What is triggering really? Possible parallels between eating disorders and child sexual abuse

By Emily T. Troscianko, University of Oxford


“Triggering” is a term that gets bandied about a lot these days. What is triggering really?


Calling something triggering is typically a shorthand for saying that it has the potential to induce distress, usually rapidly and without warning, by arousing memories and sensations associated with some kind of past trauma.


Eating disorders (EDs) and sexual trauma are two of the most frequent contexts in which the term tends to get used—and, presumably, in which the phenomenon tends to get experienced. But systematic research is scarce. Data I gathered on textual engagement amongst people with eating disorders, via a large-scale online survey study conducted several years ago in collaboration with the UK eating disorder charity Beat, suggest ways of getting more purchase on what a “triggered” response is, and what is likely to bring it about. In this post I offer some speculations on how our findings may translate to the context of child sexual abuse (CSA). These ideas are expansions on a talk I gave at the Wellcome-funded symposium “Researching representations of child sexual abuse in contemporary culture” in May 2021, and they draw on some of the other presentations given there. I’m also grateful to my friend and colleague James Carney for thought-provoking comments on a first draft.


Let’s start by taking a look at the main similarities and differences between the ED and the CSA contexts that may be relevant to understanding why “triggered” responses to texts happen so often in both, and what’s going on when they do.


CSA/ED similarities


  1. Their highly embodied nature (they’re both extreme examples of the bodily nature of trauma more generally). For example: Survey respondents reported feeling immediately fatter just because they’d been reading a novel featuring a slimmer woman. CSA-cognate reactions may be similarly strong and swift.
  2. The direct link to behaviours which makes triggering more intense and dangerous. In the ED case, for instance, a reader might stay longer on the treadmill or make herself vomit in response to what she’s just read. Behaviour change is the primary driver of most mental illnesses and recovery from them, and both eating and sexual activity are obvious realms for this to play out, in either direction.
  3. Difficult intersections with cultural tropes, especially thinness or precocious sexuality. Glamorization of dubious value systems in mainstream culture makes EDs and CSA more likely. For instance, all kinds of stimuli scream at us not only infinite variants on “thinner is better” but also subtler versions of the “be more fragile” edict. One of our survey respondents flagged the widespread literary espousal of the “if you’re a woman, stop eating when life gets difficult” instruction: “I have read books by Austen and Wollstonecraft where main characters go into crisis and stop eating, it is strongly hinted at that their eating is disordered and I find this a reminder that I should not be eating as a woman who is stressed with uni work etc.”. This type of context pushes back the point at which people are willing to say (for themselves or someone else) “OK, there’s really a problem here”. Distorted value systems normalize things that shouldn’t be, and in cultures where oppressive/distorting structures are more widespread (e.g. in religious societies where oppression of women is the norm, or within the porn, fitness, or diet industries), serious problems are even more likely to go unnoticed. Mildly “triggered” responses to textual materials may go on for years, driving vulnerability or impeding recovery.


CSA/ED differences


  1. The personal versus interpersonal core. In the simplest terms, an ED centres on actions you perform yourself, and CSA centres on actions performed by or towards or with someone else. This difference may result in perceived agency and risk (which I think are amongst the biggest underexplored aspects of triggered responses) playing out differently within the two categories. In EDs, illness/recovery status may change perceived and actual agency and risk in textual interactions: for example, being in the early “honeymoon” phase of an ED versus in a chronic disorder, in recovery, relapsed, or fully recovered, will likely involve attitudes towards textual and other stimuli ranging from deluded invincibility to exhausted caution to confident robustness. The distinction between current or past perpetrator, and between victim and survivor—and the existence of mixed states that combine two or more of these statuses—may involve similar differences. It’s useful to ask how this “status” variable might affect the likelihood of someone engaging with texts about EDs/CSA, especially via perceived control over one’s responses. Some of our survey respondents mentioned, for instance, that they had insight into the fact that feeling vulnerable made them more likely to want to seek out texts that would likely make them even more vulnerable, but that they chose to do it anyway, in some cases assuming that the fallout would be radically more manageable than it turned out to be.

