What Anorexics Really Feel About Food

One of the most pervasive and pernicious myths about anorexia is that people suffering from it don’t feel hunger as normal people do. The etymology reflects this misconception: the Latin ‘anorexia’ comes from the Greek negative prefix 'an' plus 'oregein,' to reach after, desire. And the whole clinical history of anorexia, ever since it was first medically identified as a distinct disorder in the late 19^th century, has been preoccupied with this apparent anomaly. Now, though, it’s gradually being recognised as something of a red herring: "Generally [in anorexia] there is no true 'anorexia' (loss of appetite) as such" (Fairburn 2008: 13).

This is so important because it helps bring anorexia back from the realm of the mysterious and rarefied, and locate it where it belongs, in the messy midst of a whole lot of factors relating to diet, bodyweight, body image, control, self-esteem, obsessiveness, and all the rest. It isn’t a glamorous altered state in which hunger has vanished and foodlessness is all that’s desired. We need to do better than that if we want truly to understand anorexia.

That isn’t to say that in anorexia, disturbances in the experience of hunger aren’t common, even ubiquitous, in some form or another. It’s occasionally true that a genetic peculiarity leads to abnormalities in how hunger is experienced (Vandereycken 2006: 353), sometimes to the point where it seems largely absent. It’s also true in almost all cases that semi-starvation and the physical changes it leads to, like hormonal imbalances and shrinkage of the stomach, alter the experience of hunger, in particular often meaning that sensations of fullness occur sooner and more intensely. Nonetheless, the basic point to hold on to is that anorexia isn’t defined by some mysterious absence of hungereven if that’s what sufferers would like you to believe. Anorexia, for most people, is an experience of chronic hunger: it’s the baseline to everything, accompanied increasingly by feelings of cold and weakness as weight loss progresses. Hunger can be the ultimate enemy or the point of everything, or both at once. The most important thing of all is that it isn’t allowed to shatter the illusion of control by culminating in eating in a way that breaks anorexia’s rules. These require hunger to be denied, resisted, and overcome—even though only ever partially.

Indeed, if there’s a truly defining feature of anorexia, it’s the denial of hunger, not its absence. This is true from the very early stages when, for so many people, a restrictive eating disorder arises out of ordinary dieting, and mealtimes become times of saying ‘I’m not hungry,’ or ‘I’ll have something later.’ And it remains a feature throughout the more advanced stages of illness, when the denial is less to other people—who have probably given up asking—and more to oneself: I will not eat now, I do not need to act on my hunger. The pattern of almost automatic denial recurs in its various guises in the most famous memoirs of anorexia, like Marya Hornbacher’s Wasted: ‘Not hungry, I’d say. […] I’m on a diet, I say. […] Well, I wasn’t hungry anyway. […] I push my plate away, say loudly, I’m full’ (1998: 10-12).

Of course, denying hunger doesn’t make it not exist: in one early study, ‘the [anorexic] patients perceived hunger in a manner similar to the controls, but they were more preoccupied with thoughts of food, had a stronger urge to eat, and were more anxious when hungry’ (Garfinkel 1974). And for many people, it’s the existence of hunger combined with the refusal to act on it that represents the greatest exercise of power. After all, it’s more powerful, a better demonstration of self-control, to feel hungry and not respond to it than never to feel it in the first place. As an illness of semi-starvation, anorexia is usually characterised by the broadly predictable effects of weight loss, including a salient obsessiveness when it comes to food-related stuff: hoarding recipes, reading about food, weighing it, planning it, and—though this is often not realised by onlookers—often delighting in its carefully orchestrated consumption. Profound pleasure taken in eating certainly isn’t universal in anorexia, but it’s very common, and a completely natural consequence of the semi-starved state. And this state of course comes about in the first place because hunger isn’t being felt or responded to in the normal way. And so a powerful feedback loop develops. As with semi-starvation that happens for externally imposed reasons, but incorporating extra factors, in anorexia a number of psychological factors combine with the physiological ones to maintain the underweight state. One of these interactions starts with the serotonin-driven ‘hunger high’ which can make hunger feel euphoric, turning it into something positively addictive rather than merely the negative foil to positive feelings of self-control and so on. This simple physiological mechanism may in turn help maintain the metaphorical associations of hunger and thinness with a whole host of positive traits like specialness, power, purity, cleanness, and self-control.

