ORIGINAL

ISSN 0120-0534

Volumen 58

* Corresponding author.

E-mail: qvanetza@u.uchile.cl https://doi.org/10.14349/rlp.2026.v58.7 0120-0534/© 2026 Fundación Universitaria Konrad Lorenz. This is an open access article under the CC BY-NC-ND license (https://creativecommons.org/licenses/by-nc-nd/4.0/). Revista Latinoamericana de Psicología (2026) 58, e5807 https://doi.org/10.14349/rlp.2026.v58.7 https://revistalatinoamericanadepsicologia.konradlorenz.edu.co/ Role of early adverse experiences and intolerance of uncertainty in fear, avoidance, and relief learning María Consuelo San Martín a , Rodrigo C. Vergara c,d , Mario A. Laborda b

,

Gonzalo Miguez b , Paulina Pino-Ruy-Pérez b , Andrea Sánchez-Corzo e

,

Bram Vervliet f,g , Vanetza E. Quezada-Scholz b, * a School of Psychology, Universidad de los Andes, Santiago, Chile b Department of Psychology, Universidad de Chile, Santiago, Chile c Department of Kinesiology, Faculty of Arts and Physical Education, Universidad Metropolitana de Ciencias de la Educación, Santiago, Chile d National Artificial Intelligence Centre CENIA, Santiago, Chile e Department of Psychology, Ludwig-Maximilians-Universität München, München, Germany f Department of Brain and Cognition, KU, Leuven, Belgium g Leuven Brain Institute, KU, Leuven, Belgium Received 20 November 2025; accepted 22 June 2026 Abstract | Background and Objectives: This study replicates a previous online study in vivo, examining fear discrimina­ tion learning and avoidance generalisation in a laboratory setting. We investigated whether trait anxiety, intolerance of uncertainty (IU), and early maltreatment experiences are associated with variability in fear conditioning, avoidance be­ haviour, and relief-related responding in a non-clinical adult sample. Method: Forty-two university students completed a three-phase paradigm (fear conditioning, avoidance conditioning, generalisation test). Conditioned stimuli were coloured lights (avoidable CS+, unavoidable CS+, CS−); the US (unconditioned stimuli) consisted of aversive images from the Interna­ tional Affective Picture System IAPS (#1304, #1525). Outcomes were expectancy and anxiety ratings, avoidance frequency, relief ratings, and skin conductance responses (exploratory). Individual differences (trait anxiety, intolerance of uncer­ tainty, early adverse experiences) were also examined. Results: Participants discriminated CS+ from CS− in expectancy and anxiety, replicating prior findings. Skin conductance did not show differential responding. Trait anxiety and intol­ erance of uncertainty showed no consistent effects. In an exploratory MAES (Maltreatment Abuse and Exposure Scale) subsample (n = 33), higher exposure to childhood sexual abuse was associated with higher relief during avoidance. An ex­ ploratory cluster analysis identified two response profiles differing in anticipatory expectancy and avoidance consistency. Conclusions: These findings replicate discrimination learning under controlled laboratory conditions and extend prior work by characterising avoidance behaviour, relief responding, and individual differences in learning trajectories. Results should be interpreted with caution given the sample size and the exploratory nature of several analyses. Keywords: Human fear conditioning, avoidance learning, intolerance of uncertainty, early adversity, experimental psy­ chopathology © 2026 Fundación Universitaria Konrad Lorenz. This is an open access article under the CC BY-NC-ND license (https://creativecommons.org/licenses/ by-nc-nd/4.0/).

2/13 M. C. San Martín et al.

Fear is an adaptive response enabling organisms to anticipate and respond to threats. Fear conditioning paradigms have traditionally focused on acquisition, but fear learning also involves discrimination between threat and safety cues, generalisation, avoidance, and relief-related responding (Hofmann, 2008; Vervliet et al., 2017). These processes are particularly relevant for understanding anxiety vulnerability, as individuals differ in how they perceive and regulate threat-related cues (Andreatta et al., 2020; Flores et al., 2018; Herzog et al., 2021). Briefly, discrimination = distinguishing threat (CS+) from safety (CS−); generalisation = responding to new cues similar to the CS+; avoidance = action that pre­ vents the aversive event; relief = positive affect when an expected aversive event does not occur. Within this framework, avoidance plays a central role. It is maintained through negative reinforcement: when an expected aversive event is omitted following the avoidance response, the resulting relief reinforc­ es and strengthens the behaviour (Vervliet et al., 2017). Over time, persistent avoidance may interfere with discrimination learning and contribute to fear gener­ alisation (Declercq & De Houwer, 2008; Lovibond et al., 2000; San Martín et al., 2023; Zbozinek et al., 2021). The present study uses a generalisation test design based on two theoretical gradients — a safety gradient (threat → safety cues) and an avoidability gradient (avoidable → unavoidable threat) — building on classic work on ex­ citatory and inhibitory generalisation (Hovland, 1937; Lissek et al., 2008).

The role of contextual and discriminative cues in modulating avoidance and fear responses has been ex­ amined across animal and human paradigms (Declercq & De Houwer, 2008; Leising et al., 2025; Lovibond et al., 2000; Zbozinek et al., 2021). Such processes are con­ El papel de las experiencias adversas tempranas y la intolerancia a la incertidumbre en el aprendizaje del miedo, la evitación y el alivio Resumen | Introducción/Objetivo: Este artículo replica un estudio previo realizado en línea, ahora en formato presencial, examina el aprendizaje de discriminación del miedo y la generalización de la evitación en condiciones de laboratorio. Se investigó si la ansiedad rasgo, la intolerancia a la incertidumbre (IU) y las experiencias tempranas de maltrato se asocian con variabilidad en el condicionamiento del miedo, la conducta de evitación y las respuestas de alivio en una muestra adulta no clínica. Método: 42 estudiantes universitarios completaron un paradigma de tres fases (condicionamiento del miedo, condicionamiento de evitación y prueba de generalización). Los estímulos condicionados fueron luces de colores (EC+ evitable, EC+ inevitable y EC−); el EI consistió en imágenes aversivas del IAPS (#1304, #1525). Se evaluaron expectativa y ansiedad, frecuencia de evitación, alivio y respuestas electrodérmicas (exploratorias). También se examinaron diferen­ cias individuales (ansiedad rasgo, intolerancia a la incertidumbre, experiencias tempranas). Resultados: Los participantes discriminaron EC+ y EC− en expectativa y ansiedad, replicando hallazgos previos. La respuesta electrodérmica no mostró diferenciación. La ansiedad rasgo y la intolerancia a la incertidumbre no presentaron efectos consistentes. En la submues­ tra MAES (n = 33), la mayor exposición a abuso sexual infantil se asoció con mayores niveles de alivio durante la evitación. Un análisis exploratorio de conglomerados identificó dos perfiles que difieren en expectativa anticipatoria y consistencia en la evitación. Conclusiones: Los hallazgos replican el aprendizaje de discriminación en condiciones de laboratorio con­ troladas y extienden investigaciones previas al caracterizar la conducta de evitación, las respuestas de alivio y las diferen­ cias individuales en trayectorias de aprendizaje. Los resultados deben interpretarse con cautela dado el tamaño muestral y el carácter exploratorio de varios análisis. Palabras clave: Condicionamiento del miedo humano, aprendizaje de evitación, intolerancia a la incertidumbre, trauma temprano, psicopatología experimental © 2026 Fundación Universitaria Konrad Lorenz. Este es un artículo Open Access bajo la licencia CC BY-NC-ND (https://creativecommons.org/licenses/ by-nc-nd/4.0/).