Another question worth asking here is how much perceived agency and risk feed into degrees of ambivalence about illness/trauma-related experiences, and hence into the potential desire for reimmersion in them. As I understand it, for at least some victims of CSA, the trauma attaches less to being forced to do something against one’s will than to being drawn into the act and made to feel complicit in it, for example because the abuser offers rewarding sensations or emotions. This type of trauma is likely to be associated with a form of shame that may bear some relation to the shame ED sufferers often feel about having “brought their illness upon themselves”: Once something has gone wrong, perceived agency is always more uncomfortable than perceived helplessness. But there are obviously differences between the two contexts too, in the degree to which self and others are performing actions and/or suffering consequences. The phenomenology of “being triggered” cannot be a unitary thing, and I suspect that the presence/absence of ambivalence is probably key to differentiating between some of its variants. In many cases, if you never wanted to engage with any material about EDs/CSA, it probably wouldn’t be terribly hard to manage that. If you’re drawn to such material as well as repelled and disturbed by it, that’s quite another matter: The frequency as well as the potency and complexity are bound to be amplified. Authors instinctively recognise this. In the CSA case, we might even go so far as to say that the choice of (C)SA as a plot element rather than, say, nonsexual violence is probably made primarily because it does so reliably induce ambivalence: because representations of sex are almost always physiologically arousing even when they’re also erotically repellent. Thus a reader’s ambivalence is being orchestrated in ways that echo a sexual subject’s, whether for admirable aims or not or (most likely) some inchoate mixture.

  1. Prevalence of graphic references versus allusions. A large amount of anecdotal evidence and a small amount of more formal self-report data make clear that there exists a cluster of ED-related narratives that “trigger” people. A similar state of emerging evidence exists in the (C)SA context, where it also seems that narratives of various kinds include CSA representations that serve various narrative purposes (e.g. as plot device or tool for emotional alignment, as discussed by the symposium convenor Ailise Bulfin) and that at least some of these representations trigger people. At the symposium, Xavier Aldana Reyes suggested that because the main aim of horror movies is to generate negative emotional responses rather than to deal with CSA responsibly, they may often risk crossing the line into emotional manipulation or downright exploitation. Hazel Katherine Larkin suggested that media evocations sometimes lead victims to question the validity of their own experiences, for example via the lazy and misleading trope of the abuser as monstrous stranger. In general my sense is that graphic evocations of the relevant states and actions are probably far less common in easily available narratives of (C)SA than in ED narratives. Direct descriptions of physical features like extreme thinness (with or without encouragements to perceive this as either beautiful or ugly) as well as of associated behaviours like food avoidance, purging, or excessive exercise, seem far more publicly visible than explicit CSA evocations. Thanks presumably to their criminal and widely vilified nature, textual evocations of CSA tend to occur far more often in the form of indirect allusion or via nonrealist techniques that may legitimize a degree of voyeuristic fascination. Further along the spectrum, explicitly pro-CSA literature no doubt exists, but it’s far less easy to stumble across than pro-ED content. These differences may mean that we need to be circumspect when attempting a mapping between the structures and effects of ED and CSA triggers.
  2. Range of relevant practices. Social practices around eating and sexuality are bound to have differential effects on the nature of textual triggers. Both sexuality and eating can be healthy or disordered, but there is probably a lot more potential for variation in healthy sexuality than in healthy eating. Once consent is agreed, pretty much anything goes in sexual behaviour. This may presumably include rehearsing CSA, if that were what a survivor wished to do. By contrast, it seems bizarre even to ask whether it could be considered healthy for someone recovered from bulimia to stage fake bulimia. This means that CSA-related triggersare likely to have a much bigger range of variation than ED triggers, and that CSA-related textual triggering could take on more challengingly complex (as well as far less easily avoided) phenomenologies.




Having surveyed some of the converging and diverging characteristics of CSA and EDs as they relate to reading, we can turn to one of the central questions that the concept of triggering poses: How exactly do “triggered” responses come about in engagement with a text? Here are the top three suggestions our survey data generated as to the mechanisms of the reported effects, with suggestions for what carryover there might be to the CSA context.


