1. Sartre, Beauvoir, and gaslighting
Recent years have seen an explosion of interest in the topic of “gaslighting,” Merriam-Webster’s Reference Shammas2022 word of the year.Footnote 1 There, it is defined as the “psychological manipulation of a person, usually over an extended period of time, that causes the victim to question the validity of their own thoughts, perception of reality, or memories, and typically leads to confusion, loss of confidence and self-esteem, uncertainty of one’s emotional or mental stability, and a dependency on the perpetrator.”
This definition tracks the everyday use of the term: most take “gaslighting” to refer to instances where A misleads B in order to make B doubt their basic rational competence.Footnote 2 The term originates from the 1944 movie, Gaslight, which centrally features such a phenomenon. There, Gregory actively tries to convince his wife, Paula, that she’s going insane. His hope is that this will allow him to steal her valuable jewels. Every time he turns on the light in their attic to search for the jewels, their gas light dims. Paula notices the dimming and asks Gregory for an explanation, but he denies any responsibility. This leads Paula to doubt what she knows and to become convinced that she herself dimmed the light but has somehow lost all memory of having done so.Footnote 3 This conversation repeats itself so many times that Paula eventually concludes that she must be losing her mind: after all, she seems fundamentally incapable of remembering actions that she apparently performed.
Although a great deal of the rising interest has occurred within strictly popular discourse, gaslighting is one of the rare topics that has enjoyed a parallel amount of interest in the philosophical domain. There have been a number of attempts made to provide necessary and sufficient conditions for what exactly gaslighting is and what exactly it isn’t (Podosky Reference Podosky2021; Kirk-Giannini Reference Kirk-Giannini2022). Further, many have investigated related questions, such as determining what the moral wrong of gaslighting is (Abramson Reference Abramson2014), exploring unique sub-phenomena like “moral gaslighting” (Manne Reference Manne2023) or “self-gaslighting” (Dandelet Reference Dandelet2021), and arguing that it’s possible for gaslighting to be a structural, rather than a strictly interpersonal, phenomenon (Ruíz Reference Ruíz2020; Berenstain Reference Berenstain2020).Footnote 4
Accounts that articulate conditions for gaslighting fall into roughly three camps: (i) intentionalist; (ii) background condition; and (iii) dilemmatic. For intentionalist accounts, an act counts as gaslighting if it is performed with the (conscious or unconscious) intention to make the gaslit come to doubt their basic rational competence (Abramson Reference Abramson2014; Spear Reference Spear2019). For background condition accounts, an act counts as gaslighting if “the gaslit [has been] subject to systematic epistemic injustice,” such that they are predisposed to doubt their capacities in light of what the gaslighter says (Podosky Reference Podosky2021; Stark Reference Stark2019). For dilemmatic accounts, an act counts as gaslighting if the gaslit is presented with a choice between trusting their own basic rational competence or deferring to the testimony of the gaslighter (who by stipulation they think highly of) (Kirk-Giannini Reference Kirk-Giannini2022). The example from Gaslight counts as gaslighting on all extant accounts.
Beyond Gaslight, another common example in the literature comes from Simone de Beauvoir’s autobiography, Memoirs of a dutiful daughter (Abramson Reference Abramson2014, Reference Abramson2024; Fricker Reference Fricker2007; Manne Reference Manne2023). Here, Beauvoir describes her early interactions with Jean-Paul Sartre, her classmate at the time but also her future romantic and philosophical partner:
Day after day, and all day long I set myself up against Sartre, and in our discussions I was simply not in his class. One morning in the Luxembourg Gardens, near the Medici fountain, I outlined for him that pluralist morality which I had cooked up to accommodate the people I liked but whom I didn’t want to resemble: he soon demolished it. I clung to my system, because it authorized me to look upon my heart as the arbiter of good and evil; I argued with him about it for three hours. In the end I had to admit I was beaten: besides, I had realized, in the course of our discussion, that many of my opinions were based only on prejudice, dishonesty, or hastily formed concepts, that my reasoning was at fault and that my ideas were in a muddle. “I’m no longer sure what I think, nor whether I can be said to think at all,” I noted. (Reference Beauvoir1958, 344)
The italicized portion in particular is often used to illustrate that Beauvoir was gaslit; that is, to pick out the paradigmatic state of not just doubting the truth of some particular belief she has, but doubting her ability to have truth-tracking beliefs, and indeed her ability to tell whether or not her beliefs are truth-tracking. And authors assume that this was the result Sartre intended to bring about. For instance, per Kate Abramson, he behaved the way he did out of either consciously intending “to have De Beauvoir sit admiringly and in awe at his feet” or an unconscious intention “to have others prop up his fragile ego.” Given the conjunction of Sartre’s assumed intention and Beauvoir coming to doubt “whether [she] can be said to think at all,” this case also fits the dictionary definition.
Indeed, this is a case of gaslighting under all extant philosophical accounts. It satisfies accounts with intention as a necessary condition because Sartre is assumed to have such an intention. It satisfies accounts with background societal inequalities as a necessary condition because it’s a pugilistic encounter that occurred between a man and a woman in 1930s France. And it satisfies dilemmatic accounts because Beauvoir faces the dilemma of choosing between trusting her basic rational competence and deferring to Sartre (whom she thinks of highly).