sidered central to the development of anxiety-related disorders (Cobos et al., 2022). Experimental paradigms assessing discrimination, avoidance, and relief-relat­ ed responding within a single procedure allow exam­ ination of how these components co-occur and vary across individuals, rather than establishing causal relationships.

Individual differences are central to fear learning. Childhood maltreatment heightens threat sensitivity, impairs discrimination, and promotes fear over-gen­ eralisation (the spread of conditioned fear responses to stimuli that resemble but were not paired with the threat cue; Duits et al., 2015; McLaughlin et al., 2016; Ma­ chlin et al., 2019; Oar et al., 2022; Pérez-Edgar et al., 2017; Segal & Gobin, 2022). Intolerance of uncertainty (IU) is a transdiagnostic trait — the tendency to experience uncertainty as threatening — linked to distress in am­ biguous contexts and excessive avoidance (Cobos et al., 2022; Flores et al., 2018; Freeston et al., 1994); early adver­ sity may contribute to its development (San Martín et al., 2023).

A recent study by San Martín et al. (2023) examined these processes online, assessing expectancy, anxiety, relief, and avoidance behaviour. They reported success­ ful discrimination learning and associations between early adversity, trait anxiety, IU, and variability in fear/ avoidance responses; however, the online format limit­ ed experimental control over the learning environment. The present study replicated and extended these findings in an in vivo laboratory paradigm, examin­ ing whether the behavioural patterns observed online could be reproduced under controlled conditions. Skin conductance was included as an exploratory psycho­ physiological measure; given prior evidence of limited electrodermal sensitivity in similar paradigms (Queza­

3/13 Role of early adverse experiences and intolerance of uncertainty in fear, avoidance, and relief learning da et al., 2018), it was treated as exploratory. Finally, to better capture variability in learning processes across individuals, we explored differences in response trajec­ tories using a data-driven clustering approach (Reyn­ olds et al., 2006). This approach allowed the identifica­ tion of distinct response profiles based on expectancy, avoidance, and relief measures, providing a more nu­ anced characterisation of individual differences in fear learning.

The aims were to (1) replicate, in an in vivo laborato­ ry paradigm, the discrimination learning, avoidance acquisition, and relief responding observed in prior online work; (2) characterise generalisation along two theoretical gradients — a safety gradient (excitatory/ inhibitory, from CS+ to CS−) and an avoidability gra­ dient (from avoidable to unavoidable CS+); (3) examine associations between individual-difference variables (early adverse experiences, intolerance of uncertainty, and trait anxiety) and these learning processes; and

(4) explore variability in response patterns through

clustering. The study describes patterns of association rather than causal mechanisms.

Method Participants A total of 42 university students participated (11 men, 31 women; M = 21 years, range = 19–26) and were entered into a US$70 lottery. All provided informed consent un­ der procedures approved by the Ethics Committee of the Faculty of Social Sciences, University of Chile (ap­ proval No. 08-07/2020). Given the sensitive nature of the data (retrospective self-report of childhood maltreat­ ment; exposure to aversive images), a trained research­ er was present throughout, participants were informed of aversive stimuli and their right to withdraw, confi­ dentiality was ensured, and standardised debriefing was provided; no participant required referral. An a priori power analysis (G*Power 3.1; Faul et al., 2009) for an F test for a RM-ANOVA with one with­ in-subjects factor (three levels), assuming a medium effect (Cohen’s f = 0.25), a = .05, 1−b = .95, r = .50, and e = 1, indicated N = 43; we recruited 42, leaving the study slightly underpowered. The ANCOVA and correlation­ al analyses involving individual-difference predictors were not specifically powered and are interpreted as exploratory.

Early adverse experiences (MAES) were assessed in a subset (n = 33) during a follow-up procedure required by the Ethics Committee. This requirement limited full data availability for this measure, and analyses involv­ ing early adversity are interpreted as exploratory. Stimuli and apparatus Conditioned stimuli and generalisation stimuli. The conditioned stimuli (CS) were images of an office room with a desktop lamp whose colour varied. Three colours served as CS: yellow (580 nm), green (502 nm), and red (642 nm), adapted from San Martín et al. (2020). Yellow was always the avoidable CS+ (CSav+); the assignment of green and red to the CS− and unavoidable CS+ (CSunav+) roles was counterbalanced across participants. Coun­ terbalancing was not included as a between-subjects factor in the primary RM-ANOVAs, as the sample size was insufficient to test for colour-driven effects; inspec­ tion of the data did not suggest systematic differences between subgroups.

Generalisation stimuli (GS) consisted of intermedi­ ate colours along the same perceptual dimension: two between yellow and red (600, 620 nm) and two between yellow and green (560, 540 nm). Together with the three CS, these stimuli form a graded perceptual continu­ um spanning two theoretical generalisation gradients (Hovland, 1937; Lissek et al., 2008): (a) a safety gradient running from the CS+ (threat) to the CS− (safety), on which conditioned responses are expected to decrease with perceptual distance from the CS+ (excitatory gra­ dient) and to be lowest at the CS− (inhibitory gradi­ ent); and (b) an avoidability gradient running from the avoidable to the unavoidable CS+, reflecting variation in stimulus controllability (adapted from San Martín et al., 2020).