  1. The narrowness of the interpretive filter that the ED context can apply to any textual encounter. One person wrote, as if it were the most natural thing in the world, “Reading the Hobbit, of course, just makes one want food”; others talked about envying the dead protagonists of anorexia memoirs. Eating disorders are, in a crucial behavioural sense that’s bizarrely often overlooked or denied, disorders of eating. But they are also, inseparably from their behavioural core, disorders of interpretation—in everything from the sensorimotor specifics of how narrow a gap has to be for you to turn sideways to get through it, to the conceptual associations you have with the experience of hunger. The distorting effects of embodied trauma on risk signalling, emotional regulation strategies, and interpretive flexibility (as described e.g. in Van der Kolk’s The Body Keeps the Score) and the potency with which these distortions drive behavioural maintaining factors suggest that the filter may be just as brutal post-CSA.
  2. The potency of positive (self-reinforcing) feedback For instance, you feel more obsessive, you’re less able to concentrate on anything else, so you seek out books on this topic, you read them more selectively, you get more obsessed, you eat less and exercise more, you exacerbate the cognitive rigidity of malnutrition, and you grow more obsessive, etc. Or feedback can also be negative, i.e. self-stabilizing. For instance, you feel nervous about eating, you read something engrossing, you get calmed by it, you find eating easier. There are few forces in the universe more ubiquitously powerful than feedback, and this is just one context in which its capacity to tend towards or away from equilibrium can make all the difference. We might safely assume that any post-traumatic context will involve an enhanced likelihood of unstable feedback structures amongst the physiological, psychological, and/or behavioural components of the broader system, any of which may be either exacerbated or alleviated by textual engagement.
  3. Competitive or otherwise ambivalent self-comparisons with protagonists, often appearance-focused. Here we circle back to the cultural ambivalence part of the picture and the fact that body comparisons are both so easy to make and (not coincidentally) so freighted with cultural baggage. Appearance comparison arises again and again in existing research on the dangerous effects of engagement with social media, and likewise in our survey data, mentions were rife of using a brief verbal character sketch as a basis for comparison that usually led to self-directed critique. In the CSA context, comparisons may or may not be as squarely appearance-focused, and may involve more complex emotional responses and moral judgements.


Textual features


As for what textual features are most likely to help determine whether a given response is therapeutic or anti-therapeutic, my hunch is that they may include the following:


  1. Narrative perspective. This is likely to affect mental imagery (via the visual perspective on a given scene) plus value-based appraisal of the perspectives presented in the text (especially if the narrator or focalizer falls into one or more of the perpetrator/victim/sufferer/survivor categories).
  2. Visual descriptions. It’s easy to assume that more detail = more danger, but my research on the fragmentary nature of visual perception and mental imagery, as well as numerous testimonies to the power of fleeting sketches of a character’s body or some aspect of their habits, suggest otherwise. One survey respondent mentioned, for example: “Reading bios I often pick up tricks (something will stick in my head, eg, ‘I enjoy violent purging’ and now I silent[ly] berate myself when I purge to increase the force of the purging”. All it took was the word “violent” in conjunction with “enjoy” to generate an enactive image, some positive associations, and the corresponding real-life action. Many other personal and social actions in the ED and CSA contexts could be triggered in similarly verbally efficient ways, perhaps offsetting the difference I suggested earlier in frequency of explicit evocations: Brief oblique mentions may be perfectly capable of doing a large proportion of the triggering work in many cases.
  3. Metaphorical webs of association are threaded through the ED value system (linking experiential elements like hunger or thinness with abstractions like power or purity), and full recovery always requires that this type of cognitive architecture be dismantled wholesale. Something similar may apply in CSA, possibly with a more explicitly interpersonal structure. Conceptual metaphor theory gives us a framework for predictions that the metaphorical choices an author makes (whether filtered through a salient narrator/focalizer or not) are likely to affect interpretive activity within and beyond the textual encounter.


Beyond texts and readers


Of course, none of this is just about the text in a vacuum. There’s accumulating evidence in many strands of empirical literary studies on the importance of reader variables (e.g. age) versus textual ones, as well of contextual versus textual (e.g. receiving a reading guide or doing post-processing via book-focused discussion) in shaping what the upshots of the encounter are.


We must remind ourselves, too, that reading is just one small part of what’s in play here. Humans are great at narrativizing things, and enjoying doing it, and the storytelling bias is common in therapy, even in CBT—where in ED treatment, for example, some therapists remain uninclined to do the more behaviour-oriented and therefore also more confrontational things that need to happen to generate and monitor progress, like food diaries and regular weighing. The storytelling bias is visible in publishing too. In the publication of survival memoirs, there seems to be a general lack of reflectiveness about the merits of publishing (as opposed to writing) such books: I often have the impression that authors assume that because the writing process felt satisfying to them, it therefore follows that 1) it was in fact good for them in substantial ways, and 2) it has a strong chance of being good for anyone else to read.


