Women are substantially more likely than men to be diagnosed with stress- and anxiety-related disorders, including post-traumatic stress disorder and generalized anxiety disorder (Kessler et al., 2005; McLean et al., 2011; Haering et al., 2024). These conditions are highly comorbid, share overlapping symptoms (Price and van Stolk-Cooke, 2015; Afzali et al., 2017), and rank among the most prevalent mental health challenges globally, contributing to significant societal and economic costs (Fineberg et al., 2013; Trautmann et al., 2016). Despite the clear sex disparity in prevalence, the underlying neurobiological mechanisms driving these differences remain poorly understood (Lebron-Milad and Milad, 2012; Day and Stevenson, 2020).

A core behavioral symptom and transdiagnostic feature across these disorders is maladaptive avoidance, where individuals excessively avoid situations or stimuli perceived as threatening (Dymond, 2019). Avoidance frequently arises from attempts to regulate intense negative emotions, including preventing contact with sources of distress by avoidance of external triggers and the avoidance of internal thoughts and feelings (Barlow, 2008; Ellard et al., 2010). Because avoidance can severely impair daily functioning, it is a key target for clinical intervention (Salters-Pedneault et al., 2004; Blakey and Abramowitz, 2016). Examining the basis of sex differences in the expression of avoidance, which allows for better understanding of the dysfunctional mechanisms underlying the mental disorders, is therefore a critical step toward developing more effective, personalized treatments.

Rodent models have been instrumental in elucidating the neural circuits of threat conditioning and avoidance (LeDoux, 2000; Maren, 2001; Bienvenu et al., 2021; Diehl et al., 2024), yet preclinical research has historically focused almost exclusively on male subjects (Beery and Zucker, 2011; Dalla and Shors, 2009; Shansky, 2018). This persistent exclusion of females has limited our ability to model the full spectrum of anxiety pathology and has created significant gaps in our knowledge of the female neurobehavioral response to threat (Shansky, 2019; Shansky and Murphy, 2021).

The limited studies that have included both sexes reveal important sexual dimorphisms. For instance, female mice can demonstrate more persistent avoidance following extinction training (Halcomb et al., 2024), and both female rats and mice often employ different defensive response strategies compared to males, such as active darting or “anxioescapic” behaviors instead of passive freezing (Gruene et al., 2015a; Shanazz et al., 2022; Halcomb et al., 2024). This divergence may reflect fundamentally different cognitive strategies for assessing risk. For example, in value-based decision-making, females are more likely to use a steady, long-term “less risky” strategy, whereas males often change their approach based on recent outcomes (Chen et al., 2021a, 2021b). Such findings suggest that sex-based variations could arise from sex-biased cognitive approaches (Grissom et al., 2024). Sex-biased behaviors are often assumed to arise from gonadal hormones, but hormonal contribution to avoidance-related phenomenas is not straightforward. For example, although hormonal fluctuations across the estrous cycle in females can modulate the consolidation of safety learning (Maeng et al., 2017) and increase vulnerability to deficits in contextual threat extinction in a pharmacological OCD model (Reimer et al., 2018), they show little effect on other behaviors. These include, for example, open-field exploration (Levy et al., 2023), and in interactions in the resident-intruder assay (Zeng et al., 2023).

A second challenge in existing literature is that it depends heavily on paradigms that may not model clinically relevant avoidance (Aupperle and Paulus, 2010; Kirlic et al., 2017; Ball and Gunaydin, 2022). Avoidance tasks are typically categorized as passive (withholding an action to avoid threat) or active (performing one). Focusing on the latter, many standard active avoidance paradigms, like the shuttle box, have translational limitations (Diehl et al., 2019). The shuttle avoidance (Mowrer and Lamoreaux, 1946; Krypotos et al., 2015), for example, requires an animal to repeatedly re-enter a compartment where it was just shocked. This creates a condition where there is no well-defined safe location, leading to an ambiguous response: it promotes freezing behavior that directly competes with, and can be mistaken for, the avoidance strategy being measured. Furthermore, these tasks typically omit any cost for the avoidant action, failing to model the clinically relevant situations where individuals sacrifice valuable opportunities by engaging in excessive, maladaptive avoidance that severely impairs their daily functioning (Aupperle et al., 2015; Pittig et al., 2021).