Unconditioned stimuli. The unconditioned stimuli (US) were two aversive images from the International Affective Picture System (IAPS; Lang et al., 2008), #1304 and #1525, both depicting attacking dogs (low valence, high arousal; Chilean validation: Moreno et al., 2016). The images used can be seen in Figure 1. Questionnaires The Intolerance of Uncertainty Scale (IUS) includes

27 items (Freeston et al., 1994; Spanish adaptation:

González Rodríguez et al., 2006). The State-Trait Anxie­ ty Inventory (STAI; Spielberger et al., 1982; Chilean adap­ tation: Vera-Villarroel et al., 2007) includes 20 State and

20 Trait items. The Maltreatment Abuse and Exposure

Scale (MAES) consists of the 52 retrospective items of the MACE (Teicher & Parigger, 2015); the chronological/ visual component of the original MACE — whose va­ lidity has been questioned — was not used, hence the alternative label MAES. The scale was translated and content validity was assessed by three judges; a formal interrater reliability coefficient (k) was not computed, which is a limitation of the present adaptation. Procedure After providing informed consent, participants com­ pleted the IUS and STAI in PsychoPy (version 2020.2.3; Peirce et al., 2019) and were fitted with electrodes; with­ in the same week they completed the MAES via Sur­ veyMonkey. The task closely followed San Martín et al. (2023) and comprised three phases: fear conditioning, avoidance conditioning, and generalisation (Figure 2). Fear conditioning phase. The phase included 16 tri­ als (4 CSav+, 4 CSunav+, 8 CS−) in pseudorandom order. Each trial began with a 3-s baseline (lamp off) followed by 6-s CS presentation. Intertrial intervals ranged from

12 to 18 s (M = 15 s). The US followed the offset of all CS+

trials; the CS− was never reinforced.

Measures. Expectancy ratings (E-VAS) measured, on each trial, how strongly the participant expected the aversive image to follow the lamp cue. Ratings were made during CS presentation on a horizontal visual

4/13 M. C. San Martín et al.

analog scale (0 = Not likely, 100 = Very likely; Figure 2A). Anxiety ratings (A-VAS) were collected after the phase for each stimulus type (0 = Not anxious, 100 = Very anx­ ious; Figure 2B). Skin conductance responses (SCR) were recorded continuously. Participants practiced before completing the task.

Avoidance conditioning phase. The phase includ­ ed 24 trials, divided into two blocks. In each block, each stimulus type (CSav+, CSunav+, CS−) was presented four times. Trial structure was identical to the Pavlovian phase (3-s baseline followed by 6-s CS presentation). During each CS presentation, a red button appeared for

2 s (seconds 2–4), indicating that a response could pre­

vent the US.

Avoidance was deterministic: responses successful­ ly prevented the US for the CSav+ but had no effect for the CSunav+. The CS− was never followed by the US. Tri­ als were presented in pseudorandom order. Measures. Avoidance behaviour (button presses) and SCR were recorded. Relief was rated once per stimu­ lus type at the end of the phase using a horizontal visual analog scale (0 = No relief, 100 = Very strong relief; Fig­ ure 2C); a single post-phase rating per stimulus was used (rather than trial-by-trial rating as in San Martín et al., 2023) to limit interruptions during the in vivo ses­ sion. Participants practiced before completing the task. Generalisation phase. The generalisation phase comprised two consecutive blocks: an avoidance block and an expectancy block. Each included 14 trials (7 stimuli: 3 CS and 4 GS, each presented twice). No US were delivered.

Trial structure was identical to the previous phases (3-s baseline followed by 6-s CS presentation). No uncon­ ditioned stimuli (US) were delivered during this phase. In the avoidance generalisation block, participants responded on every trial; avoidance frequency was recorded. In the expectancy generalisation block, no avoidance responses were made; expectancy ratings were collected at the end of the phase (0 = Not likely, 100 = Very likely; Figure 2A). Relief ratings were also collect­ ed after the phase (0 = No relief, 100 = Very strong relief; Figure 2C).

Data analysis Analyses were organised into three components. Pri­ mary analyses used repeated-measures ANOVAs with Stimulus as within-subject factor. RM-ANCOVAs includ­ ed trait anxiety (STAI-T), intolerance of uncertainty (IUS), and early adverse experiences (MAES) as continu­ ous predictors; these were not specifically powered and are interpreted with caution. Sphericity was assessed with Mauchly’s test and Greenhouse–Geisser correc­ tions applied when violated. All pairwise post hoc com­ parisons were Bonferroni-corrected (k = 3 in acquisi­ tion; k = 21 in generalisation); reported post hoc p-values are corrected. Effect sizes are partial h² with 95% CI for omnibus effects and Cohen’s dz for pairwise contrasts. Secondary analyses included Pearson correlations. Electrodermal activity (EDA) signals — from which SCR was computed — were recorded at 2000 Hz (MP160 Biopac System with BioNomadix wireless module). Sig­ nals were visually inspected and discrete movement Figure 1. Experimental design Note. During the fear conditioning phase, the yellow and red lamps were followed by the aversive US (CS+), but the green lamp was not (CS-). During avoidance conditioning, space bar clicking during the red button effectively cancelled the aver­ sive US at the offset of the yellow lamp (CS+ avoidable), but not the red lamp (CS+ unavoidable). During the generalisation phase all CS were presented, and also 2 additional (similar but not identical) colour lamps between CS- and CS+ avoidable (safety dimension: GS1, GS2) and 2 colour lamps between CS+ avoidable and CS+ unavoidable (avoidability dimension: GS3, GS4) were presented. None of the stimuli were followed by an aversive stimulus.

5/13 Role of early adverse experiences and intolerance of uncertainty in fear, avoidance, and relief learning artifacts removed manually (AcqKnowledge; BIOPAC Systems, 2021). Clean signals were z-standardised and decomposed into tonic and phasic components using the eda_phasic function from NeuroKit2 (Makowski et al., 2021). SCR was scored trial-by-trial as the peak SCL during the 6-s CS window minus the 2-s pre-CS base­ line; negative values were set to zero and positive values were square-root transformed.

Figure 2. A: Expectancy; B: anxiety and; C: relief rating scale Exploratory analyses examined heterogeneity in learning trajectories with a clustering approach. A sim­ ple linear regression was fit to each participant’s tri­ al-by-trial scores in expectancy, avoidance, and relief per condition; intercept, slope, and R² were extracted (27 features). One participant with missing expectancy data was excluded (final n = 41). Features were z-stand­ ardised and clustered with PAM (Reynolds et al., 2006) using Euclidean distance; the number of clusters was selected with the Silhouette method (k = 2–6) and val­ idated in PCA space. Cluster differences across the 27 features were tested with t-tests or Mann–Whitney U tests; given the exploratory nature and the low cluster separation observed, these comparisons are reported as descriptive characterisations and no formal multi­ ple-comparison correction was applied.

