350.7100.758
CS discr.0.3220.2740.368
GeneralizationCS+0.7200.6930.744
CS-0.4700.4290.510
CS discr.0.3460.2990.391
GS10.6560.6250.684
GS20.5500.5120.585
GS30.5610.5240.596
GS40.2880.2400.336

  1. Note. CS = conditioned stimulus, GS = generalization stimulus, CS discr. = CS discrimination.

Appendix 1—table 2
Resu of the repetition of our main analyses using linear mixed models for skin conductance response (SCR) (A), arousal (B), valence (C), and contingency ratings (D).
A
CS discrimination SCR during ACQCS discrimination SCR during GENLDS SCRGeneral reactivity SCR
PredictorsEstimatesCIpdfEstimatesCIpdfEstimatesCIpdfEstimatesCIpdf
(Intercept)0.18–0.03–0.400.11367.260.03–0.25–0.300.8411347.550.02–0.12–0.160.7711387.920.16–0.10–0.430.23130.64
Age0–0.00 – 0.000.3271388.920–0.00 – 0.000.3341388.970–0.00 – 0.000.6551163.350–0.00 – 0.000.4291386.32
Sex [1]–0.02–0.03 to –0.01<0.0011388.870–0.02–0.010.8181388.990–0.00–0.010.4271074.140.01–0.00–0.020.1761386.33
dummy(School level)10.12–0.18–0.410.44813860.24–0.14–0.630.2213860.09–0.11–0.280.3821386.160.02–0.32–0.360.9131386
dummy(School level)2–0.15–0.37–0.070.1931386.70.03–0.26–0.310.8441386.540.02–0.12–0.170.7531386.6–0.05–0.30–0.200.7061386.03
dummy(School level)3–0.09–0.33–0.140.4311386.54–0.06–0.36–0.240.6991386.420.01–0.14–0.160.9311388.1–0.06–0.33–0.200.6541386.02
dummy(School level)4–0.16–0.37–0.060.1531386.670.04–0.23–0.310.7751386.520–0.14–0.130.9511387.75–0.03–0.27–0.220.831386.03
dummy(School level)5–0.17–0.39–0.050.1321386.360.05–0.23–0.330.7081386.28–0.01–0.16–0.130.8461388.84–0.04–0.29–0.210.7541386.02
dummy(School level)6–0.13–0.35–0.080.2211386.470.02–0.25–0.300.8711386.36–0.01–0.14–0.130.9361388.97–0.04–0.28–0.200.7511386.02
dummy(School level)7–0.14–0.35–0.070.1931386.450.05–0.23–0.320.7451386.34–0.01–0.15–0.130.8981388.96–0.04–0.28–0.200.7541386.02
dummy(School level)8–0.14–0.36–0.070.1811386.380.03–0.24–0.310.8091386.29–0.01–0.14–0.130.9281388.97–0.05–0.29–0.200.7131386.02
dummy(School level)9–0.21–0.51–0.090.1631386.59–0.09–0.47–0.300.6591386.45–0.1–0.30–0.090.291387.6–0.05–0.40–0.290.7541386.03
Childhood adversity [1]–0.02–0.04 to –0.000.0171386.71–0.02–0.04 to –0.000.0241386.55–0.01–0.02–0.000.1781386.58–0.02–0.04 to –0.000.0241386.03
Random effects
σ20.010.0200.02
τ000.00 Site0.00 Site0.00 Site0.01 Site
ICC0.030.040.41
N4 Site4 Site4 Site4 Site
Observations1402140214021402
Marginal R2/Conditional R20.021/0.0490.012/0.0490.006/NA0.004/0.414
B
CS discrimination arousal ratings during ACQCS discrimination arousal ratings during GENLDS arousal ratingsGeneral reactivity arousal ratings
PredictorsEstimatesCIpdfEstimatesCIpdfEstimatesCIpdfEstimatesCIpdf
(Intercept)4.730.46–9.000.031387.123.89–0.92–8.690.1131386.011.37–1.27–4.010.3081388.453.220.85–5.590.0081333.36
Age–0.02–0.04–0.010.191381.610–0.02–0.030.8541384.270.01–0.00–0.030.0631213.66–0.01–0.03 to –0.000.0341388.87