If the story ends here, then that seems plausible. However, let’s keep reading past the point where writers typically end this quote:
But I didn’t let myself be discouraged; the future suddenly seemed as if it would be much more difficult than I had reckoned but it had also become more real and more certain; instead of undefined possibilities I saw opening out before me a clearly marked field of activity, with all its problems, its hard work, its materials, its instruments, and its inflexibility. I no longer asked myself: what shall I do? There was everything to be done, everything I had formerly longed to do: to combat error, to find the truth, to tell it and expound it to the world, perhaps to help to change the world. I should need time and it would need hard work to keep to my purpose, if it meant keeping only a small part of the promises I had made myself: but that didn’t frighten me. Nothing had been done: but everything was possible. (Reference Beauvoir1958, 345)
If you read Beauvoir genuinely here, it seems that she wasn’t, after all, plunged so severely into insecurity that she came to doubt her basic rational competence. Rather, it seems that the temporary doubt generated by her conversation with Sartre served as a catalyst to double down on her commitment to philosophical inquiry. And this shows that Beauvoir was likely not gaslit by Sartre, because she seems more confident than ever in her potential.
Even with the inclusion of this further passage, the philosophical accounts on offer would still count it as gaslighting. But interestingly, the dictionary definition would not count this as gaslighting, given that Beauvoir is not actually caused to “question the validity of [her] own thoughts, perception of reality, or memories,” nor is she “le[d] to confusion, loss of confidence and self-esteem, uncertainty of [her] emotional or mental stability, and a dependency on the perpetrator.” This shows that extant philosophical accounts of gaslighting fail in their extensional adequacy, and their divergence from the dictionary definition suggests they have strayed from common-sense understandings of gaslighting. I contend that this is because they fail to attend to what’s going on in the head of the gaslit and instead attend entirely to what is going on in the head of the gaslighter or the pre-existing social background.Footnote 5 What we should be doing is locating (at least some) conditions that make an act gaslighting in the head of the victim.Footnote 6 Crucially, this will allow us to account for the fact that there is an important, but often forgotten, affective element at play in cases of gaslighting.Footnote 7
I will begin by assessing extant accounts and identifying problems with each. I will surmise from this that looking to the head of the gaslit is key, explain why the current internal experience account on offer fails, and then provide a novel analysis of gaslighting which includes the internal experience of the gaslit in the right way.
Here is a quick word about method. Gaslighting, to no one’s surprise, is not a natural kind. To make matters worse, it is a relatively new term with a wide variety of use-cases. As a result, attempting to provide a conceptual analysis of gaslighting puts us in somewhat muddy territory. You might think that there’s no real puzzle here because what gaslighting is is ultimately up to us. However—and this is taken as implicit in the literature—“gaslighting” seems to have some independent criteria that set it apart from other related phenomena, meaning that it would be inapt to endorse broad linguistic relativism about the term.
That said, I take the point that there is room for conceptual engineering in building out our definition of gaslighting. So my method in this paper will be partially descriptive—showing that extant philosophical accounts overgenerate or undergenerate instances of gaslighting—and partially ameliorative—showing that our concept of gaslighting would be ‘better off’ if we adopted/discarded certain conditions.
2. Intentionalist and background-condition accounts
In this section, I show that intentionalist and background-condition accounts either overgenerate or undergenerate cases of gaslighting. Having done that, I will then propose that the common feature these accounts fail to include is the internal experience of the gaslit.
2.1. Intentionalist accounts
One family of accounts maintains that some act counts as gaslighting iff the gaslighter intends to gaslight their victim. Kate Abramson’s (Reference Abramson2014) seminal paper on the topic is one such account.Footnote 8 Although she notes that “those who engage in this form of emotional manipulation are often not consciously trying to drive their targets crazy … [and] they often seem not to have any clear end-in-view,” the intention of the gaslighter creeps in elsewhere:
The central desire or aim of the gaslighter … is to destroy even the possibility of disagreement—to have his sense of the world not merely confirmed, but placed beyond dispute. And the only sure way to accomplish this is for there to be no source of possible disagreement—no independent, separate, deliberative perspective from which disagreement might arise. So he gaslights: he aims to destroy the possibility of disagreement by so radically undermining another person that she has … no standpoint from which her words might constitute genuine disagreement. (10)
Indeed, this is the intention that Abramson attributes to Sartre in the Luxembourg Gardens, and which she takes to render it an instance of gaslighting. She defines gaslighting as a form of “emotional manipulation in which the gaslighter tries (consciously or not) to induce in someone the sense that her reactions, perceptions, memories and/or beliefs are not just mistaken, but utterly without grounds—paradigmatically, so unfounded as to qualify as crazy.” Abramson’s central project is to identify the moral wrong of gaslighting, and she claims that it is in the gaslighter’s intention(s) that we find the moral wrong: that the gaslighter aims “to make it the case that a person is in fact no longer properly regarded as a full member of the moral community.”Footnote 9
I am unconvinced that intention is a necessary condition for gaslighting.
First, while the intention condition may fit nicely with paradigmatic cases like Gregory in Gaslight, who truly is engaged in a malicious long-con toward some specific end, it is unlikely to apply in garden-variety cases. After all, as Abramson herself notes, most gaslighters don’t have such a clear-cut intention or goal in mind (this is why she permits unconscious desires to count as intentions).