Results Results are organised into primary analyses examin­ ing conditioning and generalisation effects, followed by analyses testing the role of individual differences, and finally exploratory analyses of response profiles. Across all analyses, we trace how findings from different re­ sponse systems converge on a common pattern, against which the exploratory analyses are then interpreted. Conditioning and generalisation effects Fear conditioning phase Expectancy Visual Analog Scale (E-VAS). No differences were found between CSav+ and CSunav+ (p = .816, dz = −0.04); both were collapsed. A RM-ANOVA with Stimulus (CS+/CS−) as within-subject factor showed a significant main effect, F(1, 40) = 69.72, p < .001, MSE = 547.52, hp² = .635, 95% CI [.43, .74] (Mauch­ ly’s test is not applicable for k = 2; n = 41), indicating that CS+ elicited greater US expectancy than CS− (Cohen’s dz = 1.30) (Figure 3A). A three-stimulus analysis confirmed this pattern, F(1.29, 51.52) = 60.78, p < .001, MSE = 418.75, hp² = .603, 95% CI [.46, .69] (Mauchly’s W = .447, p < .001, e = .64).

Anxiety Visual Analog Scale (A-VAS). CSav+ and CSunav+ were collapsed (no difference, p = .491, dz = −0.11). A RM-ANOVA with Stimulus (CS+/CS−) as with­ in-subject factor revealed a significant main effect, F(1,

41) = 17.273, p < .001, MSE = 919.45, hp² = .296, 95% CI [.08,

.48] (Mauchly’s test is not applicable for k = 2), indicating higher anxiety ratings for CS+ than CS− (Cohen’s dz = 0.64) (Figure 3B).

Skin conductance response. A repeated-measures ANOVA compared CS+ (avoidable, unavoidable) and CS− conditions. No significant differences emerged between CS+ and CS− (p = .738; Figure 3C); therefore, no further analyses were conducted. These results indicate that SCR did not show reliable discrimination between threat and safety cues in this paradigm and are there­ fore not further interpreted.

Avoidance Conditioning phase Avoidance frequency. A RM-ANOVA examined avoid­ ance frequency with Stimulus (CS−, CSav+, CSunav+) as within-subject factor. The Greenhouse–Geisser correct­ ed analysis showed a main effect, F(1.72, 70.38) = 6.71, p =

6/13 M. C. San Martín et al.

.003, MSE = 0.92, hp² = .143, 95% CI [.02, .27] (Mauchly’s W = .835, p = .027, e = .86; n = 42). Bonferroni-corrected post hoc tests revealed that CSunav+ elicited more avoidance than CS− (corrected p = .011, dz = 0.48); the remaining contrasts were not significant (CSav+ vs. CSunav+ cor­ rected p = .201, dz = −0.29; CSav+ vs. CS− corrected p = .096, dz = −0.34), suggesting limited sensitivity to con­ trollability (Figure 4A).

Relief ratings. A RM-ANOVA with Stimulus (CS−, CSav+, CSunav+) as within-subject factor showed a main effect, F(2, 82) = 26.97, p < .001, MSE = 357.14, hp² = .397, 95% CI [.23, .52] (sphericity met: Mauchly’s W = .97, p = .545) (Figure 4B). Bonferroni-corrected post hoc tests showed lower relief for CSunav+ than CS− (corrected p < .001, dz = 0.71) and than CSav+ (corrected p < .001, dz = 1.18); the CS− vs. CSav+ contrast was not significant (corrected p = .092, dz = −0.35).

Generalisation test phase The generalisation test characterises response gradi­ ents across stimuli varying in perceptual similarity to the conditioned cues. Under successful discriminative generalisation, responses should be highest at the CS+ and decrease with perceptual distance from it (excita­ tory gradient); they should also be lowest at the CS− and increase with perceptual distance from it (inhibitory gradient). A flat profile is consistent either with com­ plete generalisation or with extinction-like attenuation given that no US was delivered. Differential gradients across response systems (e.g., emotional vs. behav­ ioural) would indicate dissociation between response domains.

Expectancy Visual Analog Scale (E-VAS). A repeat­ ed-measures ANOVA examined expectancy ratings Figure 3. Fear conditioning phase: Expectancy US, Anxiety rating and Skin conductance response Note. The results encompass three key areas in the fear conditioning phase: expectancy US (A), anxiety ratings (B), and skin conductance response (C). For each area, the black bar represents CS- and a light gray or gray bar represents the combined average of CS+ avoidable and CS+ unavoidable. Standard errors of the mean are indicated by error bars, with significance levels marked as *p < 0.05, **p < 0.01, and ***p < 0.001. Detailed explanations are available in the accompanying text.

7/13 Role of early adverse experiences and intolerance of uncertainty in fear, avoidance, and relief learning with Stimulus (CS−, GS1, GS2, CSav+, GS3, GS4, CSunav+) as within-subject factor. The Greenhouse–Geisser cor­ rected analysis showed a main effect of stimulus, F(4.44, 173.32) = 22.13, p < .001, MSE = 802.03, hp² = .362, 95% CI [.25, .43] (Mauchly’s W = .31, p = .002, e = .74; n = 41). Bonfer­ roni-corrected post hoc tests (k = 21) revealed that the unavoidable CS+ differed significantly from each of the other six stimuli (all corrected p < .01; Cohen’s |dz| range 0.68–1.35). Two additional contrasts also survived cor­ rection: GS1 vs. GS4 and GS2 vs. GS4 (corrected p ≤ .004, |dz| ≥ 0.66). No other contrasts reached significance. These results indicate an expectancy gradient consist­ ent with excitatory generalisation centred on the una­ voidable CS+ (Figure 5A).