Sex [1]–0.72–0.96 to –0.49<0.0011380.21–0.68–0.94 to –0.42<0.0011383.34–0.1–0.24–0.050.1981151.26–0.22–0.35 to –0.090.0011388.9
dummy(School level)1–2.02–7.98–3.950.5071386.011–5.70–7.700.7691386.01–1.49–5.17–2.190.4281386.121.74–1.56–5.030.3011386
dummy(School level)2–1.93–6.33–2.480.3911387.53–1.08–6.03–3.870.6681387.35–1.2–3.92–1.520.3881388.080.66–1.78–3.090.5961386.48
dummy(School level)3–1.96–6.58–2.670.4071387.23–1.29–6.49–3.910.6271387.07–1.02–3.88–1.840.4841388.80.45–2.11–3.010.731386.37
dummy(School level)4–1.46–5.71–2.790.4991387.45–0.86–5.64–3.910.7231387.28–1.2–3.82–1.430.3721388.61.26–1.09–3.610.2941386.46
dummy(School level)5–0.71–5.06–3.640.7471386.81–0.42–5.31–4.470.8681386.71–1.8–4.49–0.890.1891388.621.57–0.83–3.980.21386.25
dummy(School level)6–0.8–5.06–3.460.7121387.04–0.48–5.26–4.310.8451386.91–1.34–3.97–1.290.3171388.981.16–1.19–3.510.3331386.32
dummy(School level)7–1.27–5.49–2.960.5561387.01–0.47–5.22–4.270.8451386.89–1.19–3.80–1.420.3731388.981.28–1.05–3.620.2811386.3
dummy(School level)8–1.29–5.52–2.930.5491386.85–0.42–5.17–4.330.8621386.75–1.23–3.84–1.380.3561388.81.29–1.04–3.630.2781386.26
dummy(School level)91.99–3.99–7.960.5141387.322.92–3.79–9.640.3931387.16–3.15–6.84–0.550.0951388.591.59–1.72–4.890.3461386.4
Childhood adversity [1]0.26–0.06–0.590.1081387.550.06–0.30–0.420.7521387.370.04–0.16–0.240.671388.050.03–0.15–0.210.7381386.48
Random effects
σ24.625.831.761.41
τ000.05 Site0.08 Site0.00 Site0.06 Site
ICC0.010.0100.04
N4 Site4 Site4 Site4 Site
Observations1402140214021402
Marginal R2/Conditional R20.036/0.0460.021/0.0340.009/0.0100.019/0.061
C
CS discrimination valence ratings during ACQCS discrimination valence ratings during GENLDS valence ratingsGeneral reactivity valence ratings
PredictorsEstimatesCIpdfEstimatesCIpdfEstimatesCIpdfEstimatesCIpdf
(Intercept)3.58–0.74–7.900.1041377.813.4–1.67–8.480.1891388.662.39–0.25–5.020.0761382.585.823.94–7.70<0.0011365.32
Age–0.03–0.06 to –0.010.0051388.31–0.02–0.04–0.010.2661243.670–0.01–0.020.8841387.16–0.01–0.02–0.000.0621388.96
Sex [1]–0.55–0.79 to –0.31<0.0011388.11–0.66–0.94 to –0.38<0.0011195.8–0.07–0.21–0.080.3521386.74–0.01–0.11–0.100.9161388.92
dummy(School level)1–2.03–8.05–3.980.5071386–1.02–8.11–6.080.7791386.1–1.75–5.42–1.930.3511386–0.84–3.46–1.770.5271386
dummy(School level)2–2.53–6.98–1.910.2631386.89–1.19–6.44–4.050.6551388.63–2.08–4.79–0.640.1341387.08–1.14–3.07–0.790.2461386.68
dummy(School level)3–0.48–5.15–4.190.841386.7–1.2–6.71–4.300.6681388.97–1.33–4.18–1.520.3611386.84–1.55–3.58–0.480.1341386.52
dummy(School level)4–0.31–4.59–3.980.8891386.850.19–4.87–5.250.9411388.88–1.91–4.53–0.710.1541387.02–0.85–2.72–1.010.3711386.65
dummy(School level)50.07–4.32–4.460.9761386.460.58–4.60–5.750.8271388.45–2.3–4.99–0.380.0921386.56–0.36–2.27–1.550.7111386.35
dummy(School level)60.07–4.22–4.370.9731386.60.11–4.95–5.180.9651388.89–1.81–4.44–0.810.1761386.72–0.91–2.78–0.960.3381386.45
