The Loss of Autonomy in Abused Persons: Psychological, Moral, and Legal Dimensions

: This paper tries to resolve a tension in popular conceptions of intimate partner violence (IPV). On the one hand, we correctly assume that all abused persons are not the same: they have irreducibly plural personalities. On the other hand, we correctly assume that abused persons suffer from a loss of autonomy. The puzzle is: if abused persons share deﬁcits in autonomy, why does it not follow that they share a set of personality traits? I argue that the psychological states implicated in autonomy-impairment in abused persons are situation-sensitive responses to salient eliciting conditions, not personality traits. This view has substantive moral and legal implications, as it implies that abusers are responsible for inﬂicting severe moral harms on victim-survivors, and they may also be liable for unlawful abduction and rape, in case the abused person lives with or has sexual contact with the abuser. This is because the conditions of abuse undermine the victim-survivor’s ability to autonomously consent to cohabitation and sexual contact with the abuser. I argue that the best way of protecting people from autonomy-undermining abuse is public education.


Tension: Psychological Pluralism versus Shared Autonomy Deficits
Amy Schumer's recent autobiography, 'Girl with the Lower Back Tattoo' (2016), is a mix of light-hearted anecdotes and narratives of personal struggle. The latter category includes a chapter on Schumer's experience as a victim-survivor of intimate partner violence (IPV). She writes about being threatened, harassed, confined against her will, and subjected to violence by her intimate partner of almost a year. At one point, she "was sure he was going to kill [her]" (Schumer 2016, p. 300). One of her reasons for sharing this narrative is to debunk the cultural myth that only 'certain kinds of women'-those who fit dominant cultural paradigms of 'victimhood' and 'femininity'-can be abused by intimate partners, and only such people can provide credible testimony about IPV. In defiance of this cultural stereotype, Schumer says, I'm a strong-ass woman, not someone most people picture when they think 'abused woman' . . . It can happen to anyone. You're not alone if it's happening to you, and you're not exempt if it hasn't happened to you (Schumer 2016, p. 300).
Schumer's testimony is a useful contribution to public discourse because it rebukes cultural stereotypes about whose narratives are epistemically valuable. Her narrative disrupts the "history of silence" imposed on women by traditional binary exclusions and hierarchies of respect (Schumer 2016, p. 17). This helps reduce what Miranda Fricker refers to as "hermeneutical gaps" in our shared interpretive resources (Fricker 2007, p. 160)-resources that favor the situated perspectives and shared values of cisgender white males. As such, Schumer's work can be seen as an applied piece of social-feminist epistemology, something that 'troubles' cultural understandings of IPV and recognizes a plurality of voices. While Schumer's narrative is undoubtedly epistemically valuable, the claim that 'anyone can be a victim of IPV' is not uncontroversial. Schumer seems to be saying that abused women 1 do not share individuating personality traits that predict vulnerability to sexual violence. This is not to deny that they share common experiences, situated insights, or psychological injuries (such as trauma); but these experiences are distinct from personality traits, which are relatively situation-invariant patterns of thought and behavior that define people as distinct individuals. Gilbert Harman defines a personality trait as a "robust disposition . . . that would manifest [itself] also in counterfactual situations," in contrast to 'shallow' behaviors that are elicited by situational factors (Harman 2009, p. 235, cf. Harman 1999. Personality traits are also distinct from psychological disorders, which may be situation-invariant, but are not features of an agent's personality or character (ibid.). Personality traits, in this view, are relatively deep features of an agent's personality, as opposed to shallow responses and/or superficial (non-individuating) psychological structures.
The image of victim-survivors as people with shared experiences, but no shared personality traits in the operative sense, calls for a psychological explanation-specifically, an explanation that respects the irreducible characterological plurality of victim-survivors, but recognizes that they share a set of autonomy-undermining psychological states. How can we distinguish the states constitutive of autonomy deficits from the personality traits of abused persons? What superficial states are incurred by IPV?
It is not too contentious to think that abuse causes autonomy deficits. Marilyn Friedman provides a philosophical account of this phenomenon. She defines 'autonomy' as the ability to reflect on one's "deeper values and concerns and act on them" (Friedman 2003, p. 141). IPV undermines the abused person's autonomy (so defined) in three senses: (1) it threatens the abused person's survival and safety, goals that the person would pursue under less oppressive conditions; (2) it focuses the abused person's attention on the interests of the abuser, preventing her from pursuing the basic human goals of survival, safety, self-actualization, and wellbeing; and (3) it submits the abused person to the will of the abuser, causing her to prioritize his goals above her own basic human needs (Friedman 2003, p. 141).
The operative notion of autonomy is first and foremost the ability to preserve one's own survival and safety, and secondly the ability to pursue the primal goals of self-actualization and wellbeing. Conditions (2) and (3) are autonomy-undermining in part because they undermine condition (1), and in part because they undermine the abused person's ability to pursue other basic interests, such as minimal self-actualization and wellbeing. In what follows, I will use 'autonomy' to denote the ability to pursue the basic human goals of survival, safety, self-actualization, and wellbeing, piggy-backing on Friedman's definition (similar definitions of autonomy appear in Raz 1986;Oshana 2006;and Maslow 1943). This provides an account of the 'shallow autonomy' possessed by ordinary democratic citizens, not the 'deep autonomy' that may come with absolute freedom from coercion and robust social, political, and epistemic support (Sneddon 2001).
Friedman qualifies her view by stipulating that autonomy is merely undermined by IPV, not extinguished by it; abused persons retain autonomy over some aspects of their lives, presumably outside of the abusive relationship. Furthermore, the loss of autonomy induced by IPV is short-term and not permanent (Friedman 2003, p. 150). I agree with these stipulations, and I seek to accommodate them in my psychological model of impaired autonomy in victim-survivors. Specifically, I aim to provide a psychological model of loss of autonomy that accommodates Friedman's and Schumer's descriptions of abused persons.
The puzzle that I try to solve is how IPV victim-survivors can share diminished autonomy without sharing individuating personality traits. Diminished autonomy is a psychological phenomenon triggered 1 In this paper, I will use feminine pronouns (she/her) to refer to victim-survivors, but I acknowledge here that men can also be victim-survivors of IPV. That said, women comprise the majority of victim-survivors, and suffer worse abuse than men (e.g., homicide). This is why theorists like Ann Cudd describe rape as a crime against women (Cudd 2006). and sustained by external factors of abuse and coercion. If these psychological mechanisms are shared by IPV victim-survivors, how do they not also share personality traits? Pervasive stereotypes may lead one to believe that people with diminished autonomy have 'passive' personality traits-that they are 'passive victims' (Mills 2006). An adequate psychological model must reject this faulty logic. It must capture the irreducible tension between the loss of autonomy caused by IPV, and the psychological pluralism and latent autonomy shared by victim-survivors of trauma.
In what follows, I advance a psychological theory that accommodates these tensions. Specifically, I argue that IPV imposes short-term, situation-sensitive, autonomy-undermining psychological states in abused persons, which are not caused or constituted by personality traits. Autonomy-undermining states are situation-sensitive responses that mediate between the eliciting condition and overt behavioral responses. This account provides a psychological basis for the claim that the loss of autonomy in victim-survivors is temporary, shallow, and domain-specific, and it rejects blanket statements about the personalities of abused persons. It also explains how abused persons can regain lost autonomy under less oppressive conditions and can exercise a greater degree of autonomy in non-abusive contexts, where triggers are not salient. This model, then, explains why we cannot identify abused persons on the basis of simplistic cultural stereotypes.
Notably, this psychological model has substantive moral and legal implications. It implies that victim-survivors are not blameworthy for their plight, but (I shall argue) abusers are blameworthy for a severe moral harm-undermining the autonomy of another person through the use of violence. Abusers are blameworthy, in my view, because their abusive actions are performed in autonomy-conducive conditions. Abusers, furthermore, can and should be held liable for sexual assault and abduction insofar as they undermine the abused person's ability to voluntarily consent to cohabite and have sex with them. Thus, living with and having sex with an abused person under autonomy-undermining conditions created and sustained by the abuser may be criminal as well as immoral.
In my view, the mental states constitutive of impaired autonomy (as defined by Friedman) include: (1) low self-efficacy, (2) learned helplessness, (3) obedience to authority/foot in the door effect, (4) social-role conformity, and (5) reward-seeking behavior/dependency. This is not an exhaustive list of autonomy-impairing states (which would surpass the scope of a single article), but it provides a set of prototypical autonomy-undermining states revealed by well-known social psychology studies. This same research shows that the prototypical activating conditions for these states include: (i) escalation of demands and violence, (ii) social isolation, (iii) imposed anxiety and disorientation, (iv) putative asymmetrical relations of power and authority, and (iii) a variable ratio reward schedule. These eliciting conditions are common to some of the most psychologically damaging and morally problematic social psychology experiments (many of which cannot be performed on human subjects) and abusive intimate partnerships. In both cases, these conditions impair the autonomy of the agent on whom they are inflicted. Thus, their use outside of experimental contexts-and in some cases, also within experimental contexts-is grossly unethical, and potentially criminal.
Because the activated states are, according to the research, relatively situation-sensitive, they are not incorporated into the personality structures of the agent in whom they are induced. Even if these states are not directly cognitively penetrable, they are nonetheless highly amenable to extinction under non-coercive conditions, which is why they are superficial response patterns as opposed to situation-invariant traits (more on which in Section 2).
My argument will proceed along the following lines. In Section 2, I will explain what I mean by 'IPV,' 'victim-survivor,' and 'non-coercive conditions'; in Section 3, I will identify the autonomy-undermining states reliably induced in victim-survivors by abusive activating conditions, based on relevant social psychology research; in Section 4, I will define the salient activating conditions for these states, and show that they are common to situation psychology experiments and (a subset of) paradigmatic cases of IPV; in Section 4, I will argue that abusers, but not abused persons, are responsible for the moral harms caused by the abusive relationship; in Section 5, I will argue that abusers may also be liable for charges of criminal abduction and rape; and in Section 6, I will argue that educating people about IPV, including its salient features and psychological effects, may be the best way of empowering people to protect themselves and others from abuse. That said, introducing criminal sanctions against abusers could help to empower victim-survivors to preserve and express their autonomy through legal means, should they so choose.