Avoidance frequency. A Greenhouse–Geisser cor­ rected RM-ANOVA with Stimulus (7 levels) as with­ in-subject factor showed a significant but small main effect, F(4.55, 186.66) = 3.12, p = .012, MSE = 6.80, hp² = .071 (Mauchly’s W = .436, p = .042, e = .76); however, no pairwise Bonferroni-corrected contrast reached sig­ nificance (all corrected p ≥ .063; |dz| ≤ 0.49). This essen­ tially flat pairwise profile is consistent either with com­ plete generalisation of avoidance across the perceptual continuum or with extinction-like attenuation in the absence of US delivery; the divergence from the differ­ entiated expectancy and relief gradients suggests a dis­ sociation between behavioural and emotional response systems during generalisation (Figure 5B). Relief ratings. A Greenhouse–Geisser corrected RM-ANOVA showed a main effect of Stimulus, F(3.98, 163.03) = 12.02, p < .001, MSE = 333.12, hp² = .227, 95% CI [.13, .30] (Mauchly’s W = .25, p < .001, e = .66). Bonferroni-cor­ rected post hoc tests (k = 21) revealed that CSunav+ elic­ ited significantly lower relief than each of the other six stimuli (all corrected p < .001; Cohen’s |dz| 0.74–0.83); no other contrasts reached significance. The gradient is consistent with excitatory generalisation along the safety dimension: relief is highest at safety-related cues and lowest at the non-controllable threat cue (Figure 5C). Figure 4. Avoidance conditioning: Avoidance Frequency, Relief pleasantness and Skin Conductance Response Note. The results display data from three measures: (a)frequency of button pressing, (b) relief pleasantness, and (c) skin conductance response during avoidance conditioning. Each measure uses bars to represent CS- (black bar), avoidable CS+ (dark gray bar), and CS+ unavoidable (black or white bar). Generalisation stimuli, represented by bars between CS+ unavoidable and avoidable, are similar to both. Error bars denote standard errors of the mean, with significance levels marked as *p < 0.05, **p < 0.01, and ***p < 0.001. For detailed explanations, see the associated text.

8/13 M. C. San Martín et al.

Effects of individual differences Descriptive statistics for all psychometric measures are presented in Table 1. Scores on the IUS and STAI-T were consistent with a non-clinical university sample. In the MAES subsample (n = 33), exposure to early ad­ versity was prevalent across most subscales — every participant endorsed at least one item on the total scale — with the highest endorsement rates observed for Non-Verbal Emotional Abuse (32/33, 97.0%), Peer Verbal Abuse (31/33, 93.9%), and Parental Verbal Abuse (25/33,

75.8%).

A RM-ANCOVA examined individual differences across all phases. No consistent effects were observed for STAI-T or IUS across outcomes. The only significant finding was in the avoidance phase: an association be­ tween relief ratings and childhood sexual abuse expo­ sure, F(2, 62) = 9.053, p = .027, MSE = 2907.878, hp² = .226, 95% CI [0.049–0.340]. Given the reduced subsample, this Figure 5. Generalisation phase: Expectancy rating scale, Relief pleasantness, Avoidance and Skin Conductance responses Note. The image displays experimental results across four categories: Relief Pleasantness (A), Expectancy Rating Scale (B), Avoidance Frequency (C), and Skin Conductance Response (D). Each category depicts bars for three stimulus types: CS+ unavoidable (white bar), CS+ avoidable (light gray bar), and CS- (dark bar). Intermediate bars represent generalisation stimuli, akin to both CS+ types and CS-. Error bars indicate the standard error of the mean, with *p < 0.05, **p < 0.01, and ***p < 0.001 denoting statistical significance. Further details are provided in the accompanying text.

9/13 Role of early adverse experiences and intolerance of uncertainty in fear, avoidance, and relief learning finding is interpreted as exploratory. A follow-up corre­ lation confirmed the direction: higher severity of child­ hood sexual abuse was associated with higher relief to CSav+ (p = .012).

Exploratory analyses: Response profiles Differences in the process of conditioning. The Sil­ houette method suggested two clusters. When project­ ed onto the first two PCA components, clusters did not overlap, supporting the two-cluster solution. However, Silhouette scores were low (range = 0.05–0.08), indicat­ ing limited cluster separation; results are therefore pre­ liminary and descriptive.

To understand the obtained clusters, we began by ex­ ploring how they differ on the regression features used to obtain the clusters. Figure 6 presents the two cluster solutions for the 9 regression features used within the three experimental conditions.

Of the 27 features, two primarily drove cluster differ­ ences: Pavlovian expectancy intercept and R² in avoid­ ance. The High Responsiveness (HR) cluster showed higher anticipatory expectancy and greater model fit in avoidance, indicating more consistent responding; the Low Responsiveness (LR) cluster showed lower expec­ tancy and less predictable avoidance. Clusters were not interpreted as functional profiles given that generalisa­ tion avoidance was not correlated with acquisition ex­ pectancy. Silhouette values were low (range 0.05–0.08) and hierarchical clustering (Ward’s method) yielded near-zero concordance (ARI = −0.02), confirming limit­ ed structural robustness; profiles are treated as prelim­ inary descriptions.

Taken together, these results trace a coherent pat­ tern rather than a set of isolated effects. Threat acqui­ sition was evident in self-report (expectancy and anx­ iety) but not in skin conductance, and this self-report signal carried forward into the operant and generali­ sation phases. During avoidance learning, participants reliably distinguished the unavoidable threat cue from the safety cue while showing limited sensitivity to con­ trollability, and the relief data converged on the same conclusion by isolating the non-controllable threat cue as least relieving. In the generalisation test, expectancy and relief produced differentiated gradients centred on the threat and safety dimensions, whereas avoidance Table 1. Descriptive Statistics for Psychometric Measures (IUS, STAI-T) and Early Adversity Exposure (MAES) Measure n M SD Mdn Range Endorsement n (%) Full Sample (n = 42) IUS Total

42

73.21

18.77

73.00

42–107 — Inhibitory IU

42

43.52

12.07

42.50

23–67 — Prospective IU

42

29.62

8.23

32.00

13–43 —

STAI-T

41

27.85

11.07

27.00

7–52 — Subsample (n = 33) Sexual Abuse*

33

0.79

1.41

0.00

0–4

9 (27.3%)

Parental Verbal Abuse

33

4.70

3.61

5.00

0–10

25 (75.8%)

Non-Verbal Emotional Abuse

33

4.18

2.24

4.00

0–8

32 (97.0%)

Parental Physical Maltreatment

33

2.82

2.88

2.00

0–10

21 (63.6%)

Interparental Violence

33

1.39

2.81

0.00

0–8

8 (24.2%)

Witnessing Sibling Abuse

33

0.73

1.31

0.00

0–3

8 (24.2%)

Peer Verbal Abuse

33

6.24

2.99

6.00

0–10

31 (93.9%)

Peer Physical Abuse

33

2.24

3.15

0.00

0–10

14 (42.4%)