dummy(School level)7–0.53–4.79–3.730.8071386.57–0.12–5.14–4.910.9641388.89–1.85–4.46–0.750.1631386.7–0.74–2.59–1.110.4331386.43
dummy(School level)8–0.4–4.67–3.860.8531386.480.04–4.99–5.070.9871388.64–1.83–4.43–0.780.171386.59–0.75–2.61–1.100.4261386.36
dummy(School level)92.2–3.83–8.230.4741386.752.68–4.43–9.800.4591388.9–3.61–7.30–0.070.0551386.91–0.41–3.04–2.210.7571386.57
Childhood adversity [1]–0.02–0.35–0.310.9021386.9–0.01–0.40–0.370.951388.6–0.03–0.23–0.170.7541387.09–0.04–0.18–0.100.5811386.69
Random effects
σ24.76.531.760.89
τ000.10 Site0.01 Site0.03 Site0.03 Site
ICC0.0200.020.03
N4 Site4 Site4 Site4 Site
Observations1402140214021402
Marginal R2/Conditional R20.035/0.0550.021/0.0220.006/0.0230.015/0.044
D
CS discrimination contingency ratings during ACQCS discrimination contingency ratings during GENLDS contingency ratingsGeneral reactivity contingency ratings
PredictorsEstimatesCIpdfEstimatesCIpdfEstimatesCIpdfEstimatesCIpdf
(Intercept)5.56–57.39–68.520.8621381.9983.325.26–141.340.0051384.6743.2510.35–76.140.011379.6521.72–3.20–46.650.0881169.04
Age–0.18–0.53–0.160.3031387.390.2–0.12–0.520.2171385.880.03–0.15–0.210.7281388.010–0.15–0.150.9631170.37
Sex [1]–6.28–9.74 to –2.82<0.0011387.01–3.72–6.91 to –0.530.0221385.230.58–1.22–2.390.5261387.75–1.58–3.05 to –0.110.0361159.35
dummy(School level)194.827.05–182.590.03413860.2–80.74–81.140.9961386.010.03–45.82–45.880.99913869.17–25.54–43.880.6041169
dummy(School level)221.75–43.07–86.580.511387.05–45.11–104.90–14.670.1391387.21–30.03–63.89–3.830.0821386.959.97–15.78–35.730.4471170.68
dummy(School level)335.78–32.32–103.880.3031386.82–34.55–97.35–28.250.2811386.96–29.19–64.76–6.380.1081386.7413.87–13.07–40.810.3131170.47
dummy(School level)446.02–16.56–108.600.1491386.99–32.68–90.40–25.030.2671387.15–29.54–62.23–3.150.0761386.920.52–4.28–45.310.1051170.59
dummy(School level)555.51–8.54–119.570.0891386.54–35.9–94.97–23.170.2331386.63–35.79–69.25 to –2.330.0361386.4923.91–1.66–49.490.0671169.58
dummy(School level)649.68–12.99–112.350.121386.7–31.66–89.46–26.130.2831386.82–32.52–65.26–0.220.0521386.6421.1–3.75–45.940.0961170.1
dummy(School level)753.68–8.52–115.880.0911386.68–29.33–86.69–28.030.3161386.79–30.67–63.16–1.820.0641386.6118.22–6.39–42.830.1471170.18
dummy(School level)856.22–6.02–118.460.0771386.57–30.69–88.08–26.710.2941386.67–30.51–63.02–2.000.0661386.5117.44–7.18–42.070.1651169.91
dummy(School level)989.881.91–177.850.0451386.89–57.56–138.69–23.570.1641387.04–35.92–81.87–10.030.1251386.8–0.23–35.05–34.580.9891170.57
Childhood adversity [1]–0.86–5.61–3.890.7231387.06–1.26–5.65–3.120.5721387.231.13–1.35–3.620.371386.960.77–1.27–2.810.4611170.8
Random effects
σ21001851.3273.12156.5
τ0018.22 Site12.82 Site5.58 Site0.88 Site
ICC0.020.010.020.01
N4 Site4 Site4 Site3 Site
Observations1402140214021184
Marginal R2/Conditional R20.029/0.0470.009/0.0240.008/0.0280.020/0.025