Second, Cameron Kirk-Giannini’s (Reference Kirk-Giannini2022) rebuttal of the intentionality condition is difficult to dismiss. He notes that many paradigmatic examples of gaslighting in the literature make no appeal to the intention of the gaslighter. Take, for instance, Abramson’s (Reference Abramson2014)
Skeptical Peers: I moved out of one field of philosophy in grad school due to an overwhelming accumulation of small incidents … When I tried to describe to fellow grad students why I felt ostracized or ignored because of my gender, they would ask for examples. I would provide examples, and they would proceed through each example to “demonstrate” why I had actually misinterpreted or overreacted to what was actually going on.
Many take this to be an obvious case of gaslighting. However, the case contains no information about the mental life of the fellow grad students, suggesting that whatever drives the intuition that this is gaslighting has little to do with intention. Thus, Kirk-Giannini argues, intention cannot be a necessary condition on gaslighting.
Finally, I think that we would be right in following Goethe in thinking that “misunderstandings and lethargy … produce more wrong in the world than deceit and malice do,”Footnote 10 even when we allow “malice” to include unconscious desires. This is true especially for wrongs that occur in virtue of the reification of pre-existing societal imbalances.Footnote 11 Much work has been done recently to suggest that a great number of sexist, racist, or otherwise bigoted acts, such as microaggressions, are totally unintentional: this is what makes them pervasive and difficult to rectify (Brennan Reference Brennan, Hutchison and Jenkins2013; Rini Reference Rini2021). Microaggressions often occur when someone is ignorant of societal imbalances, rather than being robustly bigoted themselves. Why not, then, apply the same idea to gaslighting, if so often it is just another sexist wrong? This is not to excuse the behavior—after all, negligence is certainly morally criticizable—but rather to motivate the idea that it’s plausible that many of the harms that turn on systematic injustices come not from actively malicious intent to harm or unconscious desire to “destroy even the possibility of disagreement.” They frequently occur due to just being dense and careless. We therefore ought to allow for the possibility that gaslighting could also occur due to the gaslighter being dense and careless.
I wager that this is true not just for descriptive reasons, but also for ameliorative reasons: that is, our moral progress would suffer if we didn’t allow for the possibility of unintentional gaslighting.Footnote 12 This is because it will very often be difficult to tell whether someone really has the intention to gaslight: it’s not as though we can open up their minds to reveal whether the intention was there all along. To make matters worse, as Abramson (Reference Abramson2014) herself points out, these intentions might well be hidden to the gaslighter themselves. If the gaslighter isn’t even able to identify these intentions, then an external party will hardly be better off in being able to reliably locate them. Limiting cases of gaslighting to cases that are intentional will result in us being unable to identify them a lot of the time. And insofar as we have a general moral goal of being able to identify and call out wrongdoings, this conceptual limitation would prevent us from doing so much of the time.Footnote 13
It’s worth noting that this is also a common view in the literature: many accounts push back on intention as a necessary condition. Instead, they contend that intention is a sufficient but not a necessary condition for gaslighting. Paul Catapang Podosky (Reference Podosky2021) makes such a distinction by providing a disjunctive analysis of gaslighting:
Gaslighting occurs when (i) a speaker uses words and either (ii) the speaker intends for the use of such words to cause a hearer to form (iii) negative attitudes toward her own interpretive abilities, or (iv) a speaker uses words without such an intention, but (v) the use of words is apt to cause the hearer to doubt her interpretive abilities (vi) owing to the hearer being subject to systematic epistemic injustice that has disposed her to do so. (212, emphasis mine)Footnote 14
However, that there are important counterexamples to the idea that intention to “cause a hearer to form … negative attitudes toward her own interpretive abilities” is even sufficient for gaslighting. Consider the case of education: in particular, teaching young people who might think that they know a lot more than they actually do. Part of responsibly educating them might involve a sort of delicate ego-check: they need to learn that, although they are certainly capable of a lot, their interpretive abilities are not nearly as infallible as they may have previously thought. After all, they’ve entered the educational environment in order to improve their epistemic faculties, and in order to do so, they need to be educated on the ways in which they fall short. This would be especially likely in a case where the class is on a topic that necessitates being shown all the ways in which you’re rationally fallible, such as a psychology class or an epistemology class. It doesn’t seem right to call this intentional act of making a listener doubt their interpretive abilities “gaslighting,” even if the hearer was ultimately thrust into despair over their rational faculties as a result of “having been subject to systematic epistemic injustice.”
Part of why is because it does not seem to involve a moral wrong. If we called this example of the professor “gaslighting” then we would be calling a core part of education—something we normally take to be a perhaps supererogatory thing to do—something morally wrong. As such, intention to make someone doubt their basic rational competence alone is insufficient to qualify as gaslighting. So while, in some cases, intention does seem to be a key component of gaslighting, making intention a sufficient condition will overgenerate cases. Gaslighting thus could be unintentional or intentional, but the presence of intention is neither necessary nor sufficient for securing the label.