More recently, the Platform Mediated Avoidance (PMA; Bravo-Rivera et al., 2014) task was developed to address these issues, allowing for a less ambiguous behavioral response by providing a distinct safe zone while also incorporating a cost–benefit component. In PMA, animals may food-seek by pressing a lever, but to avoid a signaled foot shock, must retreat to an insulating platform far from the lever. However, this paradigm models a specific type of conflict where the threat and reward are separated in time; an animal can simply wait for the threat to pass before safely resuming reward-seeking. While this is a valuable model, it does not fully capture the decision-making characteristic of clinical anxiety, where valuable opportunities are contingent upon enduring a perceived threat. A more direct model requires forcing an explicit trade-off where the reward is only available in the presence of the threat. Investigating the potentially dimorphic strategies that emerge under such intense conditions - which can reflect differences in threat-response regulation, decision-making, or both (Orsini et al., 2016; Greiner et al., 2019; Xu et al., 2022) - is a crucial step toward understanding the neurobiological basis of avoidance circuits and how they may be altered in clinical populations (Olff, 2017; Panayiotou et al., 2017).

Therefore, the present study employed a high conflict PMA task similar to Bravo-Rivera et al. (2021), designed to create the type of intense approach-avoidance conflict needed to reveal these strategies. In our paradigm, animals could only access rewards when a threat was imminent, forcing a choice between reward-seeking and safety. Aside from having more face resemblance to clinical avoidance, this increased conflict may invoke more deliberative, value-based decision-making, which may be revealed in vicarious behaviors (Papale et al., 2012; Schmidt et al., 2013). We tracked the development of avoidance and approach behaviors over 25 days to examine how strategies were acquired and expressed once they stabilized, allowing us to compare both the learning dynamics and end-stage behavioral patterns between sexes.

Based on the clinical prevalence of anxiety disorders in women and preclinical evidence of dimorphic threat responses, we hypothesized that male and female rats would adopt distinct strategies. Specifically, we predicted that females would exhibit greater avoidance behavior, prioritizing safety at the cost of reward, whereas males would demonstrate more risk-insensitive reward-seeking. We also conducted exploratory analyses to determine whether behavioral strategies in females were modulated by the estrous cycle, given the putative influence of reproductive hormones on threat-detection / threat-response in rodents (Reimer et al., 2018; Maeng et al., 2017).

Confirming our initial hypothesis, the present study demonstrated robust and replicable sex differences in resolving approach-avoidance conflict, with females consistently prioritizing safety while males exhibit more persistent reward-seeking. These findings provide evidence for sex-specific behavioral strategies that align with and extend the existing literature on dimorphic threat responses and risk–reward preferences. This divergence seems consistent with other tasks where females often adopt stable, risk-averse strategies, while males are more influenced by recent outcomes and less deterred by negative consequences (Grissom and Reyes, 2019; Chen et al., 2021a; Grissom et al., 2024). Interpreting these results requires moving beyond a simple conclusion that females are “more responsive” to threats. Instead, as Shansky (2018) argues, these behavioral divergences likely represent qualitatively different, and potentially adaptive, strategies for navigating high-conflict situations.

A strength of the current study is the inclusion of a pre-registered replication cohort. While the significance of certain model predictors (e.g., body weight, handler) varied between cohorts - likely reflecting the sensitivity of behavioral tasks to environmental fluctuations - the core metrics defining the sexually dimorphic strategies demonstrated robust replicability. Specifically, the “safety-first” phenotype in females was confirmed across multiple dimensions: females spent significantly more time on the safety platform in both cohorts, males showed higher rates of lever-pressing under threat, and females evacuated the reward zone earlier in the threat window, as indicated by shorter latencies to reach safety. The convergence of these metrics across independent cohorts provides strong evidence that these sex differences reflect stable, biologically based differences in behavioral strategies.