  1. Note. Due to its categorical nature, we included school level as a dummy variable.

Appendix 1—table 3
Exploratory results of testing the cumulative risk model involving severity groups using ANOVAs with exposure to abuse as between-subject factor and conditioned stimulus (CS) discrimination, LDS, and general reactivity as dependent variable.
OutcomePhaseMeasureMean 'none'SD 'none'Mean 'low'SD 'low'Mean 'moderate'SD 'moderate'Mean ’severe'SD ’severe'dfNumdfDenFppartial Eta2
 CS discrimination ACQ SCR0.050.110.040.100.020.110.030.1031,3982.350.0710.00
 Arousal ratings2.802.212.762.182.992.073.132.2131,3980.860.4590.00
 Valence ratings2.052.192.212.212.112.212.002.4131,3980.540.6550.00
 Contingency ratings53.0131.8852.0632.0052.3733.1648.0335.0231,3980.510.6730.00
 GEN SCR0.050.140.050.140.010.120.040.1431,3982.360.0700.00
 Arousal ratings3.232.443.022.423.122.353.392.7731,3980.910.4370.00
 Valence ratings2.662.642.732.412.662.532.712.8931,3980.070.9750.00
 Contingency ratings58.2028.9556.1228.6656.9330.4354.5535.0931370.800.4980.13
 LDS GEN SCR0.010.070.010.070.000.060.000.0631,3980.600.6150.00
 Arousal ratings0.521.320.381.320.481.400.631.2531,3981.380.2480.00
 Valence ratings0.561.330.541.330.501.230.591.4931,3980.140.9350.00
 Contingency ratings14.1116.4213.9116.7914.8217.4316.1016.3231,3980.400.7540.00
 General reactivity ALL SCR0.100.160.090.120.070.100.070.1031,3981.640.1780.00
 Arousal ratings4.011.224.041.244.151.093.791.1231,3981.430.2340.00
 Valence ratings4.810.964.770.984.780.834.640.9231,3980.780.5080.00
 Contingency ratings39.4212.2139.0913.7140.5411.6439.5011.3231200.540.6560.15

  1. Note. ACQ = acquisition training, GEN = generalization phase, LDS = linear deviation score, SD = standard deviation. Italic lines indicate the application of robust ANOVAs. In this context, effect sizes do not indicate partial eta squared, but the explanatory measure of effect size (Mair and Wilcox, 2020). Values of 0.10, 0.30, and 0.50 represent small, medium, and large effect sizes, respectively.

Appendix 1—table 4
Exploratory results of testing the cumulative risk model involving the number of subscales exceeding an at least moderate cut-off using regressions with the number of subscales as predictor and conditioned stimulus (CS) discrimination, LDS, and general reactivity as criterion.
OutcomePhaseMeasurebetaSEbLL (95% CI)UL (95% CI)BetatdfpR2Cohen’s f2
CS discriminationACQSCR–0.010.00–0.020.00–0.07–2.621,4000.00900
Arousal ratings0.070.09–0.090.240.020.851,4000.39300
Valence ratings–0.050.09–0.220.12–0.01–0.551,4000.58200
Contingency ratings–2.531.25–4.98–0.07–0.05–2.021,4000.04400
GENSCR–0.010.00–0.020.00–0.03–1.231,4000.21800
Arousal ratings0.000.10–0.190.190.000.001,4000.99700
Valence ratings0.030.10–0.170.220.010.271,4000.78900
Contingency ratings–1.611.14–3.850.63–0.04–1.411,4000.15900
LDSGENSCR0.000.00–0.010.00–0.01–0.481,4000.62900
Arousal ratings0.030.05–0.070.130.010.561,4000.57900
Valence ratings0.000.05–0.100.100.000.041,4000.96500
Contingency ratings0.710.65–0.561.980.031.101,4000.27200
General reactivityALLSCR–0.010.00–0.020.00–0.04–1.911,4000.05700
Arousal ratings–0.030.05–0.120.06–0.02–0.651,4000.51700
Valence ratings–0.040.04–0.110.03–0.03–1.061,4000.29000
Contingency ratings0.590.55–0.481.660.031.081,1820.28100

  1. Note. ACQ = acquisition training, GEN = generalization phase, LDS = linear deviation score. Bold numbers indicate significant results (p<0.05).