Emotional Neglect

31

2.45

2.35

2.00

0–8

21 (67.7%)

Physical Neglect

31

0.77

1.69

0.00

0–6

7 (22.6%)

Total MAES Score

33

26.12

14.07

25.00

2–59

33 (100.0%)

No. of Maltreatment Types

33

2.64

2.18

2.00

0–9

27 (81.8%)

Note. IUS = Intolerance of Uncertainty Scale; Inhibitory IU = inhibitory intolerance of uncertainty subscale; Prospective IU = prospective intolerance of uncertainty subscale; STAI-T = State-Trait Anxiety Inventory, Trait subscale; MAES = Maltreatment Abuse and Exposure Scale (Teicher & Parigger, 2015). Endorsement = number (%) of participants with a subscale score > 0, computed over the number of participants with valid data on that subscale (see column n). Dashes indicate that endorsement is not applicable for continuous scales. One participant did not complete the STAI-T and two participants did not complete the Emotional and Physical Neglect items; these cases were excluded from the corresponding rows. * Subscale with a significant association with relief ratings during avoidance conditioning.

10/13 M. C. San Martín et al.

remained essentially flat—pointing to a consistent dis­ sociation between emotional and behavioural systems. Against this common backbone, the individual-differ­ ence and clustering analyses indicated where early ad­ versity (specifically childhood sexual abuse, linked to heightened relief) and response consistency (the HR/LR profiles) modulated the shared pattern; these explora­ tory findings are therefore interpreted within, rather than apart from, the primary effects.

Discussion The present study examined how childhood maltreat­ ment, trait anxiety, and IU relate to avoidance, dis­ crimination, and generalisation. During Pavlovian conditioning, participants showed clear CS+/CS− dis­ crimination in expectancy and anxiety, replicating pri­ or work (San Martín et al., 2023; Vervliet et al., 2017) and extending it to an in vivo setting. SCR did not discrim­ inate in any phase, likely because data collection took Figure 6. Two cluster solutions for the 9 regression features used Note. The graph displays box plots with the Y-axis representing regression values for expectancy, avoidance frequency, and relief. The X-axis categorises responses by stimulus type: Control, Evitative, and Non-Evitative.

11/13 Role of early adverse experiences and intolerance of uncertainty in fear, avoidance, and relief learning place in a shared space with elevated ambient noise that reduced signal-to-noise ratio. This dissociation between self-report/behaviour and autonomic arousal underscores the value of measuring several response systems (Quezada et al., 2018).

In the operant phase, participants did not show dis­ criminative avoidance between the avoidable CS+ and the CS−, but did differentiate the unavoidable CS+ from the CS−, indicating the unavoidable stimulus was per­ ceived as more threatening. Importantly, avoidance on CSunav+ trials cannot have been negatively reinforced by US omission because the US was delivered regard­ less; this likely reflects elevated threat perception rath­ er than instrumental contingencies. The absence of differences involving the avoidable CS+ suggests limit­ ed sensitivity to controllability, although a generalised threat response is an alternative explanation. During the generalisation test, relief gradients were clearly differentiated whereas avoidance was essentially flat, suggesting a dissociation between emotional and be­ havioural systems (Quezada-Scholz et al., 2022). The avoidance button may have acted as an occasion set­ ter signalling response efficacy (Declercq & De Houwer,

2008).

Childhood sexual abuse exposure (MAES) correlated significantly with higher relief ratings during avoid­ ance. This effect was specific to relief — not observed in discrimination or avoidance frequency — suggesting that early adversity preferentially influences the affec­ tive consequences of avoidance rather than learning processes per se. These findings align with prior evi­ dence that early adversity heightens threat sensitivity and avoidance motivation (Duits et al., 2015; Kreutz­ mann et al., 2021; Wicking et al., 2016), and are consist­ ent with the hypothesis — though they cannot test it directly — that early sexual abuse may amplify the affective value of successful avoidance. Given the ex­ ploratory subsample (n = 33), this should be regarded as preliminary.

Cluster analysis identified two response profiles: High Responsiveness (HR), with greater anticipatory expectancy and more consistent avoidance — resem­ bling patterns associated with anxiety propensity (Ball & Gunaydin, 2022) — and Low Responsiveness (LR), with lower expectancy and more variable avoidance. Because the cluster solution showed low Silhouette values (range 0.05–0.08) and near-zero agreement with hierarchical clustering (ARI = −0.02), these profiles are presented as preliminary descriptive patterns and should not be in­ terpreted as discrete subtypes. Larger samples and rep­ lication with hierarchical clustering are needed. Overall, the study replicates and extends San Martín et al. (2023): discrimination learning was robustly rep­ licated, and we add a more nuanced characterisation of avoidance under controlled laboratory conditions — limited sensitivity to controllability, dissociation between emotional and behavioural systems during generalisation, and a novel association between early adversity and relief. Cluster-based approaches suggest potential pathways of vulnerability and resilience that warrant replication in larger samples integrating phys­ iological, behavioural, and self-report measures. Limitations Several limitations qualify these conclusions. First, the sample was modest (N = 42) and drawn from a non-clin­ ical, predominantly female university population, which constrains statistical power for individual-dif­ ference analyses and limits generalisation to clinical or community samples. Second, the early-adversity measure (MAES) was available only for a subset of par­ ticipants (n = 33) collected during a post-participation follow-up, so all analyses involving childhood maltreat­ ment, including the association with relief, are explor­ atory and rest on retrospective self-report, with the at­ tendant risk of recall bias. Third, skin conductance did not discriminate between threat and safety cues, most plausibly because data were collected in a shared space with elevated ambient noise; the autonomic channel therefore offers no corroboration for the self-report and behavioural effects. Fourth, no US was delivered during the generalisation test, so the flat avoidance gradient remains open to two readings, complete generalisation versus extinction-like attenuation, that the present de­ sign cannot disambiguate. Finally, the response profiles identified by clustering showed low internal robustness (low Silhouette values and near-zero agreement with hierarchical clustering) and should be regarded as pre­ liminary descriptions rather than discrete subtypes. Future directions and applied implications Future studies should address these limitations by re­ cruiting larger and more diverse samples, including clinically anxious participants, and by assessing early adversity prospectively to reduce reliance on retrospec­ tive reports. Manipulating controllability more explic­ itly, delivering occasional reinforcement during the generalisation test to separate complete generalisation from extinction, and improving the recording envi­ ronment for psychophysiological measures would help adjudicate the dissociations observed here. Integrat­ ing physiological, behavioural, and self-report indices within the same design, together with replication of the HR/LR profiles in adequately powered samples, would clarify whether response consistency reflects a stable vulnerability marker. In applied terms, the specific link between childhood sexual abuse and heightened relief during successful avoidance suggests that the affective payoff of avoidance, rather than threat acquisition per se, may be a useful target in exposure-based and inhib­ itory-learning treatments for individuals with adverse developmental histories. Likewise, the dissociation be­ tween what people feel and what they do during gen­ eralisation underscores the clinical value of tracking emotional and behavioural outcomes separately, since reductions in subjective threat may not translate di­ rectly into reduced avoidance.