IPV
For simplicity and clarity, I will define intimate partner violence by reference to the guidelines offered by the Centre for Disease Control (CDC) and the World Health Organization (WHO). According to the CDC definition, IPV has two main elements. First, it is perpetrated by an intimate partner, or someone with whom one shares emotional connectedness, on-going physical contact, identity as a couple, and familiarity and knowledge about each other's lives. Second, IPV involves one or more of the following types of violence: (1) physical violence, (2) sexual violence, (3) stalking, and (4) psychological aggression. Psychological aggression is a broad range of abuses that include: expressive aggression (e.g., name-calling, humiliating); coercive control (e.g., limiting access to transportation, money, friends, and family; excessive monitoring of whereabouts); threats of physical or sexual violence; control of reproductive or sexual health (e.g., refusal to use birth control; coerced pregnancy termination); exploitation of victim's vulnerability (e.g., immigration status, disability); exploitation of perpetrator's vulnerability; and presenting false information to the victim with the intent of making them doubt their own memory or perception (e.g., mind games) (Centre for Disease Control & Prevention 2016).
WHO includes three additional characteristic features of IPV: (1) escalation of violence severity, (2) imposed isolation, and (3) the perception of "love and the hope that the partner will change" (World Health Organization 2012). I will include these features in my working definition of IPV, with the caveat that (3) is by no means necessary for IPV. That said, the perception of love can make leaving an abuser particularly psychologically difficult, even in the absence of material constraints such as poverty and lack of safe housing. Since (3) is a psychological impediment to self-actualization and self-preservation, and I am concerned with psychological obstacles to robust autonomy, I will include (3) in my definition, noting that this somewhat artificially limits the scope of my analysis. A truly comprehensive analysis would consider the role of socioeconomic oppression in IPV, as this is perhaps the primary cause of loss of autonomy in IPV situations. That this is a central cause is well-known, and explains, for example, why police-reported IPV is 2-3 times higher for Black and Hispanic women compared to white women (Lipsky et al. 2009); racialized minorities face social adversity that impairs their ability to safely leave an abusive partner and achieve self-sufficiency. That said, even when material constraints limit a person's freedom, strictly psychological impediments caused by IPV may also the prevent the person from pursuing more autonomy-conducive conditions. Therefore, a psychological framework for understanding the loss of autonomy due to IPV is compatible with a sociological framework for understanding the socioeconomic bases of IPV.
In sum, I will address cases of IPV that include: an abusive person who uses physical violence, sexual violence, stalking, and/or psychological aggression, as well as escalation of violence and coercive demands, isolation, and the superficial appearance of love and concern, in ways that harm their partner. Again, these are not necessary and sufficient conditions of IPV, but common features that pick out an easily identifiable type of abuse-one that predictably induces autonomy-undermining psychological states in abused persons.

Victim-Survivor
The use of the terms 'survivor, ' 'victim,' and 'victim-survivor' are contested in feminist philosophy (e.g., Nissim-Sabat 2009; Heberle and Grace 2009). In the popular imagination, victimhood tends to be associated with traits of passivity and helplessness, while survivorhood tends to be associated with traits of resilience and resistance. In the present proposal, abused persons have the same latent autonomy as anyone else, but their ability to exercise this capacity is suppressed by salient eliciting conditions (as I shall argue). As such, I will use the term 'victim-survivor' to capture the inherently ambiguous nature of the abused person's agency-neither completely extinguished, nor as robust as it could be.