Our results fit within a broader framework of active versus passive defensive behaviors. Preclinical literature consistently shows that when faced with a threat, males often adopt passive, immobile responses like freezing, whereas females tend to engage in active, mobile strategies (Dalla and Shors, 2009; Gruene et al., 2015a). This active strategy in females has been termed “anxioescapic behavior,” characterized by movement-based responses, such as darting or escape, rather than simple threat response potentiation (Shanazz et al., 2022; Bangasser and Cuarenta, 2021). The PMA task is exceptionally well-suited to capture this strategic divergence. By providing an unambiguous safe location, it allows for the expression of an active avoidance response that might be masked in paradigms that rely solely on freezing. The greater time females spent on the platform, therefore, likely reflects the expression of this coping strategy rather than a simple inability to approach the reward.

The high-conflict nature of our paradigm, where rewards were only available under imminent threat, is crucial for revealing these sex-specific priorities. Rather than simply reflecting greater anxiety, this strategic divide may represent a sex difference in decision-making, as females in both human and animal studies show a strong tendency to avoid options associated with frequent negative outcomes (van den Bos et al., 2013; Grissom and Reyes, 2019). This strategic divide is not unique to PMA and has been observed in other approach-avoidance models. For instance, in the Vogel conflict test, where water-deprived rats are punished with a shock for drinking, females also display greater behavioral inhibition, accepting significantly fewer punished responses than males (Basso et al., 2011). This parallel finding reinforces the idea that females, across different types of conflict, adopt a more cautious, safety-prioritized strategy. Notably, Basso et al. (2011) also reported a pharmacological dissociation: while classic benzodiazepines were anxiolytic in both sexes, other compounds like SSRIs and buspirone only produced anxiolytic-like effects in males. This pharmacological dissociation provides an important indication: while some core circuits for processing conflict may be shared between sexes, the systems that modulate these circuits and ultimately shape the behavioral strategy seem sexually dimorphic. This suggests that the decision-making process involves both shared and sex-specific aspects of neurocircuitry and circuit functioning, which may be differentially influenced by neuromodulatory systems. It is important to note that the “safety-first” strategy could partially reflect sex differences in pain sensitivity, as females generally exhibit lower thresholds than males (Mogil and Bailey, 2010; Sorge and Totsch, 2017).

While our avoidance/approach metrics differed between sexes, reward attentiveness showed no consistent sex difference across cohorts. This suggests that the strategic divergence may not lie in the amount of deliberation itself; both sexes appear to engage in a similar degree of conflict processing, weighing the value of the reward against the imminent threat. The difference, therefore, appears to be in the outcome of this deliberative process. Despite their engagement with the reward, females seem to more consistently suppress the final motor-approach response in favor of a safety-prioritized strategy. This dissociation between deliberation and action strongly supports the hypothesis that males and females adopt qualitatively different strategies for resolving the conflict (Shansky, 2018). Thus, the dimorphism observed here may not arise from the assessment of the conflict per se, but rather in the downstream decision-making circuits that translate the output of that assessment into a behavioral choice.

These distinct behavioral strategies likely arise from sexually dimorphic neural and endocrine mechanisms. The work of Bravo-Rivera et al. (2021), which characterized these behavioral phenotypes in rats, offers a potential neural basis for our findings. They demonstrated that an avoidance-preferring strategy, which we observed in females, was associated with heightened activity in the amygdala, a key hub for threat response. Other threat-related behaviors, such as conditioned freezing and escape, have been linked to amygdala and particularly its reciprocal interactions with prefrontal cortex (Diehl et al., 2024; Plas et al., 2024). Those circuits may also be dimorphic – one study suggested that females and males “solve” extinction learning differently, with females depending less on infralimbic cortex (Gruene et al., 2015b). Thus, identifying the neural basis of sexually dimorphic avoidance likely requires simultaneous measurement of multiple structures within the canonical fear circuitry.