Funding This research was supported by the Agencia Nacional de Investigación y Desarrollo de Chile (ANID) through the fund granted to Maria Consuelo San Martín [Fonde­ cyt #11250037], to Vanetza E. Quezada-Scholz [Fondecyt

12/13 M. C. San Martín et al.

#11170143], and by funding from the Centro Nacional de Inteligencia Artificial CENIA, FB210017, BASAL, ANID granted to Rodrigo C. Vergara.

Conflict of interest There is no conflict of interest to disclose. Acknowledgements We would like to thank all the participants who chose to take part in the experiment.

References Andreatta, M., Neueder, D., Herzog, K., Genheimer, H., Schiele, M. A., Deckert, J., Domschke, K., Reif, A., Wieser, M. J., & Pauli, P. (2020). Generalization of conditioned contextual anxiety and the modulatory effects of anxiety sensitivity. Neurotherapeutics, 17(3), 1239-1252. https://doi.org/10.1007/ s13311-020-00831-8 Ball, T. M., & Gunaydin, L. A. (2022). Measuring maladaptive avoidance: From animal models to clinical anxiety. Neuro­ psychopharmacology, 47(5), 978-986. https://doi.org/10.1038/ s41386-021-01263-4 BIOPAC Systems. (2021). AcqKnowledge (Version 5.0) [Software]. BIOPAC Systems. https://www.biopac.com Cobos, P. L., Quintero, M. J., Ruiz-Fuentes, M., Vervliet, B., & López, F. J. (2022). The role of relief, perceived control, and prospective intolerance of uncertainty in excessive avoid­ ance in uncertain-threat environments. International Jour­ nal of Psychophysiology, 179, 89-100. https://doi.org/10.1016/j. ijpsycho.2022.07.002 Declercq, M., & De Houwer, J. (2008). Evidence for the inter­ changeability of an avoidance behavior and a negative occasion setter. Learning & Behavior, 36(4), 290-300. https:// doi.org/10.3758/LB.36.4.290 Duits, P., Cath, D. C., Lissek, S., Hox, J. J., Hamm, A. O., Engelhard,

I. M., van den Hout, M. A., & Baas, J. M. P. (2015). Updated

meta-analysis of classical fear conditioning in the anxiety disorders. Depression and Anxiety, 32(4), 239-253. https://doi. org/10.1002/da.22353 Faul, F., Erdfelder, E., Lang, A.-G., & Buchner, A. (2009). Statisti­ cal power analyses using G*Power 3.1: Tests for correlation and regression analyses. Behavior Research Methods, 41(4), 1149-1160. https://doi.org/10.3758/BRM.41.4.1149 Flores, A., López, F. J., Vervliet, B., & Cobos, P. L. (2018). Intoler­ ance of uncertainty as a vulnerability factor for excessive and inflexible avoidance behavior. Behaviour Research and Therapy, 104, 34-43. https://doi.org/10.1016/j.brat.2018.02.008 Freeston, M. H., Rhéaume, J., Letarte, H., Dugas, M. J., & Ladou­ ceur, R. (1994). Why do people worry? Personality and Indi­ vidual Differences, 17(6), 791-802. https://doi.org/10.1016/0191-

8869(94)90048-5

González Rodríguez, M., Cubas León, R., Rovella, A. T., & Darias Herrera, M. (2006). Adaptación española de la Escala de Intolerancia hacia la Incertidumbre: procesos cognitivos, ansiedad y depresión. Psicología y Salud, 16(2), 219-233. Herzog, K., Andreatta, M., Schneider, K., Schiele, M. A., Dom­ schke, K., Romanos, M., Deckert, J., & Pauli, P. (2021). Reduc­ ing generalization of conditioned fear: Beneficial impact of fear relevance and feedback in discrimination training. Frontiers in Psychology, 12, 665711. https://doi.org/10.3389/ fpsyg.2021.665711 Hofmann, S. G. (2008). Cognitive processes during fear acqui­ sition and extinction in animals and humans: Implica­ tions for exposure therapy of anxiety disorders. Clinical Psychology Review, 28(2), 199-210. https://doi.org/10.1016/j. cpr.2007.04.009 Hovland, C. I. (1937). The generalization of conditioned re­ sponses: I. The sensory generalization of conditioned re­ sponses with varying frequencies of tone. The Journal of General Psychology, 17(1), 125-148. https://doi.org/10.1080/00

221309.1937.9917977

Kreutzmann, J. C., Marin, M. F., Fendt, M., Milad, M. R., Ress­ ler, K., & Jovanovic, T. (2021). Unconditioned response to an aversive stimulus as predictor of response to conditioned fear and safety: A cross-species study. Behavioural Brain Re­ search, 402, 113105. https://doi.org/10.1016/j.bbr.2020.113105 Lang, P. J., Bradley, M. M., & Cuthbert, B. N. (2008). Internation­ al affective picture system (IAPS): Affective ratings of pictures and instruction manual (Technical Report A-8). University of Florida.