Non-Coercive Conditions
In the introduction, I said that victim-survivors suffer an incomplete, domain-specific loss of autonomy, which is susceptible to extinction in 'non-coercive conditions.' Autonomy in Friedman's sense is ostensibly an evolutionarily deep, genetically human capacity to pursue the primal needs of survival, self-protection, and self-actualization. As social animals, however, human beings cannot secure these ends without healthy (non-abusive) social relationships. Thus, autonomy should be seen through a contemporary feminist-philosophical lens, as an inherently relational capacity-one that depends on non-abusive social relationships. Hence, when I say that autonomy deficits are remediated in 'non-coercive conditions,' I do not mean just any non-violent ecological conditions, but rather, social conditions enjoyed by ordinary citizens of liberal democracies, who participate in social networks, epistemic communities, and democratic institutions. These conditions provide the materials required, as Susan Brison puts it, to 're-make' an autonomous self (2002).
Brison relates her experience of recovering from the loss of autonomy imposed by post-traumatic stress disorder (PTSD) in the aftermath of rape by "caring for and being cared for by others" (Brison 2002, p. 47), and "connect[ing] to humanity in ways [she] value[d]" (Brison 2002, p. 46). Like Brison, I see autonomy as relational and thus enabled by social scaffolding, particularly healthy relationships, medical resources such as therapy and peer support groups, and laws and policies that empower survivors to move freely in the world, according to their conception of the good. When I say that autonomy is activated or enhanced in 'ordinary social conditions,' I mean conditions that include these agency-enabling resources, e.g., friendship, information, peer support, and political freedom.
Of course, some people do not have access to these resources because their societies are not non-coercive. Indeed, some societies are macrocosms of an abusive intimate partnership, characterized by male domination, misogyny, gaslighting, torture, and systemic oppression. Women living in authoritarian regimes that sanction rape may never realize even 'shallow autonomy,' either because they cannot frame IPV as wrong due to political oppression, or because they are cognizant of their plight, but they lack the social, political, and legal resources needed to resist systemic oppression. Some people living in liberal democracies, too, may lack access to agency-enabling resources because they are victims of sex trafficking, forced marriage, or other oppressive circumstances that resemble, or include, IPV. 2 Fortunately, however, most democratic citizens do have access to self-constituting resources, though these resources can be blocked by features of IPV, such as violence, social isolation, and imposed distress. Removing these obstacles provides access to the conditions that foster basic autonomy.
Childhood abuse is another contested area of inquiry. Whether childhood abuse victims have the capability to acquire autonomy in the first place is a substantive empirical question. Research indicates that complex trauma in children can cause "lifelong problems that place them at risk for additional trauma exposure and cumulative impairment" (Cook et al. 2017), which might be taken to support the pessimistic conclusion that childhood trauma survivors cannot develop robust autonomy under any circumstance. However, research on treatment outcomes does not support this conclusion. On the contrary, research shows that trauma-focused treatments are effective at treating both adult and childhood complex sexual trauma and complex PTSD, and some studies have found no 2 Victims of sex-trafficking and other social injustices can also achieve autonomy if the abusive relationships, and additional constraints, are removed. Thus, the removal of the abusive triggers is part of the liberation of oppressed groups, along with social and legal reform. difference in treatment response between those with and without childhood trauma (Walter et al. 2014;Wagenmans et al. 2018). We cannot assume, then, that victim-survivors of childhood trauma do not have the capacity to develop basic autonomy. Perhaps some cannot, but the fact that someone is a survivor of autonomy-impairing childhood trauma does not support the inference that recovery is impossible for that person. As Martha Nussbaum cautions (1992), we should be wary of presumptively discounting the latent human capabilities of any group, particularly culturally marginalized groups whose members are systemically underestimated and discredited by epistemic injustice.

A Social Psychology Model of Impaired Autonomy
What happens when someone is subjected to coercive violence? More specifically, how does the experience of intimate partner violence impair the abused person's autonomy?
According to the social psychology literature, certain kinds of aversive stimuli reliably produce states that are 'autonomy-impairing' in Friedman's sense, viz., they prevent the agent from acting in self-preserving and self-actualizing ways. In domestic contexts, where there is virtually no limit on the use of 'aversive treatment' (including severe violence and murder threats), activating conditions for loss of autonomy may be severe, constant, and virtually inescapable. Even when the victim-survivor is not in the presence of the abuser, she may experience hyper-focus on the person, which is a normal response to unpleasant stimuli (Hollins et al. 2009), and this can limit her autonomy even in non-domestic contexts. Thus, the autonomy-undermining effects of salient activating conditions will be much more severe in victim-survivors of IPV than in subjects of regulated scientific experiments. Many of the experiments that we will be considering are unreproducible in modern times due to ethical concerns, and others were only permitted to use non-human animal subject because using the adverse stimuli on human subjects was deemed unethical. The use of these stimuli on human subjects in unregulated contexts, then, is patently morally problematic.
In this section, I will identify autonomy-undermining states caused by eliciting conditions common to social psychology experiments and IPV; namely, (i) low self-efficacy, (ii) learned helplessness, (iii) obedience to authority, (iv) social-role conformity, and reward-seeking behavior/dependence. These states can be induced in any sentient creature (not just human beings) using activating triggers such as violence, escalating demands and violence, imposed isolation, and anxiety-inducing prompts. In some theories, the states induced by these conditions are instrumental coping mechanisms that provide adaptive advantages in stressful situations (Fogle 1978). But these states are maladaptive in non-coercive conditions, and can undermine the agent's ability to cultivate robust autonomy, i.e., the ability to act on her values and interests across a range of circumstances. Thus, in a pluralistic, democratic society, which provides access to a range of circumstances in which autonomy would be adaptive, these states are ultimately more of a liability than an adaptive advantage.

Low Self-Efficacy
Self-efficacy is one's sense of personal competency with respect to a particular task. If the task is protecting oneself from abuse, then low self-efficacy may impair the abused persons' ability to pursue this task and secure a safe and self-actualizing environment. Low self-efficacy may therefore be implicated in the loss of autonomy induced by IPV.
Research shows that low self-efficacy can impair one's ability to persevere in the face of obstacles, cope with environmental stress (Bandura 2006, p. 309), and deal with trauma-related distress and negative moods (Benight and Bandura 2004). This supports the view that low self-efficacy is a factor in the loss of autonomy. That said, interventions designed to enhance self-efficacy tend to improve recovery outcomes in survivors of violence (Froeschle 2009), suggesting that low self-efficacy is a situation-sensitive response to abuse.
Consistent with this view, social cognition theory describes self-efficacy precisely as a situation-sensitive state that mediates between situational triggers and overt behavior, in contrast to a stable personality trait, which is defined within this paradigm as a relatively situation-invariant disposition (Bandura 1997(Bandura , 2006. The social cognition model is supported by research showing that, e.g., low self-efficacy is easily induced using non-conscious primes. In one study, older employees primed with negative stereotypes of aging performed worse on mail-sorting tasks than controls, exhibiting low 'mail-sorting self-efficacy' (Kirchner et al. 2017;cf. Kirchner et al. 2015). Conversely, self-efficacy can be enhanced by certain conditions, e.g., subjects who participate in highly interdependent groups show increased self-efficacy, but their self-efficacy later returns to baseline (i.e., the pre-experimental level) (Kerr and Hertel 2011). Thus, self-efficacy is highly sensitive to activating conditions, including non-conscious stimuli that are first-personally inaccessible. This is relevant because features of IPV can be subtle and difficult to identify in the initial stages of a relationship (without proper training).
Unlike social psychological experiments, in which participation is voluntary and highly regulated, IPV can be severe, chronic, and inescapable, particularly in relationships involving imposed isolation, stalking, and threats. Thus, the loss of self-efficacy in violent partnerships is likely to be far more intense than its counterpart in approved scientific studies.

