According to World Health Organization data (Krug et al., 2002), IPV may lead to different patterns of sanitary issues, such as HIV infection, sexually transmitted diseases, induced abortion, low birth weight, premature birth, alcohol consumption, suicide, self-harm, and death by homicide. The consequences of domestic violence could therefore have important effects on the health of victims (Lutgendorf, 2019). More generally, victims of GBV seem to be at greater risk of unwanted pregnancies, infections, sexual dysfunctions, and abortion (Pallitto et al., 2013); moreover, in some countries, such as the United States of America, GBV seems to be the primary cause of injury in women (Tjaden and Thoennes, 2001).

Even though most of these findings are based on studies about domestic violence, these consequences are not limited to intra-family violence only: a study conducted in six different European countries (Lukasse et al., 2015) found that among women who reported having an unwanted pregnancy, 24.5% were abused during their lifetime and 38.5% were recently abused. The consequences of GBV are far-reaching, extending beyond immediate health concerns to impact the long-term wellbeing and decision-making capacity of victims. Such experiences have been observed to manifest in various psychological reactions, including depression, anxiety, and a diminished sense of self-esteem.

The health consequences of abuse may also have long-term secondary outcomes which could be related to the primary ones. Individuals suffering from depression may experience impairments in the executive (Snyder, 2013) and cognitive functions, e.g., attention and memory. Particularly, a recent study (Lawlor et al., 2020) found that individuals suffering from Major Depressive Disorder seem to need more time to make decisions and show biased decision-making strategies (Adolphs et al., 1996; Bechara et al., 1998). Moreover, some relationships between the physiological response to stress and functions such as attention, executive functions, decision-making have been highlighted (Lebois et al., 2016). Symptoms' severity seems to be related to the individual tendency to suppress painful contents and mitigate planification strategies (Huys et al., 2012; Montague et al., 2012); deciding implies a simultaneous evaluation of present stimuli and possible choices (Aupperle et al., 2012; Palmiero et al., 2020; Gambetti et al., 2022) which subconsciously involves the information reduction while reasoning (Kahneman and Frederick, 2002; Shah and Oppenheimer, 2008; Evans and Stanovich, 2013), also affected by moods and emotions in the post-reasoning evaluations and therefore decision-making processes (Bower, 1981; Forgas, 1989; Schwarz and Clore, 1996; Yildiz and Eldeleklioglu, 2021). While considerable research has explored the physical and mental health consequences of GBV, there remains a paucity of studies specifically investigating the effects of violence on the decision-making processes of victims (Wiebe and Janssen, 2001; Leung et al., 2002; Pallitto et al., 2005; Bourassa and Bérubé, 2007; Ely and Otis, 2011; Roth et al., 2011; Lausi et al., 2023). The complex nature of decision-making within the context of GBV presents challenges in defining a cohesive framework for examination, as outcomes may vary considerably based on diverse factors, including the victim's perception of violence, control dynamics, and emotional attitudes. To advance our understanding, it is crucial to explore the interplay between GBV and decision-making processes comprehensively.

The present research seeks to obtain data which will help to address the research gap within experiencing GBV and the long-term effects of trauma on cognitive functions. More specifically, this paper aims to explore differences in emotional regulation, decision-making strategies, direct and indirect strategies of psychological abuse and anxiety, stress and depression between people who have experienced different types of violence and between people who have experienced different forms of abuse (Daigle et al., 2008; Walsh et al., 2020; Daigle and Hawk, 2022) and in assessing decision-making and risk-taking through the use of an implicit measure, in a sample of victims of abuse recruited from anti-violence centers and a control group. Since to the best of our knowledge, no previous studies investigated the differences among the different forms of abuse in cognitive decision-making and risk-taking, the analyses concerning the cognitive task are explorative only.

In particular, the following assumptions were made:

H1: There is a relationship between implicit measurements (cognitive task) and stress, anxiety, and depression; we expect that people with higher scores in the DASS will show a longer response time (Lebois et al., 2016; Lawlor et al., 2020).

H2: The implicit measurement of decision-making (measured by the Millisecond Gambling Task) may be predicted by the explicit variables (measured by the questionnaire) with differences among the no-violence, single-violence, and complex violence victims.

A non-probabilistic and convenience sampling was used, and a total of 83 participants joined the research. Inclusion criteria were being at least 18 years old, being a female and being Italian speaker (Table 1).

Data have been collected throughout Italy using the Qualtrics Platform for online surveys. The questionnaire was launched in January 2022 and data were collected until April 2022. All study procedures were carried out according to Helsinki Declaration and approved by the Institutional Review Board of the Department of Psychology, the University of Rome “Sapienza” with protocol number 0001446 (5/07/2021).