2.2. Background social condition accounts
So the intention side of the disjunction is insufficient, which means the disjunction qua disjunction won’t work. But what about the other side of the disjunction, which says that gaslighting occurs if there are social conditions present that make it likely that the gaslit will come to doubt their basic rational competence? For Stark (Reference Stark2019), when unintentional gaslighting occurs, it is a form of testimonial injustice—a credibility deficit that occurs due to prejudice on the part of the gaslighter.Footnote 15 For Podosky, when unintentional gaslighting occurs, it is because the words in question are liable to make the gaslit doubt their basic rational competence because the gaslit is “subject to systematic epistemic injustice that has disposed her to do so.”Footnote 16
If pre-existing social conditions are necessary for unintentional gaslighting to occur, then any instance of an otherwise marginalized individual making an otherwise privileged individual come to doubt their basic rational competence (even with malicious intent) could not qualify as a case of gaslighting. Descriptively, this just seems inaccurate: while it’s possible to grant that many cases of gaslighting might occur via reification of pre-existing power dynamics, it hardly seems like a necessary condition. Kirk-Giannini (Reference Kirk-Giannini2022), for instance, raises the intuitive plausibility that “two cisgender gay men who have never experienced significant prejudicial treatment on the basis of either their gender or their sexual orientation and who harbor no identity-based prejudices against each other” could nonetheless gaslight one another. Hence, making this condition necessary seems like it would undergenerate cases.Footnote 17
In addition, the thought that gaslighting is necessarily a species of testimonial injustice doesn’t seem quite right. After all, testimonial injustice requires that a testifier is assigned a credibility deficit due to prejudice on the part of the hearer (e.g., a hearer regards a Black person’s testimony as less reliable than the hearer’s evidence suggests because they hold prejudice against Black people). But even our paradigmatic case of gaslighting, Gaslight, does not meet this condition. After all, Gregory doesn’t think Paula is unreliable at all! He knows full well that what she is saying is true: indeed, that’s precisely why his trying to convince her otherwise seems so villainous to us.
3. Gaslit-focused gaslighting
Notice that all of these accounts locate the qualifying feature of the gaslighting encounter outside of the head of the gaslit. When we make intention to cause another to doubt their basic rational competence the central feature of gaslighting, we locate gaslighting in the head of the gaslighter. When we make implicit bias or identity prejudice the central feature of gaslighting, we also locate gaslighting in the head of the gaslighter. When we make pre-existing societal imbalances the central feature of gaslighting, we locate gaslighting in the background context of the dynamic. Perhaps we will have better luck, then, if we include features of the gaslit’s experience.
3.1. The dilemmatic account
Kirk-Giannini (Reference Kirk-Giannini2022) aims to do just that. He does so by proposing that the paradigmatic feature of gaslighting is the dilemma that the gaslit experiences: a dilemma that may certainly come about unintentionally and/or occur without background conditions of epistemic injustice and identity prejudice.Footnote 18 His account goes as follows:
For all persons A, B, and propositions p: A gaslights B with respect to p iff (i) A intentionally communicates p to B, (ii) B knows (and A is in a position to know) that if p is true, then B has good reason to believe that she lacks basic epistemic competence in some domain D, (iii) A does not correctly and with knowledge-level doxastic justification believe p, and A does not correctly and with knowledge-level doxastic justification believe that B lacks basic epistemic competence in D, and (iv) B assigns significant weight to A’s testimony. (13)
The gaslit, then, is faced with a dilemma: they must either deny A’s testimony (which would be particularly bold, given that they “assign significant weight” to it), or they must come to doubt their own basic rational competence. It is this dilemma that is supposed to capture the confusing tension that is so core to gaslighting encounters.
There’s a lot in this account that I find compelling. In particular, I like that it moves away from talk of the gaslighter’s intentions or the external conditions of the encounter and instead focuses on the experience of the victim and the dilemma they are faced with. However, I submit that two of its conditions prevent it from adequately picking out all possible cases of gaslighting.
First, Kirk-Giannini’s condition (ii) requires that “B knows … that if p is true, then B has good reason to believe that she lacks basic epistemic competence in some domain D.” This condition will rule out too many cases that we plausibly want to call “gaslighting.” It’s perfectly possible that B might incorrectly, irrationally, or without good reason come to believe that p’s being true would make her lack basic epistemic competence, and that the case could still count as gaslighting. In the original case of Gaslight, although Paula hesitantly accepts Gregory’s first claim that she dimmed the gaslights but forgot, it takes several iterations of this before she becomes convinced that she’s insane. Consider an alternate version, where it only takes one instance for her to become convinced that she’s insane. This would intuitively still count as gaslighting, but it’s far from reasonable to conclude that quickly that you’re insane. Here, Paula would seemingly still be gaslit, but she doesn’t have good reason to derive the conclusion that she’s insane from one (apparent) instance of her forgetting something.
Second, endorsing (iv) crowds out cases that intuitively count as gaslighting in a similar way. For instance, B could not consciously grant all that much weight to A’s testimony, but A just gets so incensed in their disagreement with B that B starts to doubt themselves. If they were pressed on the matter, it wouldn’t be that B in fact believes that A is more likely to be correct about p than B is, but rather that A’s vehement insistence provokes a sub-doxastic response in B to start doubting their basic rational competence. The case still appears to be one of gaslighting, insofar as the kind of felt reaction is the same as it would be otherwise and is prompted by the same kind of behavior, but Kirk-Giannini’s account would not permit this.