Learned Helplessness
Learned helplessness, like low self-efficacy, is a characteristic psychological response to triggers. Early research on learned helplessness placed laboratory animals (because it was unethical to use human subjects) in a confined space and submitted them to painful electric shocks; then the animals were placed in a second enclosure with an open door, where they were again shocked (Seligman 1972). The animals that had been prevented from escaping in the first trial did not try to escape in the second trial, even though escape was physically possible, demonstrating 'learned helplessness.' When the experimenters later helped the animals escape by picking them up and moving their legs, they regained the ability to avoid the shocks (Seligman 1972). Arguably, these experiments were unethical even though they used non-human animals (specifically, dogs), and should never be replicated with sentient creatures. That said, they lend support to the view that learned helplessness is not a fixed trait; it is easily induced in sentient animals by painful stimuli, and is readily 'unlearned' in non-coercive conditions. While the absence of the trigger is sufficient to extinguish learned helpless over time, the introduction of rehabilitative resources expedites this process, restoring the agent's self-protective reflex in a matter of minutes.
Learned helplessness experiments involve eliciting conditions similar to IPV, particularly painful abuse and imposed isolation. Victim-survivors are often isolated in the shared home and subjected to repeated physical and/or psychological violence. There is evidence that victim-survivors exhibit emotional numbing and passivity characteristically associated with learned helplessness (Peterson and Seligman 1983). This suggests that learned helplessness may be a salient factor in loss of autonomy due to IPV, but in the absence of the abusive stimuli, the person's autonomy should return to its pre-abuse baseline. With rehabilitative resources, self-efficacy could even surpass previous levels.
It should be noted that some theorists take learned helplessness models of IPV to be disparaging to victim-survivors, because they ostensibly depict them as 'mere passive victims' of violence (e.g., Gondolf and Fisher 1988;Mills 2006). The social cognition model of learned helplessness avoids this criticism, as it refuses to attribute a 'victim personality' to abused persons; rather, it frames learned helplessness as a context-sensitive response to IPV that can be induced in anyone (even non-human animals) under salient activating conditions. The objection to the learned helplessness model, by contrast, fails to explain the loss of autonomy caused by IPV, which requires some psychological basis. The social cognition model succeeds in explaining the autonomy deficits imposed by violent abuse, but avoids the unsavory implication that the abused person has a 'passive personality,' unlike other people who have an 'active' and 'resilient' personality profile.

Obedience to Perceived Authority, Foot-in-the-Door Phenomenon (Escalation)
Obedience to authority was famously documented by Stanley Milgram, who designed a set of experiments to explain how the Nazi Party was able to co-opt so many ordinary German civilians to collaborate in the Holocaust. In a series of experiments, Milgram (1965) instructed subjects to deliver (simulated) electric shocks to a mock learner (really a confederate), gradually increasing the intensity (in 15-volt increments) from 15 volts to 450 volts. At 150 volts, the mock-learner shouted and pleaded to be let out, and at 330 volts, he stopped responding and seemed to fall unconscious. Thinking that the mock learner was a fellow participant, most subjects (65%) nonetheless continued to shock the person to the 450-volt mark, the hypothetical point of death (1963). Even though their participation went against their deep-seated values, they were unable to defect from the mock-learner's instructions.
The high compliance rate is especially surprising given that the experimenter was allowed to use only four verbal prods: (1) 'Please continue,' (2) 'The experiment requires that you continue,' (3) 'It is absolutely essential that you continue,' and (4) 'You have no other choice; you must go on. ' This presents what appears to be quite a low threshold of pressure-much less than the amount that would ordinarily be labeled as 'coercion.' For this reason, a group of experts surveyed prior to the experiment predicted that all subjects would defect before the 150-volt mark. They were shocked when most continued to the end.
Although some subjects defected (35%), there was surprisingly little variation amongst the group, suggesting that compliance was a normal psychological response to the salient features of the experiment, not a feature of the subjects' personalities. On scrutiny, 'salient features of the situation' include: (i) escalating demands, (ii) forced isolation, (iii) imposed anxiety and distress, and (iv) violence. The escalating prompts constitute (i) escalating demands for compliance, issued under pressure. In alternate experimental paradigms in which subjects were allowed to communicate with other people, full compliance dropped to a meagre 10% (Myers 2010, p. 202), showing that (ii) isolation is also motivationally significant. Experimental subjects, furthermore, displayed visible distress, including sweating, trembling, and moaning, even as they continued to shock the mock learner (Myers 2010, p. 197;Marcus 1974), showing that (iii) imposed anxiety and distress motivated compliance. Some theorists have suggested that induced distress may have 'swamped' the subject's decision-making capabilities, preventing them from acting on their evaluative judgments (Doris 1998), and something similar might explain the loss of autonomy in victim-survivors of IPV. Finally, (iv) violence is present in both contexts-in the violent shocks, isolation, and pressure administered in the Milgram experiments, and in the similarly coercive features of IPV.
It should be noted that variations of the original experiment produced very different compliance rates. For example, full compliance fell to 21% when subjects received prods by telephone as opposed to in person, i.e., when physical proximity was reduced. Full compliance also dropped when the experimenter was replaced by a less authoritative 'clerk' (actually a second confederate) (20%), when the experiment was moved from Princeton to a less authoritative lab in Bridgeport, Connecticut (48%), and when two co-instructors (actually confederates) revolted against the lead experimenter (10%) (Myers 2010, p. 209). This does not challenge the claim that compliance is a situation-sensitive state; rather, it shows that compliance is sensitive to a multitude of situational variables, some of which enhance compliance and others of which enhance autonomy.
Violent intimate partnerships most closely resemble the original (and most coercive) experimental paradigm (1963), which produced the highest compliance rate. IPV conditions involve not only (i) escalating demands, (ii) forced isolation, (iii) imposed anxiety and distress, and (iv) violence, but also (v) close physical proximity and trust between the abuser and the abused person, (vi) the perception of authority and credibility fostered and exploited by the abuser, and (vii) imposed isolation and restricted access to the outside world. Thus, mediating factors that reduce compliance in later Milgram experiments are not salient in abusive intimate partnerships, whereas mediating factors that enhance compliance are multiply salient. IPV, then, bears striking similarities to the high-compliance version of the obedience experiment, and predicts an acute loss of autonomy in abused persons.
Similar compliance effects are induced by a popular advertising strategy called foot-in-the-door phenomenon, which uses a gradual escalation of demands to induce compliance in potential customers. Researchers have studied this effect by asking a group of subjects to place a large, unattractive 'Drive Carefully' billboard in their front yard, to which most predictably said 'no,' and then asking a second group to place a small 'please drive carefully' sign in their front window, to which most predictably said 'yes'; then, two weeks later, the experimenters returned to ask the second group to install the ugly billboard, and most consented (Freedman and Fraser 1966). This shows that smaller demands prime people to acquiesce to larger demands, to which they would not otherwise consent. Abusers often make increasingly imposing demands on their partner, asking for bigger and bigger sacrifices (of time, attention, and money). Thus, foot-in-the-door phenomenon, like Milgram's escalating prods, could induce compliance in victim-survivors. Since this psychological technique is not patently coercive, it may be difficult to recognize as a compliance-induction technique without prior theoretical knowledge. Yet, precisely because it is so insidious, it can be a particularly effective tool of persuasion.
People who isolate, intimidate, and abuse their partners use Milgram-style coercion to induce compliance in their partner, and they sometimes use these inducements to prevent their partner from contacting the police, giving a deposition, and pursuing litigation. Friedman controversially argues that, because some abused persons cannot exercise their autonomy to report their abusers to the police, we should support "mandated legal procedures" that force the person to participate in an investigation (Friedman 2003, p. 148). I am skeptical of this conclusion, though I agree with Friedman's claim that IPV can undermine a person's ability to report abuse when she has good reason to do so; yet, in some cases, there are valid reasons not to involve law enforcement officials.
To give a few examples: police brutality and housing discrimination are real threats to Women of Color, deportation and separation from children is a valid concern for immigrants and undocumented women, and lack of safe accommodations is a valid worry amongst socioeconomically disadvantaged persons (WOCN 2017). The fact that socially marginalized groups face discrimination by the state implies that they state might not be in a position to restore their autonomy, because state interventions can be autonomy-undermining. For this reason, non-coercive initiatives, like education, are preferable to mandatory interventions by the state. But I will return to this complicated issue in Section 6.