One confusion that probably feeds in here is the often-ignored gulf between awareness-raising (in a general readership) and therapeutic purposes (for affected people). Little thought appears often to be given to how a book will reach a general readership, if that’s what it’s designed for (why should anyone who knows nothing about CSA or EDs buy a book about it?) and what might happen if the specific readership gets hold of them instead (as seems much more likely). There is probably a significant ED/CSA difference here, though, tied to the personal/interpersonal difference I flagged earlier. CSA is (at least in part) done to you by someone who doesn’t understand enough to not do it, and who is probably unhappy and/or unwell themselves. EDs often involve trigger events like comments from parents/friends on bodyfat or eating habits, but then require long chains of action and reaction sustained by the individual themselves. This is a difference of degree, however, and in both contexts it must matter how other people perceive the problem, do or don’t play a role in its development, do or don’t get involved in recovery from it, etc. So thinking clearly about what we’re doing when we publish, beyond scratching our own storytelling itch, is clearly crucial.


Finding out more


After thinking clearly and sceptically to generate hypotheses, the best way to follow up is to put them to the empirical test. I’m currently running an experiment designed to find out, pre-publication, whether the recovery memoir I’ve written is helpful or harmful to people with a restrictive eating disorder, with clear quantitative cutoffs for ditching the publication plans if the answer is harmful. I’ve designed the book to forestall triggering and deliberate self-triggering as far as possible, but I’m also trying to be as sceptical of my own motivations and capabilities as possible.


There are many options available to us for empirical investigation (many of which are also relevant for assessing a theoretical evidence base and narrowing down a research project more generally). Some key variables to consider when designing an experiment to answer a specific research question in this field include:


  • Reading type: group or individual; independent or supported
  • Study type: observational or experimental (only the latter lets one draw causal conclusions)
  • Demographic: general population (e.g. for awareness-raising or prevention) or victim/sufferer, survivor, perpetrator, etc. (for a therapeutic intervention)
  • Cutoffs:g. clinical diagnostic measures versus gen-pop vulnerability measures
  • Timing: acute illness or soon after the event, versus later in recovery/rehabilitation
  • Control condition [for experimental designs]: another book about the phenomenon, an unrelated book, a book without accompanying reading guide, etc.
  • Outcome measures: pre/post diagnostic or vulnerability measures as above, other trauma/ED-relevant scales, readiness for change scales, open-ended responses to generate thematic clusters, etc.


In general we do well to reject the “quant”/“qual” divide by using mixed methods in any study we’re conducting, and by pushing forward the development of better methods that span the gap, e.g. sensitive computational linguistics methods for analysis of free verbal responses.


Overall, we know extremely little about all of this, so we need to correspondingly careful any time we make any general statements about textual interactions in any health-or trauma-related context. Above all, we should notstart with the assumption that literary texts are a good thing and we just need to find ways to prove that to the world. (It’s terrifying how common this position is amongst humanities researchers.) Truisms about the healing power of literature are wrecked on impact by the mere existence of triggering as a phenomenon observed right across the genre spectrum from The Girl with the Dragon Tattoo to Maria: or, The Wrongs of Woman.



If you’re thinking about conducting research on triggering in any context, or are already doing so, I’d love to hear from you. You can get in touch via my website at https://troscianko.com.



Read more on EDs and reading



  • Troscianko, E.T. (2018). Literary reading and eating disorders: Survey evidence of therapeutic help and harm.Journal of Eating Disorders, 6,  [open access here]



  • Troscianko, E.T. (2017). Feedback in reading and disordered eating. In M. Burke and E.T. Troscianko (Eds), Cognitive literary science: Dialogues between literature and cognition(pp. 169-194). New York: Oxford University Press.
    [download: Feedback-in-reading-and-disordered-eating_Troscianko-2017]


Emily T. Troscianko, from University of Oxford, is a writer, researcher, blogger, and recovery coach, with particular interests in eating disorders, consciousness, and the psychological effects of reading narrative.