The present study employed a quantitative approach, through an online questionnaire, which was used to collect the target information, and a cognitive task used to collect implicit measures. The questionnaire was administered through Qualtrics platform; at the end of the questionnaire a short link redirected participants to the Inquisit web platform to perform the Millisecond gambling task. Both questionnaire and task were administered through computers and laptops.

The online survey was spread through various channels, such as working platforms, word of mouth and social networks, to reach as many participants as possible from the general population. Participants were recruited through a snowball sampling method: the participants from the researchers' networks referred to other people who accepted participating to the study.

The victims of violence (N = 56) were recruited through shelters and in addition, we made sure to contact women's anti-violence and victim support centers to promote the research and propose the participation of victims on a voluntary basis. This was done to ensure that the voices of those affected by violence were heard and to recognize and value the contribution of victims to the research. Furthermore, special efforts were made to ensure that the survey was accessible so that victims could participate without fear of retribution or discrimination. Two victims did not declare the kind of violence episodes (single or multiple violence); therefore, they were not considered for the final analysis, leading to a total number of 54 victims.

The Millisecond Gambling Task is a task developed to assess decision-making and risk-taking, adapted from the Cambridge Gambling Task (Rogers et al., 1999).

Participants are given 10 boxes which are red and blue on a different ratio (i.e., 1:9; 2:8; 3:7; 4:6) and told that a yellow token is hidden under one of these boxes. Once they decide on the color, participants are asked to bet points on their choice. The bets are either in ascending or descending order. If they win, the bet number is added to the total points, if they lose the number is taken away. The MGT can measure several aspects: risk-taking, represented by the betting ratio where participants choose the optimal color; the quality of the decision-making process, represented by the percentage of trials in which the participant chooses the color with the most boxes; the impulsiveness/aversion to delay, revealed by the choice of the percentages that are shown first deliberation time, given by the time between the start of the trial and the choice of the percentage; and risk adjustment, which captures the tendency of participants to bet more when the odds are in their favor (Zois et al., 2014; Yazdi et al., 2019; Romeu et al., 2020).

Statistical analyses were conducted using SPSS (Statistical Package for Social Science; version 27.0; IMB SPSS; Armonk, NY, USA). Descriptive analyses of sample characteristics were performed; then, the assumption for normality was checked for all the scales. Since the normality assumption was violated, a non-parametric One-Way ANOVA (Kruskal-Wallis) was performed among the three groups according to the self-reported type of the experienced form of abuse (i.e., no victimization, single victimization, complex victimization) (Daigle et al., 2008; Walsh et al., 2020; Daigle and Hawk, 2022). Statistical significance in the post-hoc analysis was determined using Dwass-Steel-Critchlow-Fligner (DSCF) pairwise comparisons and defined as p < 0.05. A Spearman non-parametric correlation was performed among the variables measured through the DASS subscales and the Millisecond Gambling Task outcomes. A linear regression analysis was performed among the three groups with Millisecond Gambling Task total point as dependent variable. The idea of including questionnaire scales as predictors is given by the possible variables that generally influence decision-making and risk-taking.

A non-parametric One-Way ANOVA was performed to reveal any statistically significant differences in Depression, Anxiety and Stress among victims of single and complex violence and non-victims. Results revealed no statistically significant differences in any of the subscales; the Dwass-Steel-Critchlow-Fligner (DSCF) pairwise comparisons showed that women who did not experience violence showed statistically lower scores in Stress (M = 10.9; SD = 7.26) than women who experienced single violence (M = 15.0; SD = 8.89) (Table 2).

A non-parametric One-Way ANOVA was then performed to reveal any statistically significant differences in Difficulties in Emotion Regulation subscales among victims of single and complex violence and non-victims. Results revealed significant differences in the non-Acceptance subscale. Through Dwass-Steel-Critchlow-Fligner (DSCF) pairwise comparisons differences among the three groups were investigated. Women who did not experience violence showed statistically lower scores (M = 9.04; SD = 4.31) than women who experienced both single violence (M = 12.3; SD = 5.77) and complex violence (M = 13.1; SD = 5.85) (Table 3).

A non-parametric One-Way ANOVA was performed to investigate differences between victims of single and complex violence and non-victims in the EAPA-P scale. Results showed no statistically significant differences in both Direct and Indirect strategies of psychological violence. The Dwass-Steel-Critchlow-Fligner (DSCF) pairwise comparisons showed differences among non-victims (M = 8.93; SD = 2.60) and complex violence victims (M = 11.3; SD = 5.15) in Direct strategies (Table 4).

A non-parametric One-Way ANOVA was performed to investigate differences among victims of single and complex violence and non-victims in the Melbourne Decision Making Questionnaire, showing no statistically significant differences in any of the subscales nor in the comparison among groups (Table 5).