Social Role Conformity
Social-role conformity was demonstrated by Philip Zimbardo in the 'Stanford Prison Experiment' (SPE)-an experiment that was terminated early, and has never been reproduced in an academic context, because it unexpectedly violated research ethics guidelines (Zimbardo et al. 1972). I call the main effect of the experiment 'social role conformity' because the experimental design induced subjects to "conform" to their assigned "social role," which was either dominant (the guard) or subordinate (the prisoner) (Bleske-Rechek 2001). In IPV contexts, abusers similarly assume a 'dominant,' abusive role, and victim-survivors are assigned to a 'subordinate,' oppressed role.
In 1971, Zimbardo recruited 24 psychologically normal male subjects and assigned half of them to be guards and the other half to be prisoners, to simulate a normal American prison. He then instructed the guards to try to instill a sense of powerlessness in the prisoners, stopping short of physically harming them or denying them food and drink, to replicate the conventions of a normal U.S. prison. Zimbardo's subjects were then confined to a mock-prison in the basement of Stanford University, and barred from communicating with outsiders until the end of the experiment (with the exception of one supervised visit).
By the end of the week, the guards were abusing the prisoners, subjecting them to forced exercise, barring them from using the toilet, depriving them of sleep, placing them in solitary confinement, force-feeding them, and subjecting them to sexual humiliation, which prompted Zimbardo to end the study early. He explained the surprising acceleration of abuse by reference to the 'salient features of the situation,' including forced isolation, enforced asymmetries of power and authority, sleep-derivation, and imposed stress, anxiety, and disorientation. Zimbardo later argued that similar conditions gave rise to analogous abuses perpetrated in Abu Ghraib Prison, though those abuses escalated to human rights violations due to a complete lack of oversight or regulatory initiatives on the part of the Bush Administration (which had previously licensed 'soft torture') (Zimbardo 2007). In Abu Ghraib, these abuses continued unabated for years.
Similar conditions can be seen in IPV contexts, particularly: (i) an escalation of violence and control, (ii) the production of disorientation, anxiety, and distress in the abused person; (iii) an asymmetry of power created by the abuser's violent and demeaning treatment (sometimes reinforced by the illusion of male authority created by patriarchy), and (iv) social isolation imposed on the victim-survivor. These conditions undermine the abused person's autonomy, making it difficult for her to pursue safe and self-actualizing projects.
While abused persons have a role in the relationship similar to the prisoners in the SEP, abusers are not analogous to the guards, because they are not subject to SEP-type constraints (imposed isolation, violence, distress). They are closer to the non-coerced position of the experimenter who administrates the experiment (Zimbardo), rather than the guards who are thrust into it. While patriarchy admittedly influences men's psyches, these influences are not equivalent to the violent conditions inflicted on subjects of unreproducible social psychological experiments and experiments performed on non-human animals, which preclude access to alternative perspectives. Consider that a small minority of American men are violent intimate partners, whereas virtually all abused persons lose autonomy as a result of violent treatment-this says something about the uniquely coercive force of IPV-type triggers. I will discuss this asymmetry between abusers and abused persons further in Section 4.

Reward-Seeking Responses to Variable Ratio Reinforcement Schedules
Variable ratio reinforcement schedules are schedules that offer rewards after a variable or unpredictable number of reward-seeking behaviors. This schedule induces a strong reward-seeking response in the subject, because variable rewards trigger the release of dopamine, which strengthens reward-seeking urges (Winstanley et al. 2011). A variable schedule produces the highest reward-seeking response rate and is more resistant to extinction than other schedules (e.g., fixed ratio schedule). This is why slot machines and video games are programmed on a variable ratio reinforcement schedule: the alternation of losses and rewards at unpredictable intervals provokes gamblers and gamers to play more and spend more money on the source of their reward-inducing experience.
Abusive partners are not always violent; they can also be affectionate, though their affection is unreliable. The alternation of abuse and affection characteristic of abusive intimate partnerships creates a de facto variable ratio reinforcement schedule, in which rewards (affection) are delivered at unpredictable intervals, interspersed with abuse. This pattern predictably induces reward-seeking behavior in the abused person, manifested in repetitive attempts to seek affection and affirmation (rewards) from the abuser.
Although reward-seeking urges are resistant to extinction, they eventually begin to abate in the absence of the expected reward; for example, video game addiction abates when video games are unavailable. Extinction is accelerated when constructive social activities offer an alternative to the expected reward. This was demonstrated in an experiment on addiction (Alexander et al. 1981), in which one group of rats were placed in solitary cages, and another was placed in a rat commune with recreational activities called Rat Park. Both groups were given access to morphine-laced water and plain water. The denizens of Rat Park preferred the plain water, whereas the solitary rats preferred the morphine-laced water; yet when the solitary rats were transferred to Rat Park, they quickly developed a preference for the plain water. This suggests that exposure to a natural habit (with social networks and recreational activities) extinguishes reward-seeking impulses fairly quickly, even when the reward is available in that environment. Extinction is likely to occur even more quickly when the reward is unavailable.
Interestingly, cult leaders use a strategy called 'love bombing' to recruit new members and compel existing members to conform to the cult's ethos (Singer and Lalich 1995). Love bombing is typically part of a variable interval reward schedule that alternates positive reinforcement (love) with the use of escalating demands, intimidation, and, in some cases, physical violence, according to cult researcher Margaret Singer (ibid.). Psychopaths are also known to use a variable interval reward schedule that alternates 'love bombing' with violence to coerce their victims into compliance (Hare 1999;Babiak and Hare 2006). Many non-psychopathic abusers probably use the same variable ratio reward schedule to induce reward-seeking behaviors and forced compliance in their partners.
The use of variable ratio reinforcement by an abuser is autonomy-undermining because it induces a reward-seeking pattern that impairs the abused person's ability to pursue safety, wellbeing, and self-actualization. This reward-seeking response, however, is not a personality trait but a universal human vulnerability that predatory people seek to exploit. Fortunately, the reward-seeking urge will begin to abate when the victim-survivor engages in healthy socialization, and/or when the expected reward is avoided.