Furthermore, gonadal hormones may profoundly shape these circuits. This shaping occurs not only through the immediate influence of circulating hormones in adulthood (activational effects) but also through the permanent organizing of neural pathways during critical developmental periods (organizational effects) (McCarthy et al., 2012, 2015). The influence of high estrogen levels, for instance, on facilitating the extinction of conditioned fear would be an activational effect (Graham and Milad, 2014; Ramikie and Ressler, 2018). However, findings from Halcomb et al. (2024) in a modified PMA task revealed that persistent avoidance in female mice was significantly reduced by a glucocorticoid receptor (GR) antagonist administered during learning. This suggests that GR activation during the acquisition of the threat association, rather than activational effects of circulating gonadal hormones alone, may be a key mechanism that defines and maintains this avoidance strategy in females.

While our study did not find a significant modulation by the estrous cycle in our exploratory analysis, the role of hormonal fluctuations remains an important consideration (Li and Graham, 2017). This is consistent with findings from both the modified PMA task (Halcomb et al., 2024) and the Vogel test (Basso et al., 2011), where the estrous cycle did not influence avoidance behavior. The influence of the estrous cycle can be, however, task-dependent, as Carvalho et al. (2021) found it impacted fear-potentiated startle but not conditioned freezing. Together, this suggests that while ovarian hormones are critical for the developmental organization of these sexually dimorphic circuits, the expression of a stable, learned behavioral strategy in a high-conflict task may be less sensitive to acute fluctuations than other measures of fear-like responses. However, it is important to note that our findings characterize the safety-first strategy specifically in nulliparous females. Literature on maternal behavior suggests that lactation and reproductive experience can induce a risk-prone phenotype, potentially to ensure resource acquisition for offspring (Ferreira et al., 1989; Faraji et al., 2024), suggesting that the safety-prioritization observed here is likely a state-dependent adaptation rather than an inherent sex difference.

A limitation of the current study is that the neurobiological mechanisms underlying these sex-specific strategies were not directly assessed. Future studies incorporating neural recordings and causal manipulations - particularly targeting oscillatory dynamics within amygdala-prefrontal circuitry, which has been implicated in both conflict processing and sexually dimorphic threat responses (Gruene et al., 2015b; Bravo-Rivera et al., 2021; Totty and Maren, 2022) - will be necessary to validate the proposed mechanisms and enhance translational relevance. A consideration inherent to conflict paradigms is that multiple decision-making components are integrated within the task, and the current design does not permit complete dissociation of all underlying processes. Nevertheless, our data suggest that the sex difference emerges specifically under conflict: males and females showed equivalent reward-seeking during the non-conflict Light Cue phase, similar reward attentiveness, and the statistical modeling revealed that sex predicts for how long animals will endure rather than whether they approach. Future studies utilizing shock titration to equate subjective cost across subjects, progressive ratio schedules, or sucrose preference assays could further isolate the specific contributions of punishment sensitivity, reward valuation, and risk tolerance.

In conclusion, we show that male and female rats employ fundamentally different strategies to resolve approach-avoidance conflict. Females adopt a safety-first strategy, consistent with a more active coping style that seems more resistant to change. Males, in contrast, employ a risk-prone, reward-oriented strategy. The translational importance of these findings is significant. The persistent avoidance shown by females mirrors a core clinical feature of anxiety and trauma-related disorders that is often a primary target for intervention. Understanding the distinct neurobiological pathways and mechanisms that support these strategies is a critical step toward developing more effective, personalized treatments. While acknowledging challenges in translating circuit-level findings across species (de Oliveira et al., 2021) the next critical step is to characterize how this circuit activity differs between sexes. This characterization may provide the necessary biomarkers for developing targeted neuromodulation therapies, such as deep brain stimulation, to ameliorate persistent avoidance.