The common theme between these two instances of under-generation is that Kirk-Giannini’s account of gaslighting fails to include the unreasonable, messy, irrational cases of gaslighting. This under-generation is especially problematic for an analysis of gaslighting because the felt experience so often has a robustly affective, arational component (Manne Reference Manne2023). Part of what makes the experience of being gaslit so incapacitating is that it is deeply messy. And this is attributable to the experience of being brought to the conclusion—by your own lights—that your basic rational competence is fundamentally defective. There is a deep sadness in this: insofar as most people construct their identities and ground projects through their ability to independently and coherently reason,Footnote 19 discovering that you have reason to believe your epistemic agency is fundamentally defective is deeply undercutting of your independent personhood.
But what’s especially messy about it is that it is not just sad: it is also incredibly confusing. If your own basic rational competence is what is bringing you to the conclusion that your basic rational competence is fundamentally compromised, then you also have reason to believe that you might be mistaken about this conclusion, given that the tools you used to get there are seemingly broken! If you have concluded that you ought to doubt your basic rational competence, but your basic rational competence is what has brought you to that conclusion, then it’s not at all clear what to believe, or how to go on reasoning about what to do. In this way, the experience of being gaslit is messy and confusing because it puts the gaslit in a distressing feedback loop.Footnote 20 Appeals to “good reason” just don’t get off the ground.
This feeling is also exacerbated by the fact that gaslighting often takes place within relationships where one not only cares very deeply about what the other thinks of them, but also presumably takes their gaslighter to be able to notice important things about them (e.g., intimate romantic relationships, teacher–student relationships, or supervisor–supervisee relationships). So it’s also, frankly, embarrassing to be seen that way by someone who you prize the opinion of so deeply. Relatedly, attempts to gaslight are at their most forceful when they target one’s competence in a subject matter that one cares about being competent within—for example, in the case of education, a student that cares deeply about doing well in some class will take their teacher’s regard of them to be particularly high-stakes.Footnote 21
This is all to say that, although it is tempting to view gaslighting within strictly epistemic terms, a key element of its phenomenology is the affective experience of the gaslit. Abramson (Reference Abramson2014) writes:
When gaslighting works well, its target ends up feeling “carved up”, “nobody”, or she may say she feels lost, that there’s nothing left of her, or with De Beauvoir “‘I’m no longer sure … even if I think at all”. She has lost her sense of independent moral standing. (23, emphasis mine)
And Kate Manne (Reference Manne2023) builds on this thought:
“the unifying feature common to cases which intuitively count as gaslighting is that agents are made to feel fundamentally defective—bad, mad, or similar—for harboring a mental state to which they are entitled” (142, emphasis on “feel” mine).
As a result, the fact that dilemmatic gaslighting cannot capture cases where someone comes to doubt their basic rational competence for either bad reasons or no real reason at all (i.e., when it occurs mostly or fully sub-doxastically) shows that it is inadequate as an account of gaslighting. Although Kirk-Giannini is right to include the internal experience of the gaslit, his account does not include it in the right way, such that these more affective or “unreasonable” cases would not be identified correctly.
In order to include the internal experience in the right way (i.e., such that the affective component of gaslighting is captured), I contend that we need two different elements in our account. First, we need to be very specific about the kind of uptake that gaslighting creates in its victims: we need to be clear on the precise way that, and the degree to which, gaslighting makes someone doubt their basic rational competence, rather than simply their individual beliefs. And second, we need a success condition: we need it to be the case that gaslighting only occurs when the gaslit in fact comes to doubt their basic rational competence—such that unsuccessful attempts at gaslighting are not correctly called “gaslighting.”
3.2. The phenomenology of gaslighting
The phenomenology of being gaslit has two key components, and we can identify them by looking at paradigmatic cases of gaslighting for commonalities.
First, gaslighting seems to always include coming to doubt one’s individual beliefs. This is evident in the original Gaslight case. Paula starts off believing that she did not dim the gas light, but after Gregory expresses disagreement with her on this point, she comes to doubt this belief. We see this in other cases as well. Sophie Dandelet (Reference Dandelet2021) examines the phenomenon of self-gaslighting through describing a case of public indecency. A man exposes himself to two teenage girls on the beach. The girls form the belief that this was an instance of public indecency. After self-gaslighting due to the concern that they won’t be taken seriously, the two girls come to doubt that this was actually a case of public indecency—perhaps they misremembered what they saw. All this to say, gaslighting seems to always involve instances where an initial belief comes to be doubted.
It’s crucial, though, that gaslighting does not stop here. Were someone to simply come to doubt some belief, this would not be sufficient to count as gaslighting (we doubt our beliefs all the time, but we are not gaslit all the time). What’s important is that the victim of gaslighting comes to actually doubt their fundamental rational competence: the doubt seeps into not just the content of the belief, but the method that they took to get there. Paula takes herself to be unreliable at remembering things she did; the girls on the beach take themselves to be unreliable at picking out instances of public indecency. It’s not that they just have “bad beliefs”—they are themselves “bad knowers.”