Summary
To recapitulate, low self-efficacy, learned helplessness, obedience to authority/foot-in-the-door phenomenon, social-role conformity, and reward-seeking behavior appear to be situation-sensitive psychological states induced by salient features of IPV-namely, the escalation of demands and violence, enforced isolation, imposed anxiety and distress, perceived asymmetries of power and authority, and variable ratio reinforcement schedules. The psychological effects of these conditions induce autonomy deficits in the subject.
This social-psychological model of loss of autonomy satisfies three key desiderata: (a) it explains how anyone can be a victim-survivor of IPV, regardless of individuating personality traits, (b) it explains how victim-survivors can have a plurality of personality traits in spite of lacking basic autonomy, in opposition to the notion that they can be reduced to a homogeneous group of 'passive victims,' and (c) it explains how abused persons are capable of regaining autonomy in the aftermath of an abusive relationship.
A further desideratum satisfied by this model is that it avoids blaming the victim-survivor for being abused, but it does not excuse the abuser. I will address this implication in the next section.
One might object here that trauma due to IPV can have lasting autonomy-undermining effects, so the present model is too optimistic. I would not deny that trauma can impair deep autonomy for years after a violent assault, as Brison illustrates with a combination of narrative evidence and empirical data (2002). However, autonomy, as defined by Friedman, is a minimal agential capability, which involves merely the ability to survive, protect oneself from abuse, and act on one's basic values. This degree of agency is manifested in people recovering from trauma, even if a more robust level of autonomy-perhaps the level that they previously enjoyed-is unachievable due to traumatic after-effects. When I say that ordinary social conditions 'restore autonomy,' I am referring to this minimal threshold, not the level to which a person can aspire under ideal circumstances. Thus, the current proposal, viz., that non-coercive conditions support basic autonomy, is not overly optimistic. Deep autonomy, on the other hand, may require more than ordinary social conditions-perhaps robustly supportive relationships, excellent medical services, egalitarian laws, and other agency-enhancing resources that are in short supply. PTSD may be incompatible with ideal autonomy, but it does not preclude basic autonomy.

Moral Implications
This social-psychological model has implications for the moral and legal assessment of abusers and abused persons. It clarifies who is responsible for the loss of autonomy induced in abused persons by IPV. This is a question that has received substantive debate in philosophy (e.g., Brownmiller 2013; Enns 2012).
From a lay perspective, it certainly seems that the social cognition model implies that the victim-survivor is not responsible for being in an abusive relationship, insofar as the person does not autonomously choose to be in that relationship; rather, she is coerced by the autonomy-undermining features of the relationship, which reduce or eliminate personal responsibility. This is a pre-theoretical reading of the situation. It would, however, be informative to examine how theoretical models of responsibility parse this case.
There are three popular approaches to moral responsibility: the control view, the character view, and the functionalist view. We can consider these views one at a time.

Control
According to the control view, we are responsible for choices over which we have control. There has been abundant discourse lately on whether we can be responsible for implicit biases-states over which we lack first-personal access and direct, reflective control. This is instructive for our purposes because autonomy deficits imposed by IPV may be first-personally inaccessible. There is emerging consensus that we can be responsible for overt manifestations of implicit bias, provided that we could have used remediating resources to prevent the emergence of the implicit biases or the downstream manifestations of those biases in our overt behavior. 'Remediating measures' include exposure to counter-stereotypical exemplars, intergroup contact, and implementation intentions (Holroyd 2012;Kelly et al. 2010). The upshot is that we can be responsible for the expression of non-conscious states, provided that we were capable of exercising indirect, non-immediate control over them using available resources, i.e., resources at our disposal within our environment. People who fail to take reasonable steps to control their implicit biases using these resources are responsible for the consequences of that choice, decision, or omission.
The implications of this approach for victim-survivors are more complicated than they may at first seem. The psychological effects of IPV are similar to implicit bias in that they are not subject to direct introspection or reflective mediation, which is why they are difficult to shed. Victim-survivors may genuinely deny that their relationship is abusive, particularly if they think that they are in love with the abuser, which suggests that the psychological effects of IPV are not first-personally available-they are implicit. Implicit states can be responsibility-imputing. However, notice that responsibility theorists who adopt the control model say that we are responsible for implicit states only if we have access to remediating measures within our environment, as these tools give us non-immediate, indirect control over our implicit states. One of the characteristic features of IPV is that the abuser manipulates the abused person's environment such that autonomy-conducive economic, epistemic, emotional, and other autonomy-conducive resources are unavailable. This suggests that victim-survivors of IPV and ordinary implicitly-biased people (living in what I have called 'ordinary conditions') are not equally responsible for their implicit states, because they occupy different environments: the first group has access to control-enabling resources, whereas the second group does not. Thus, even if both have the motivational capabilities to exercise control over their implicit states, these resources are suppressed in victim-survivors by the control-impairing features of their (constructed) environment. This provides reason to believe that victim-survivors of IPV are not responsible for the psychological effects of IPV, which prevent them from making fully autonomous choices.
Of course, not every victim-survivor's life situation is the same. Some may have some degree of access to control-enabling economic, epistemic, and emotional resources, and these individuals may bear responsibility for failing to make use of these resources to remediate their autonomy deficits. However, some abused persons may be so oppressed by their abusive partner that they have no direct or indirect control over their choices, e.g., if the abuser has completely cut off access to control-enabling resources. Moreover, because autonomy-undermining conditions tend not to be first-personally accessible (e.g., non-conscious primes, subtle persuasive techniques), the abused person may not have been capable of identifying or avoiding the control-undermining triggers that initiated her loss of autonomy.
Thus, the person may not be responsible for the onset of the loss of autonomy, or for the subsequent loss of autonomy caused by lack of access to control-supporting resources, on the control view.
The abusive partner, on the other hand, is in a similar position to an ordinary democratic citizen with implicit biases. People who commit sexual violence-one type of abuse-tend to be high in implicit rape-myth acceptance and implicit gender bias (Chapleau and Oswald 2010)-states over which the agent does not have direct, immediate control. However, if the abuser lives in a liberal democracy with open access to control-enabling relationships and information, he is, on the control view, responsible for failing to make use of those resources to control his abusive impulses. Even though knowledge about IPV is not as widespread as it should be, it is sufficiently 'available' to trigger responsibility in the control sense. A critical difference between abusers and abused persons is that, during the course of the relationship, abusers have the ability to make voluntary decisions about their actions within the relationship, whereas abused persons lack the (full) autonomy to do so.
Some abusers have, not only abuse-conducive implicit associations, but also explicit hatred of women, over which they uncontroversially have direct, reflective control, and over which they therefore bear responsibility. However, even if the abuser acts on implicit states, he is responsible for his actions insofar as he lives in a control-conducive environment. The salient difference between the abused person on the one hand, and the abuser on the other, is that the former lacks environmental access to control-enabling resources (because of the environment constructed by the abuser), whereas the latter is not similarly constrained (barring extenuating circumstances, such as being incarcerated in a control-disabling prison, or something of this nature). The average abuser therefore voluntarily chooses (in the ordinary compatibilist sense) to commit a grave moral harm, i.e., violently depriving another person of autonomy. Because environmental control is available, the person is responsible.