Appendix 1—table 5
Exploratory results of testing the specificity model using regressions with exposure to abuse as predictor and conditioned stimulus (CS) discrimination, LDS, and general reactivity as criterion.
OutcomePhaseMeasurebetaSEbLL (95% CI)UL (95% CI)BetatdfpR2Cohen’s f2
CS discriminationACQSCR0.000.000.000.00–0.03–1.261,4000.20900
Arousal ratings0.010.01–0.020.030.020.591,4000.55600
Valence ratings0.000.01–0.020.030.010.271,4000.78900
Contingency ratings–0.250.20–0.640.14–0.03–1.251,4000.21100
GENSCR0.000.000.000.000.031.011,4000.31100
Arousal ratings0.000.01–0.020.040.010.351,4000.72500
Valence ratings0.020.02–0.010.050.031.201,4000.23000
Contingency ratings–0.130.18–0.480.23–0.02–0.711,4000.47900
LDSGENSCR0.000.000.000.00–0.01–0.221,4000.82500
Arousal ratings0.000.01–0.020.010.00–0.081,4000.93300
Valence ratings0.000.01–0.020.01–0.01–0.471,4000.63900
Contingency ratings0.070.10–0.130.270.020.681,4000.49600
General reactivityALLSCR0.000.000.000.000.00–0.021,4000.98400
Arousal ratings0.000.01–0.010.020.020.661,4000.51100
Valence ratings0.000.01–0.010.010.00–0.191,4000.85300
Contingency ratings0.120.09–0.050.290.041.381,1820.16700

  1. Note. ACQ = acquisition training, GEN = generalization phase, LDS = linear deviation score.

Appendix 1—table 6
Exploratory results of testing the specificity model using regressions with exposure to neglect as predictor and conditioned stimulus (CS) discrimination, LDS, and general reactivity as criterion.
OutcomePhaseMeasurebetaSEbLL (95% CI)UL (95% CI)BetatdfpR2Cohen’s f2
CS discriminationACQSCR0.000.000.000.00–0.07–2.531,4000.01200
Arousal ratings0.010.01–0.020.030.010.501,4000.61500
Valence ratings0.000.01–0.020.030.000.101,4000.91900
Contingency ratings–0.410.17–0.75–0.07–0.06–2.361,4000.01800
GENSCR0.000.000.000.00–0.04–1.291,4000.19600
Arousal ratings–0.010.01–0.030.02–0.02–0.591,4000.55800
Valence ratings0.000.01–0.030.02–0.01–0.271,4000.78900
Contingency ratings–0.400.16–0.71–0.09–0.07–2.521,4000.01200
LDSGENSCR0.000.000.000.00–0.02–0.661,4000.51200
Arousal ratings0.000.01–0.020.01–0.01–0.251,4000.79900
Valence ratings0.000.01–0.010.010.000.081,4000.93500
Contingency ratings0.010.09–0.170.190.000.111,4000.91500
General reactivityALLSCR0.000.000.000.00–0.06–2.311,4000.02100
Arousal ratings0.000.01–0.020.01–0.02–0.671,4000.50400
Valence ratings–0.010.00–0.020.00–0.05–1.761,4000.07900
Contingency ratings0.070.07–0.080.220.030.911,1820.36500

  1. Note. ACQ = acquisition training, GEN = generalization phase, LDS = linear deviation score. Bold numbers indicate significant results (p<0.05).