Although this surely might happen in a variety of different ways, I propose that it often occurs because of a total mismatch between the stated belief of the gaslighter and the reasoning of the gaslit. Paula believes that she did not dim the gas light. Gregory insists that she did. Paula believes Gregory in this instance, and continues to believe him each and every time it happens. As a result, in order to maintain rational coherence, at a certain point she must start to doubt her ability to form accurate beliefs, since apparently she’s been so unreliable at forming accurate beliefs about her dimming the gas lights.
This is the most paradigmatic element of gaslighting: that someone comes to believe that they are fundamentally incapacitated as a rational agent—at least with respect to some particular domain.Footnote 22 The work that “fundamental” is doing here is picking out the particular way in which one takes oneself to be incapacitated. It’s not simply that one takes oneself to have some collection of bad beliefs. Rather, one comes to think that one’s fundamental means of forming beliefs is defective.
Now, Paula does come to think that her basic capacity for having any kind of reliable beliefs is faulty. But as I have already indicated, one need not think that this is true with respect to all domains to count as having been gaslit. One could be gaslit about a domain-specific capacity, such that one believes that one’s ability to reliably pick out true beliefs in some specific area (e.g., mathematical reasoning or identifying bigotry) is what’s broken, but is still confident in one’s competence with respect to other topics.Footnote 23
How intense must this doubt of one’s basic rational competence be in order to count as gaslighting? Abramson (Reference Abramson2014) characterizes the capacity doubt central to gaslighting as thinking that your beliefs are “not just mistaken, but utterly without grounds.” I submit that the degree of capacity doubt must be even more severe. Recall the example from earlier about the value of doubt—even of one’s basic rational competence—in certain educational settings. The PSYC 101 student learning about the ubiquity of implicit bias will likely think that the beliefs of theirs affected by implicit bias are both “mistaken [and] utterly without grounds,” thus meeting Abramson’s characterization. However, they are surely not gaslit.
This indicates that the kind of capacity doubt involved in gaslighting needs to reach a particular degree of intensity: one that involves believing they are hopeless. The PSYC 101 student who learns about the pervasiveness of implicit bias and starts to doubt certain elements of their reliability does not conclude that they are beyond help. Rather, they likely will start to enact certain practices in their life to circumvent these implicit biases. Were Paula to do the same, I don’t think we’d be inclined to say she’s been gaslit, because she would seem to have retained confidence in her ability to deliberate and accommodate potential pitfalls in her reasoning. But crucially, she does not—she comes to believe that she is beyond help.Footnote 24 And this is why she is gaslit, rather than helpfully ego-checked. In this way, my account is more future-oriented than Abramson’s: the gaslit must believe that there is no hope for them in their future exercises of rational agency.
I wager that truly coming to believe that your rational agency is hopeless rationally necessitates holding that belief for some extended period of time. Just as, per Richard Wollheim (Reference Wollheim1999), it seems “conceptually incoherent to suppose that one could have an emotion—say, an intense jealousy or a consuming rage—for only a fraction of a second,” it seems odd to think that one could suffer the intense self-doubt of being gaslit so briefly. That is part of why, even though Beauvoir severely doubts herself after her spar with Sartre, she is not gaslit: she recovers too quickly for her to have truly doubted herself that seriously. Of course, though, this self-doubt need not be permanent to qualify as gaslighting—after all, Paula’s story in Gaslight has a happy ending.Footnote 25
3.3. The success condition
One point of controversy in the gaslighting literature is whether it has a success condition: that is, whether “gaslighting” aptly refers to both successful and unsuccessful attempts at gaslighting, or whether its use is limited only to cases where the gaslit in fact comes to doubt their basic rational competence. I contend that gaslighting must have a success condition, for three reasons.
First, an appeal to intuition. To my ear, there is something paradoxical about saying “you are gaslighting me” to someone else. This is because, if the speaker really had been gaslit, then they would presumably be confused and “carved up” enough that they would not be able to detect that they are being gaslit. Notice that there is a similar oddness to the claim “you’re deceiving me”: if the speaker really has been deceived, they’re not in a position to tell that they have been deceived. (The same goes for someone’s charging another with “misleading” them.Footnote 26 ) At best, the charge “you’re deceiving me” seems like loose speech for “you’re trying to deceive me.” Similarly, my reading of assertions like “you’re gaslighting me” is that they are loose speech for picking out attempts at gaslighting. If this is right, it suggests that an act only counts as gaslighting if it succeeds in getting its target to doubt their basic rational competence (again, much as there is a success condition on deceiving or misleading someone).Footnote 27
Second, recall the Beauvoir case from the beginning of the paper. As noted at the start, every extant philosophical account would count it as an instance of gaslighting. For the intentionalists, it would count because they posit that Sartre had the intention (unconscious or otherwise) to bring Beauvoir to doubt her basic rational competence—to incapacitate her to the degree that she could not be an “independent … source of possible disagreement.”Footnote 28 For disjunctive accounts that feature a non-intentional side of the disjunction, the social backdrop and/or Sartre’s use of words being liable to produce basic rational competence doubt are sufficient for it to count as gaslighting. On the dilemmatic account, Beauvoir certainly seems presented with the characteristic dilemma of having to choose between doubting her own capacity and doubting Sartre’s testimony, and thus the conditions for gaslighting would be fulfilled.