Character
According to the character view, people are responsible for actions that stem from their character, or their 'deep self' (Wolf 1993). In this framework, the abused person is not responsible for manifestations of her IPV-induced states, because, based on social cognition theory, these states are not part of her deep self. Abusers, on the other hand, act against public norms that proscribe violence against women, and do so independent of control-disabling constraints. Thus, they seem to act on individuating character traits-specifically, on an autonomous set of desires and attitudes (whether implicit or explicit) that sanction harming, coercing, and oppressing women. While many men share this masculinist motivational structure, it is not so pervasive in modern culture as to be socially determining; rather, it is cultivated by the agent himself. It is compelling to think that abusers who live in democratic, plural societies with access to a variety of perspectives are abusive by choice, not 'against their wills.' In this respect, they are different from Susan Wolf's famous example of JoJo, a tyrant who is morally incompetent only due to the absence of alternative perspectives in his culture. In contrast, abusers ordinarily have plenty of epistemic resources at their disposal; they just don't care about them.

Functionalism
According to the 'functionalist view,' people are blameworthy for their behavior if blaming them would serve a positive social purpose, such as promoting justice, equality, or wellbeing (McGeer 2013;Bell 2013). In this view, we need to ask (i) whether it would serve society to blame abused persons, and (ii) whether it would serve society to blame abusers.
Given that our patriarchal-colonial culture is saturated with victim-blaming and woman-bashing narratives, it is unlikely that blaming victim-survivors for being victimized would promote justice, equality, or collective wellbeing. On the other hand, seeing that our society tends to excuse rapists and empathize with the cisgender-male gaze in general, blaming abusers (who are predominantly male) would putatively serve the ends of justice, equality, and collective wellbeing. Asymmetrical blame between abused persons (innocence) and abusers (reprehensible) would be admissible in functionalist terms, insofar as this asymmetrical blaming structure serves to rectify existing hierarchies of respect and authority, which harm victim-survivors of IPV and embolden abusers. This is admittedly a fairly brief treatment of the functionalist model, but I do not have time to canvas all of the prospective effects of the proposed distribution of moral regard. I will therefore present this proposal as a prima facie application of functionalism to abusive partnerships, pending further investigation into the effects of this blaming structure. Additional inquiry into this question is welcome.

Summary
The control view, the character view, and the functionalist view prima facie converge in holding abusers responsible for their abusive behavior, while finding victim-survivors not responsible. Abusers, more precisely, are responsible for abusing their intimate partner, and thereby committing a grievous moral harm-willfully undermining another person's basic autonomy using violence. Abusers are committing a moral infraction that social scientists would not be allowed to perpetrate on human subjects or even, in some cases, non-human animals, and they are committing a particularly egregious type of objectification; not simply undermining their partner's autonomy, but manipulating the person's environment to force her to participate in her own oppression. This type of manipulation is described by Kate Abramson (2014) as literal torture, as well as an extreme violation of Kantian respect: one that does not merely treat the person as if she were an object, but manipulates her into seeing herself an object and internalizing the abuser's misogynistic worldview.
This analysis avoids a mistake that Rebecca Solnit identifies in public health responses to IPV, including the CDC's approach (Solnit 2017). Solnit criticizes the CDC for identifying alcohol as a main 'cause' of injury, violence, and unplanned pregnancy amongst women, without citing the main agential cause: violent men. By omitting the perpetrator of these adverse effects, the CDC places the responsibility for avoiding violence and unplanned pregnancy on women, perpetrating a common victim-blaming narrative. While I have identified (non-agential) features of IPV that induce autonomy-impairing states in abused persons (as this is important for educative purposes), I would not deny that these conditions are perpetrated by abusers, who bear responsibility for their actions. A causal analysis of the loss of autonomy experienced by abused persons is not incompatible with a moral analysis of the role of abusers in IPV. Identifying mediating causal factors, in fact, can be informative to abused persons.
The present moral analysis is consistent with an influential approach to research ethics. Philip Zimbardo in particular has promoted the idea that experimenters are responsible for the psychological effects of their research paradigms on experimental subjects. He holds that the ill-fated subjects of the SEP were not responsible for their role in the experiment (whether guard or prisoner), but instead he, as the engineer of the environment, was responsible for their actions and interactions (2007). This is why he has advocated for legal restrictions on SEP-type environments, like Abu Ghraib, which undermine the autonomy of their victims. The social engineers, in his view, are responsible for creating an autonomy-undermining environment and submersing people in it, which is a grave moral harm, and something that, with appropriate vigilance, the orchestrators could have avoided through better research, consultations, regulations, and other readily available informational resources.
Analogously, abusers are responsible for engineering an environment that predictably, reliably, and acutely undermines the autonomy of their intimate partner-a grievous moral harm that they are capable of avoiding using culturally available resources. Abused persons, by contrast, are often unwittingly submersed in deeply autonomy-undermining conditions-conditions that undermine their ability to exercise autonomy in the normal way. Once immersed in this environment, the resources for exercising autonomy are withheld from the abused person by the abuser's violent treatment.
This psychological account also has legal implications, which I turn to in the next section.

Legal Implications
The abuser is morally responsible for severe moral harms, and he may also be legally liable for abduction and sexual assault, based on a revisionary, but sensible, interpretation of the law.

Abduction
According to the law in Canada and most U.S. jurisdictions, abduction is the confinement of a person by persuasion, fraud, open force, or violence (18 U.S. Code § 1201; R.S. C., 1985, c. C-46). Canadian law includes the stipulation that abduction entails confinement 'against the person's will.' In most abduction cases, the abductor uses direct physical restraint, or persuades the victim to enter a situation of confinement or leave the country against her will. Most victim-survivors do not charge their abuser with abduction because the person does not seem to satisfy these criteria: he does not, for example, imprison his partner in a cell, or transport her in a locked vehicle, while she is physically resisting.
That said, abusers do 'trap their partner in a confined space' in the sense that they prevent the person from choosing to leave the shared home (in cases of cohabitation and imposed isolation). More specifically, abusers undermine the abused person's ability to choose a safe living environment. Although there are no physical constraints on the abused person's freedom of movement, there are psychological constraints on that person's freedom of choice: the abused person often cannot conceive of or act on a plan to escape, due to autonomy deficits imposed by the abuser. The abuser might 'forcibly transport' the person to an unsafe environment by persuading her, through the use of violence and coercion, to move in with him, which thus involves involuntary transport and involuntary cohabitation. Abusers may use other coercive tactics, such as limiting the abused person's access to transportation, money, friends, and family, and excessively monitoring the person's whereabouts (paradigmatic features of IPV), which places stringent constraints on the person's freedom of choice and action, sometimes to the point of making escape materially impossible. While the abuser may not literally lock his intimate partner in a cell or transport vehicle, he can limit the person's ability to realistically consider alternative ways of life and act on such ideals, as these mental and physical activities require basic autonomy, as defined by Friedman. Thus, with an interpretation of 'abduction' that recognizes the evidence-based impact of IPV on the victim-survivor's freedom of choice and action, abusers who live with their partner often are guilty of abduction. They effectively coerce their partner to live and/or travel with them without voluntary consent.