According to the hypotheses, the Millisecond Gambling Task (MGT) scores were correlated with the subscales of the DASS. Positive correlations were found in the “meanrt_deliberation” scores, i.e., all those scores that measure the average reaction time taken to decide (where the numbers represent the ratio of the boxes), in both the stress and depression subscales (Table 6).

Three multiple logistic regression analyses were performed with the “totalpoint” of Millisecond Gambling Task as dependent variable in the three groups: no violence, single violence, and complex violence victims. The variables inserted as predictors were all the subscales of DASS, DERS-20, EAPA-P, and MDMQ.

Despite the considerations made so far, the research has some limitations that it is important to highlight. First, the sample size should be expanded to be able to make statistical inferences that can be discussed more reliably, even though victims come from shelters making the answers more accountable than a more shared online questionnaire. Replicating the research by collaborating with more shelters could allow the sample to be enlarged enough to consider conducting a mediation analysis to better understand the relationships between the factors and to include decision-making style as a possible variable in addition to decision-making strategies. Moreover, following the considerations that emerged from the study, one could include a specific screening for possible disorders (e.g., PTSD, substance abuse...) that may be closely linked to the variables examined, especially referred to Depression, Anxiety, Stress Scale symptoms. Finally, there is no detail on time differences between the forms of violence: this information, although important, was not included following the specific suggestion of the anti-violence centers that helped with data collection, as it is part of the sensitive information that is asked of the victims.

The limits of the present research allow us to identify which may be the future perspectives of the study and, more specifically, of cognitive research in victims of violence. First, the sample of the present study should be enlarged to carry out a more accurate investigation of the implicit variables examined. In addition, longitudinal studies should be carried out, starting from the present research, to allow a broader interpretation of the data, including the addition of a semi-structured interview, so that the individual experience of the victim is not lost when interpreting the quantitative data measuring common experience. In addition, the use of qualiquantitative data in an area that is still under-investigated within abusive contexts (i.e., cognitive processes) can be useful in the construction of training paths for those professionals working directly with victims but also in the development of evidence-based clinical practices that consider not only the short-term effects of gender-based violence but also the long-term effects.

Diversity factors can significantly impact the experiences of GBV victims. While the present sample primarily consisted of Italian-speaking females, it is important to acknowledge that the effects of GBV may vary across diverse cultural, linguistic, and socio-economic backgrounds. Future research should prioritize a more diverse participant pool to capture the experiences and consequences of GBV among various populations. Additionally, investigating the intersectionality of diversity factors, such as race, ethnicity, disabilities, and socio-economic status, is essential for understanding how these overlapping identities influence the overall wellbeing among victims.

This study set out to explore the influence of different forms of GBV experience on psychological wellbeing and cognitive outcomes in a sample of victims and non-victims. The data collected, although mainly with an exploratory purpose, allow several considerations to be made on the topic of gender-based violence. First, there are differences between those who have and have not experienced violence in their lifetime. These differences, however, do not follow specific patterns. It emerged that there are differences between those who have not experienced violence and those who have—both single and complex—but that there are no differences between groups of people who have experienced one or more forms of violence. Taken together, these results suggest that once a victim has experienced a traumatic event, such as one of the forms of abuse under investigation, they react not only in the immediate term but may develop coping strategies and forms of resilience that they will also use in the long term (McCann and Pearlman, 1990; Berkowitz, 1993; Gutner et al., 2006; Tjaden, 2009; Lawrence et al., 2012; Lausi et al., 2021b). However, these results must be approached with come caution to avoid a risk of “normalization” of the violence within an individual's life in the long term, a risk that could lead to a devaluation of the pain or even in extreme cases to the attribution of responsibility on the victim of complex violence (Humphrey, 2003; McCarry and Lombard, 2016).

Another aspect to be taken into consideration concerns, more generally, the interpretation of data. The risk, when we talk about gender-based violence and decision-making or risk-taking, is once again to find ourselves saying that the victim is such because she does not use the correct decision-making strategies or because she is “reckless”: studies in victims of violence, for obvious ethical reasons, cannot predict “pre” violence, so we cannot know the connection between the decision-making strategies before the traumatic event and the event itself. What we do know is that numerous studies have shown how, in situations of similar or even lower stress, coping strategies and decision-making and risk-taking patterns changed (Adolphs et al., 1996; Bechara et al., 1998; Huys et al., 2012; Montague et al., 2012; Lebois et al., 2016; Taylor et al., 2018; Minooee et al., 2020). And it is only in the “post-violence” phase that we are interested in knowing how victims behave, and what the cognitive effects of violence may be, not to judge but to be able to implement support programmes for victims that are effective and aimed at acquiring new skills in risk assessment, decision-making and emotional regulation.