As noted at the start of the paper, this is an undesirable conclusion. If we read Beauvoir genuinely, her spar with Sartre and coming to briefly doubt her own abilities serves to fan the flame on her confidence in her abilities and her desire to solve problems in the world. She certainly does not seem to think that there is no hope at all for her basic rational competence. And this conclusion seems worlds away from Paula’s, who believes that she is beyond help.
The reason why the intentional or background condition accounts generate the conclusion that Beauvoir was gaslit is because they lack a success condition: for them, mere attempts at gaslighting, or actions that are liable to produce a gaslighting-esque response in someone, are sufficient for counting as gaslighting. But given the further details of the Beauvoir case, we should want to avoid this conclusion.Footnote 29
Third, I want to return to the central desideratum of my account, which is to include the internal experience of the gaslit. To not endorse a success condition—to hold that mere attempts at gaslighting that receive no uptake are still aptly called “gaslighting”—is to not truly maintain this inclusion.Footnote 30 This is not to say that accounts that lack a success condition are utterly confused in virtue of lacking one. For instance, if one is an intentionalist, whether one adopts a success condition is ultimately a matter of terminological preference: one can call mere (unsuccessful) attempts at gaslighting “gaslighting” or decide to only use this term to apply to cases where the intention receives uptake. Both are fully intelligible. However, despite their intelligibility, neither are preferable to my account. Intentionalist accounts without a success condition will fail to include the experience of the gaslit, and intentionalist accounts with a success condition will undergenerate cases of gaslighting, for reasons delineated in section 2.1.Footnote 31
4. Gaslighting as (pro tanto) morally wrong
With these two criteria defended, I am now in a position to offer a first pass at my account of gaslighting.
Some actFootnote 32 A counts as gaslighting iff:
-
(1) the gaslighter’s act A expresses disagreement with the gaslit,Footnote 33
and this makes it the case that:
-
(2) the gaslit in fact comes to believe that there is no hope for their fundamental rational competence, either in general or within some specific domain.
In its current form, this first-pass account features both a success condition and some narrowing of the specific sort of doubt that occurs in cases of gaslighting. It makes no appeal to internal mental states of the gaslighter or pre-existing social contexts, which meets the desideratum that this account will include the gaslit’s own experience.
An unfortunate consequence of this tactic, however, is that we are currently left without the means to identify the moral wrong of gaslighting. Without any substantive appeal to a feature of the gaslighter, we have signed on for the possibility that someone could be perfectly kind, perfectly stand-up, perfectly good, and still end up gaslighting another person. After all, all that’s required for gaslighting here is for the gaslit to severely doubt their basic rational competence as a result of their gaslighter expressing disagreement with them. Things are worsened, too, because of the success condition: with a success condition, we are left without the means to diagnose the moral wrong of unsuccessful attempts at gaslighting.
This is a substantial problem because gaslighting is (plausibly) a pro tanto moral wrong. Yes, one might at times be in situations where gaslighting someone is the right thing to do (if Paula gaslit Gregory in order to escape, we would think this at least excusable and perhaps even right), but this is a clear exception to the rule.Footnote 34 Even if we disagree vehemently with someone, causing them to believe that their basic rational competence is so beyond help that they come to believe they’re incapacitated is a decent step beyond respectful conduct.
So, I propose a second-(and final-)pass account.
Some act A counts as gaslighting iff:
-
(1) the gaslighter’s act A expresses disagreement with the gaslit, without adequately expressing respect for the gaslit,
and this makes it the case that:
-
(2) the gaslit in fact comes to believe that there is no hope for their basic rational competence, either universally or within some specific domain.
Two initial clarifications on “expressing respect.” First, what do I mean by respect? Stephen Darwall (Reference Darwall1977) famously distinguishes between “recognition respect” and “appraisal respect.” Recognition respect refers to the kind of respect we owe all people, simply in virtue of their being people: taking their value to constrain the permissibility of your actions. Appraisal respect refers to the kind of respect we have for someone we hold in some kind of esteem. For example, I have appraisal respect for my sister in virtue of her quick wit. Here, I am using “respect” in the former sense. The gaslighter need not actually express special esteem for the victim in order to escape the charge of gaslighting: they must just express regard for them as, for example, an end in themselves.Footnote 35
Second, expressing respect requires a matching mental state–behavior pair. One cannot express respect without having a corresponding mental state of regarding someone as an end in themselves. To express respect purely phenotypically, as it were, is to not express respect at all.
“Expressing respect” might seem like far too high a bar to expect people to meet in disagreements in order to avoid the charge of gaslighting. It might sound as though you need to constantly placate and reassure your interlocutor (“I seriously disagree with you, but! don’t forget! I respect you!”), which seems like both an unrealistically high standard, as well as alienating for all parties involved. It would be especially odd to demand that you do this toward someone with whom you might intensely morally disagree. But this is not what I mean by expressing respect here. To get a better sense of what I mean, consider the following case.
disagreement & doubt: When hanging out with some colleagues after a department talk, you find yourself in a heated disagreement with a peer about what “begging the question” means. Your peer says that it refers to a logical fallacy where you presume the conclusion as a starting assumption, whereas you insist that it’s a rhetorical flourish writers use to introduce a new question. When you leave the event, you find yourself plunged into doubt: maybe you really were wrong about this, and you were too hasty in your insistence. Given that you were so confident and yet still quite possibly wrong, you begin to doubt your general ability to be correct about related facts.