Appendix 1—table 7
Exploratory results of testing the dimensional model using multiple regressions with exposure to both abuse and neglect as predictors and conditioned stimulus (CS) discrimination, LDS, and general reactivity as criterion.
OutcomePhaseMeasurepredictorbetaSEbLL (95% CI)UL (95% CI)Betatdfpadj. R2Cohen’s f2
CS discriminationACQSCRabuse0.000.00–0.010.00–0.09–1.191,3980.23400
neglect0.000.00–0.010.00–0.17–2.331,3980.02000
interaction0.000.000.000.000.191.491,3980.13700
Arousal ratingsabuse0.050.04–0.030.120.091.171,3980.24000
neglect0.040.03–0.030.100.081.091,3980.27500
interaction0.000.000.000.00–0.14–1.131,3980.25800
Valence ratingsabuse0.040.04–0.040.120.081.031,3980.30300
neglect0.030.03–0.040.100.070.881,3980.38200
interaction0.000.000.000.00–0.13–1.021,3980.30900
Contingency ratingsabuse0.810.58–0.331.950.111.391,3980.16500
neglect0.180.49–0.781.140.030.371,3980.71200
interaction–0.030.02–0.080.01–0.18–1.431,3980.15400
GENSCRabuse0.000.000.000.010.050.641,3980.52500
neglect0.000.00–0.010.00–0.11–1.421,3980.15600
interaction0.000.000.000.000.050.381,3980.70500
Arousal ratingsabuse0.030.04–0.060.120.050.631,3980.52700
neglect–0.010.04–0.080.06–0.01–0.201,3980.83800
interaction0.000.000.000.00–0.04–0.281,3980.78200
Valence ratingsabuse0.040.05–0.050.130.060.831,3980.40500
neglect–0.020.04–0.100.06–0.04–0.471,3980.63600
interaction0.000.000.000.00–0.01–0.101,3980.91800
Contingency ratingsabuse0.560.53–0.481.600.081.051,3980.29300
neglect–0.260.44–1.130.62–0.04–0.581,3980.56400
interaction–0.010.02–0.060.03–0.09–0.671,3980.50200
LDSGENSCRabuse0.000.000.000.00–0.03–0.381,3980.70200
neglect0.000.000.000.00–0.06–0.771,3980.44300
interaction0.000.000.000.000.070.531,3980.59600
Arousal ratingsabuse–0.010.02–0.060.04–0.03–0.351,3980.72600
neglect–0.010.02–0.050.03–0.04–0.501,3980.61800
interaction0.000.000.000.000.060.431,3980.66700
Valence ratingsabuse0.000.02–0.050.050.000.021,3980.98200
neglect0.010.02–0.030.050.040.511,3980.61200
interaction0.000.000.000.00–0.04–0.341,3980.73800
Contingency ratingsabuse–0.040.30–0.630.55–0.01–0.141,3980.89100
neglect–0.160.25–0.660.33–0.05–0.641,3980.52300
interaction0.010.01–0.020.030.070.521,3980.60300
General reactivityALLSCRabuse0.000.000.000.010.030.421,3980.67600
neglect0.000.00–0.010.00–0.11–1.421,3980.15600
interaction0.000.000.000.000.040.281,3980.77800
Arousal ratingsabuse0.040.020.000.090.151.941,3980.05300
neglect0.010.02–0.020.050.060.741,3980.45900
interaction0.000.000.000.00–0.19–1.501,3980.13300
Valence ratingsabuse0.010.02–0.020.050.060.731,3980.46800
neglect–0.010.01–0.040.02–0.05–0.691,3980.49300
interaction0.000.000.000.00–0.04–0.271,3980.78500
Contingency ratingsabuse–0.100.26–0.600.40–0.03–0.401,1800.68900
neglect–0.170.22–0.600.25–0.07–0.801,1800.42400
interaction0.010.01–0.010.030.130.941,1800.35000

  1. Note. ACQ = acquisition training, GEN = generalization phase, LDS = linear deviation score. Bold numbers indicate significant results (p<0.05).

Appendix 1—table 8
Pearson correlations of STAI-T and ADS-K with skin conductance response (SCR).
STAI-TADS-K
CS discrimination in SCR (ACQ)r=–0.05, p=0.06r=–0.057, p=0.033
SCR to the CS+ (ACQ)r=–0.057, p=0.032r=–0.057, p=0.032
SCR to the CS- (ACQ)r=–0.019, p=0.467r=–0.013, p=0.638
CS discrimination in SCR (GEN)r=0.001, p=0.964r=–0.032, p=0.234
LDS in SCRr=–0.045, p=0.091r=–0.038, p=0.153
Mean reactivity in SCRr=–0.019, p=0.484r=–0.006, p=0.835

  1. Note. ACQ = acquisition training; GEN = generalization phase; STAI-T = Trait scale of the State-Trait Anxiety Inventory (Spielberger, 1983); ADS-K = Allgemeine Depressionsskala - Kurzform (short version of the Center for Epidemiological Studies-Depression Scale, [CES-D; Hautzinger and Bailer, 1993]); LDS = linear deviation score. Bold numbers indicate significant results (p<0.05).

Additional files

MDAR checklist
https://cdn.elifesciences.org/articles/91425/elife-91425-mdarchecklist1-v1.pdf

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Open citations (links to open the citations from this article in various online reference manager services)

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)

  1. Maren Klingelhöfer-Jens
  2. Katharina Hutterer
  3. Miriam A Schiele
  4. Elisabeth J Leehr
  5. Dirk Schümann
  6. Karoline Rosenkranz
  7. Joscha Böhnlein
  8. Jonathan Repple
  9. Jürgen Deckert
  10. Katharina Domschke
  11. Udo Dannlowski
  12. Ulrike Lueken
  13. Andreas Reif
  14. Marcel Romanos
  15. Peter Zwanzger
  16. Paul Pauli
  17. Matthias Gamer
  18. Tina B Lonsdorf
(2025)
Reduced discrimination between signals of danger and safety but not overgeneralization is linked to exposure to childhood adversity in healthy adults
eLife 12:RP91425.
https://doi.org/10.7554/eLife.91425.3