Sexual Assault
If the victim-survivor's autonomy is impaired by IPV, then she cannot autonomously consent to sex with the abuser. Assuming that the abused person loses autonomy in the presence of triggers, including the abuser himself, she cannot autonomously consent to engage in sex with him.
There is a well-known precedent for this interpretation of sexual assault. Having sex with a severely intoxicated person is assault because the person was not capable of autonomously consenting; she lacked the decisional capacity to consent to sex under the circumstances. Even if the person said 'yes' to the rapist, this is insufficient for legal consent because the agent lacks decisional capacity. Similarly, if an abused person lacks decisional capacity because of the autonomy-undermining features of the environment, then she cannot consent to sex with the abuser. Any 'sex' that takes place is, in actuality, rape.
Prosecuting abusers for abduction and rape does not require a re-writing of the law, but an interpretation of the law that takes seriously the environmental conditions required for genuine consent. The abuser undermines the abused person's capacity to consent to living with him, traveling with him, and having sex with him. The lack of consent with respect to these activities entitles the victim-survivor to pursue criminal litigation.

Research-Based Educational Interventions
The social psychology research also has implications for how to avoid, evade, and reduce the prevalence of IPV. Freidman argues that it is permissible to force abused persons to participate in litigation against their abuser, but she would surely agree that educating people to protect themselves from abusers is preferable to mandated litigation. Education enhances the autonomy of abused persons, enabling them to choose to pursue litigation if they see fit, and if there are no valid countervailing considerations. Since socially marginalized groups face discrimination, their autonomy may not be served by state interventions that can result in police brutality, housing insecurity, deportation, and other potentially autonomy-undermining consequences. Thus, education is generally preferable to mandated legal procedures, as it places people "in the presence of [their] freedom" (De Beauvoir 2014, p. 96), rather than forcing an external conception of the good on them. (This is not to say that forced interventions are never permissible, but that they are not always beneficial).
On this point, there are some interesting pedagogical implications from the social psychology research. Studies suggests, for example, that defecting from an abuser may be easier in the early stages of a relationship, seeing that the Milgram subjects tended to defect at the 150-volt mark or else not at all; participating in healthy social activities reduces a person's susceptibility to IPV, by providing access to adaptive social roles and outside perspectives; regarding abusers as non-credible and non-authoritative reduces the influence of the abuser and facilitates non-compliance; and autonomy-deficits caused by IPV are situation-sensitive and amenable to extinction under ordinary social conditions. This information may be useful to abused persons and their friends and families.
There is general agreement amongst social psychologists that disseminating knowledge about social psychology research can help people avoid 'morally dangerous situations' (Doris 1998), and can enhance self-determination. Evidence to this effect includes studies showing that students educated about the bystander effect are more likely than controls to intervene to help a putative victim two weeks after training (Beaman et al. 1978). Philip Zimbardo offers a program called 'The Heroism Project' that teaches people about important social psychology findings to empower them to live more autonomous lives (Miller 2011). Similar educational initiatives could be used to teach people about the autonomy-undermining features of IPV, how to avoid them, and how to resist them.
Unfortunately, most American students never learn anything about intimate relationships or IPV. A recent Harvard Executive Summary reports that "we as a society are failing to prepare young people for perhaps the most important thing they will do in life-learn how to love and develop caring, healthy romantic relationships" (Weissbourd et al. 2012, p. 1). A survey of 18-to 25-year-olds revealed that 70% wished that "they had received more information from their parents about some emotional aspect of a romantic relationships" (Weissbourd et al. 2012, p. 2). Most respondents had never spoken with their parents about: "being sure your partner wants to have sex and is comfortable doing so before having sex" (61%), assuring your "own comfort before engaging in sex" (49%), the "importance of not pressuring someone to have sex with you"(56%), the "importance of not continuing to ask someone to have sex after they have said no" (62%), or the "importance of not having sex with someone who is too intoxicated or impaired to make a decision about sex" (57%). About 58% of respondents had never had a conversation with their parents about the importance of "being a caring and respectful sexual partner." (Weissbourd et al. 2012, p. 3).
Many American schools do not offer any classes on sex or relationships, let alone IPV. This means that many Americans are unequipped to identify and respond appropriately to features of IPV that may threaten their autonomy and wellbeing, and many lack the resources to adequately respond to, support, and empathize with victim-survivors. Since IPV tends to involve a gradual escalation of violence, and salient features of IPV can be particularly subtle in the initial stages of a relationship, it can be extremely difficult for ordinary people to identify the onset of abusive behavior. Furthermore, the pervasiveness of stereotypes that represent sexual violence as normal, natural, and even romantic, presents epistemic obstacles to the identification of abuse. Education can repair gaps in our "shared hermeneutical resources" about IPV (Fricker 2007, p. 1), enhance people's ability to recognize IPV, and enable people to respond adequately to the experiences of victim-survivors. (That said, the consequences of abuse are so serious that abusers have a duty to investigate the implications of their behaviour).
If people can identify IPV and understand its moral and psychological effects, they will be better able to protect themselves from abuse, and to empower their intimates with knowledge. Society does not have to mandate legal proceedings against abusers if people can avoid these relationships in the first place, and can intervene to empower others. Education can also help women find autonomy-enabling resources (such as Women's Safe Houses and stable employment) that may enable them to safely pursue litigation. Social injustice currently makes litigation a dangerous choice for many women.
Although educational resources about IPV are not as available as they should be, this does not mean that abusive people cannot control their actions in a responsibility-triggering sense. Hermeneutical gaps may mitigate responsibility, but relevant information is not so unreachable, or so unintelligible, that abusers are absolutely non-culpable. Fricker (2007), who has written extensively on culpability and cultural ignorance, would not see modern-day abusers as mere victims of epistemic bad luck: they have sufficient access to epistemic resources to be held responsible for committing abuse.

Concluding Remarks
I have argued that, from a social cognition model, victim-survivors of IPV acquire autonomy-impairing states that are shallow psychological variables rather than personality traits. These states are reliably induced in psychologically normal people in response to salient triggers, such as the escalation of violence and demands, social isolation, induced anxiety and disorientation, perceived asymmetries of power, and variable reward schedules. The induced autonomy-impairing states are highly dependent on salient situational factors within abusive relationships, and they are susceptible to extinction in ordinary social conditions. These states are universal human vulnerabilities that can be induced in anyone by abusive people, but education can empower victim-survivors to cultivate and exercise their autonomy, and, if they see fit, use available legal resources to pursue litigation against abusers.
Abusive people are responsible for inducing autonomy deficits in others, and they are therefore responsible for an egregious moral harm, viz., undermining the autonomy of their partner using violence, coercion, and isolation. In the current psychological model, the abuser could also be legally liable for abduction and sexual assault, in case he has deprived his partner of the psychological, epistemic, and material resources required to make autonomous choices about whether to cohabitate and have sex with him.

Conflicts of Interest:
The author declares no conflict of interests.