We can fill out the details of the case, and in particular, the way that the disagreement is done, in different ways. Here’s one possibility: when you disagree with your peer, they seem to be actively listening to you and hearing out your reasons. They do not belittle or demean you. When they explain their view, they provide you with reasons why they believe that it’s a logical fallacy. Better yet: not only do they provide positive reasons for their belief, but they provide negative reasons for yours, demonstrating that they are interested in genuinely engaging with you and your reasons.
Here’s another way that it could go: when you disagree with your peer, they don’t hear you out. You tell them that you think it’s a rhetorical flourish and they try (and fail) to stifle a laugh. Rather than provide reasons for why their belief is right, they just say, “I don’t know what to tell you. You’re just wrong. That’s not what it means. Are you feeling okay? Gosh.” In particular, in discussing the matter with them, you don’t seem to get the sense that they think there’s any fraction of a chance that they might be wrong, nor do they seem at all willing to begin to consider it.Footnote 36
The first possibility is a case where you are shown respect, and the second possibility is a case where you are not. The distinction between the two can be cashed out as whether your interlocutor provides you with off-ramps to concluding that you have reason to consider your basic rational competence as beyond help.Footnote 37 The first possibility involves your interlocutor engaging with your reasons: implicitly treating them as understandable, even if absolutely incorrect. This would likely prevent you from concluding that you are hopelessly rationally compromised, because they are treating you as responsive to reasons by engaging with yours and providing you with their own.
The second possibility does not involve these off-ramps. They give you reason, unintentionally or otherwise, to believe that you might indeed be hopelessly rationally compromised, because they make no attempts to treat you as an epistemic agent by actually engaging with your epistemic agency. Rather, they suggest that you might be ill (“are you feeling okay?”), as if that is the only possible explanation for why you would believe something like that.
Consider also this distinction within an educational context. One way that, for instance, a mathematics professor might correct a student’s calculation error is by engaging with the (errant) reasoning they took to get to that conclusion—once again, treating the reasoning as understandable, even if incorrect. This provides the student with the relevant off-ramps: it provides them with evidence that they simply made a mistake, not that their rational agency is hopelessly compromised. Were the professor to not engage with the student in this way, such evidence, and accordingly, such off-ramps, would not be provided to the student.
The moral importance of expressing respect is underlined when we consider that, even though gaslighting can occur without pre-existing vulnerability conditions (as defended earlier), sociologist Paige Sweet (Reference Sweet2019) has shown that gaslighting often does occur due to the presence of certain vulnerability conditions in the mind of the gaslit. This is of course empirically observable through Sweet’s work, but it can also be made plausible through more theoretical considerations about the structure of gaslighting cases.
Consider an alternate version of Gaslight, where all the same events occur, but Gregory’s accusation that Paula dimmed the light fails to get uptake. In fact, over time, Gregory’s insistence that Paula dimmed the light actually becomes higher-order evidence for her to discount his reliability as a source of information, because she’s not sure why he keeps insisting on things that didn’t happen. Because she maintains this level of self-trust in her own capacities, the attempts at making her doubt those capacities simply fail to take hold. This does not strike me as a particularly unusual way things could have gone. Our perceptual evidence tends to be the evidence that we are most confident in. Thus, if our confidence in it is so weak as to conclude that we are the mistaken ones, rather than our interlocutor (and further, to conclude that this means our basic rational competence is beyond hope), it is likely that there is some pre-existing vulnerability present.Footnote 38
As I’ve said before, I don’t think that these vulnerability conditions are necessary for gaslighting, but they tend to be part of our explanation for how someone becomes gaslit: the victim has certain epistemic vulnerabilities that are primed to turn gaslighting behavior into gaslighting uptake. And insofar as it is plausibly a moral wrong to be careless with or take advantage of the vulnerable, our account of gaslighting must feature a condition that will stamp it as a pro tanto moral wrong: something that the first-pass account currently lacks. The second-pass account maintains the desirable inclusion of the inner life of the gaslit, but incorporates the interpersonal nature of moral wrongs by including a condition for what must be done in order to avoid the charge of gaslighting.
It ought to come as no surprise that our interactions with one another are often fraught: we are far from ideal discursive agents. Although “gaslighting” is a term that is relatively new in its ubiquity, it captures one such non-ideal phenomenon; one that has been around long before we had the hermeneutical resources to recognize it. As a result, it is vitally important that we refine our conceptual resources to pick out all and only the cases that are and should be called “gaslighting.” I maintain that my account succeeds on this front.
Acknowledgments
Thank you to Isabel Herburger, Devin Lane, Adrian Liu, Ram Neta, Sara Purinton, Gus Turyn, Susan Wolf, Elise Woodard, and Alex Worsnip for incredibly helpful feedback on this paper. Thanks as well as to audiences at the Applied Epistemology Project Work-in-Progress series, the British Society for Ethical Theory’s 2023 annual conference, and to the editors and anonymous reviewers at Hypatia for helping improve this paper.
Ripley Stroud is a PhD candidate in philosophy at the University of North Carolina, Chapel Hill with primary interests in epistemology and ethics.