Leising, K. J., Nerz, J., Solorzano-Restrepo, J., & Bond, S. R. (2025). Are you studying occasion setting? A review for inquiring minds. Comparative Cognition & Behavior Reviews, 20, 5-44. https://doi.org/10.3819/CCBR.2025.200002 Lissek, S., Biggs, A. L., Rabin, S. J., Cornwell, B. R., Alvarez, R. P., Pine, D. S., & Grillon, C. (2008). Generalization of conditioned fear-potentiated startle in humans: Experimental valida­ tion and clinical relevance. Behaviour Research and Therapy, 46(5), 678-687. https://doi.org/10.1016/j.brat.2008.02.005 Lovibond, P. F., Davis, N. R., & O’Flaherty, A. S. (2000). Protection from extinction in human fear conditioning. Behaviour Re­ search and Therapy, 38(10), 967-983. https://doi.org/10.1016/ S0005-7967(99)00121-7 Machlin, L., Miller, A. B., Snyder, J., McLaughlin, K. A., & Sher­ idan, M. A. (2019). Differential associations of deprivation and threat with cognitive control and fear conditioning in early childhood. Frontiers in Behavioral Neuroscience, 13, 80. https://doi.org/10.3389/fnbeh.2019.00080 McLaughlin, K. A., Sheridan, M. A., Gold, A. L., Duys, A., Lambert, H. K., Peverill, M., Heleniak, C., Shechner, T., Wojcieszak, Z., & Pine, D. S. (2016). Maltreatment exposure, brain struc­ ture, and fear conditioning in children and adolescents. Neuropsychopharmacology,

41(8),

1956-1964.

https://doi.

org/10.1038/npp.2015.365 Makowski, D., Pham, T., Lau, Z. J., Brammer, J. C., Lespinasse, F., Pham, H., Schölzel, C., & Chen, S. H. A. (2021). NeuroKit2: A Python toolbox for neurophysiological signal process­ ing. Behavior Research Methods, 53(4), 1689-1696. https://doi. org/10.3758/s13428-020-01516-y Moreno, C. P., Quezada, V. E., & Antivilo, A. (2016). Identificando un conjunto de imágenes provocadoras de miedo del Inter­ national Affective Picture System (IAPS) en una muestra chilena. Terapia Psicológica, 34(3), 209-215. https://doi. org/10.4067/S0718-48082016000300005 Oar, E. L., Johnco, C. J., Waters, A. M., Fardouly, J., Forbes, M. K., Magson, N. R., Richardson, C. E., & Rapee, R. M. (2022). Eye-tracking to assess anxiety-related attentional biases among a large sample of preadolescent children. Behaviour Research and Therapy, 153, 104079. https://doi.org/10.1016/j. brat.2022.104079 Peirce, J. W., Gray, J. R., Simpson, S., MacAskill, M. R., Höchen­ berger, R., Sogo, H., Kastman, E., & Lindeløv, J. (2019). Psy­ choPy2: Experiments in behavior made easy. Behavior Research Methods, 51(1), 195-203. https://doi.org/10.3758/ s13428-018-01193-y Pérez-Edgar, K., Morales, S., LoBue, V., Taber-Thomas, B. C., Al­ len, E. K., Brown, K. M., & Buss, K. A. (2017). The impact of negative affect on attention patterns to threat across the first 2 years of life. Developmental Psychology, 53(12), 2219-

2232. https://doi.org/10.1037/dev0000408

13/13 Role of early adverse experiences and intolerance of uncertainty in fear, avoidance, and relief learning Quezada, V. E., Laborda, M. A., Díaz, M. C., Navarro, V. M., Mallea, J., Repetto, P., Orellana, G., & Betancourt, R. (2018). Extinc­ tion cues do not reduce recovery of extinguished condi­ tioned fear in humans. International Journal of Psychology and Psychological Therapy, 18(1), 39-53.

Quezada-Scholz, V. E., Laborda, M. A., San Martín, C., Miguez, G., Alfaro, F., Mallea, J., Bacigalupo, F., & Miño, V. (2022). Cued fear conditioning in humans using immersive vir­ tual reality. Learning and Motivation, 78, 101803. https://doi. org/10.1016/j.lmot.2022.101803 Reynolds, A. P., Richards, G., de la Iglesia, B., & Rayward-Smith,

V. J. (2006). Clustering rules: A comparison of partitioning

and hierarchical clustering algorithms. Journal of Mathe­ matical Modelling and Algorithms, 5(4), 475-504. https://doi. org/10.1007/s10852-005-9022-1 San Martín, C., Jacobs, B., & Vervliet, B. (2020). Further charac­ terization of relief dynamics in the conditioning and gen­ eralization of avoidance: Effects of distress tolerance and intolerance of uncertainty. Behaviour Research and Therapy, 124, 103526. https://doi.org/10.1016/j.brat.2019.103526 San Martín, C., Laborda, M. A., Miguez, G., Sánchez, A., Vervliet, B., & Quezada-Scholz, V. (2023). Relation among trait anxie­ ty, intolerance to uncertainty and early maltreatment ex­ periences on fear discrimination learning and avoidance generalization online task. Journal of Behavior Therapy and Experimental Psychiatry, 81, 101886. https://doi.org/10.1016/j. jbtep.2023.101886 Segal, S. C., & Gobin, K. C. (2022). Threat-biased attention in childhood anxiety: A cognitive-affective developmen­ tal model. Journal of Affective Disorders Reports, 8, 100315. https://doi.org/10.1016/j.jadr.2022.100315 Spielberger, C. D., Gorsuch, R. L., & Lushene, R. E. (1982). STAI: Cuestionario de ansiedad estado-rasgo. Manual. TEA Edi­ ciones.

Teicher, M. H., & Parigger, A. (2015). The “Maltreatment and Abuse Chronology of Exposure” (MACE) scale for the ret­ rospective assessment of abuse and neglect during devel­ opment. PLOS ONE, 10(2), e0117423. https://doi.org/10.1371/ journal.pone.0117423 Vera-Villarroel, P., Celis-Atenas, K., Córdova-Rubio, N., Bue­ la-Casal, G., & Spielberger, C. D. (2007). Preliminary analysis and normative data of the State-Trait Anxiety Inventory (STAI) in adolescent and adults of Santiago, Chile. Terapia Psicológica, 25(2), 155-162. https://doi.org/10.4067/S0718-

48082007000200006

Vervliet, B., Lange, I., & Milad, M. R. (2017). Temporal dynam­ ics of relief in avoidance conditioning and fear extinction: Experimental validation and clinical relevance. Behaviour Research and Therapy, 96, 66-78. https://doi.org/10.1016/j. brat.2017.04.011 Wicking, M., Steiger, F., Nees, F., Diener, S. J., Grimm, O., Ruttorf, M., Schad, L. R., Winkelmann, T., Wirtz, G., & Flor, H. (2016). Deficient fear extinction memory in posttraumatic stress disorder. Neurobiology of Learning and Memory, 136, 116-126. https://doi.org/10.1016/j.nlm.2016.09.016 Zbozinek, T. D., Wise, T., Perez, O. D., Qi, S., Fanselow, M. S., & Mobbs, D. (2021). Pavlovian occasion setting in human fear and appetitive conditioning: Effects of trait anxiety and trait depression. Behaviour Research and Therapy, 147,

103986. https://doi.org/10.1016/j.brat.2021.103986