So hunger and its denial, and the association of this denial with all sorts of positive attributes, is clearly a crucial part of how anorexia takes hold and retains its grip. The question remains, though, how and why exactly anorexia develops: how and why the experience of hunger and the intense liking and enjoyment of food can remain so pronounced, but eating can nonetheless happen so little that severe weight loss ensues, setting all the rest of the feedback in motion. How, to take my own experience as an example, could there be so much hunger (‘I’m desperate for food, & sleep,’ 4:08 a.m., 9 June 2008), and so much pleasure taken in eating when it was finally allowed (‘divine food—the last bites, of squished soft bread & lots of fat & the lingering taste of strong garlic, & salt to perfect it all,’ 4:53 a.m., 8 June 2008), and yet my BMI still be so firmly in the critical range and my life reduced to little but illness?

Recent eating-disorders research has begun to answer this question by distinguishing between the ‘liking’ and the ‘wanting’ aspects of responses to food in anorexia (based on Berridge et al. 2009). The traditional clinical wisdom is that anorexia is characterised by an ‘anhedonic’ mind state, in which no reward or pleasure is experienced, but recent experiments suggest that this is an over-simplification, and that people suffering from anorexia still like food but just don’t want it in the usual way; they experience ‘partial reward’ when it comes to food, rather than no reward at all. Furthermore, it’s been suggested that anorexics may not just want positive food-related things less, they may also want punishing stimuli (like self-starvation or excessive exercise) more than their healthy counterparts (Keating 2010).

A recent experiment (Cowdrey et al. 2013) indicates that both current and weight-restored (but not fully recovered) anorexics ‘want’ high-calorie foods less and low-calorie foods more, which is the opposite from average preferences in those who have never been ill. Wanting was here measured both directly, by indicating a point on a line from ‘not at all’ to ‘extremely’ in response to the question ‘How much do you want some of this food now?’, and indirectly, as reflected by reaction times in a forced-choice task (choosing between foods with different flavours and calorie contents, e.g. high-calorie savoury versus low-calorie sweet, according to which they ‘most want to eat now’). Compared to the much decreased motivation to eat high-calorie foods and increased incentive to eat low-calorie foods, the difference in explicit liking responses (in response to the question ‘How pleasant would it be to experience the taste of this food now?’) between anorexic and controls was much less, suggesting that the sensory pleasure of eating is still experienced, but is being overridden by higher cognitive inputs. This conclusion is supported by brain-imaging studies that find neural activation suggesting greater top-down cognitive control in response to food images in participants with anorexia or bulimia—with bulimics also showing increased activation in reward and somatosensory areas, possibly impinging on the control over eating which is exerted more consistently in anorexia (Brooks et al. 2011).

Two experiments have investigated the distinction between liking and wanting a little more closely, asking whether the difference between sweet and fatty foods might have a bearing on it. Drewnowski and colleagues (1987) found that anorexic patients estimated the sweetness and fat content of mixtures of milk, cream, and sugar the same as healthy control— i.e., their sensory perception wasn’t impaired in this context, as may be suggested anecdotally by anorexics’ sometimes extreme preferences for very sweet foods. But while controls preferred the high-fat mixtures over the sweet ones—as indicated on a 9-point rating scale ranging from ‘dislike extremely’ to ‘like extremely’—the anorexic and underweight bulimic patients liked the sweet ones but disliked the fatty ones, and showed elevated optimal sugar:fat ratios. This pattern was unaltered after weight restoration (but see my thoughts below on what counts as weight-restored). Normal-weight bulimic patients preferred sweeter stimuli than the healthy controls, but didn’t differ from controls in their optimal fat preferences.

Anorexia Nervosa Essential Reads

Anorexia Nervosa and Superior Mesenteric Artery Syndrome

Anorexia Nervosa: Addressing the Bugs in the System

A later study (Simon et al. 1993) found that anorexics and healthy controls assessed sweetness equivalently, but that the anorexics rated fat/sugar mixtures as both higher in fat and less pleasant than controls. Like the earlier experiment, this one found a heightened dislike for fatty foods in anorexics as opposed to controls, again with equivalent perceptions of sweetness; the difference here was that there was no difference between the two groups in how much they liked sweet stimuli, so fattiness seemed to be the major dimension of difference. This study also assessed the short-term effect of eating a meal (including both fats and sugars) on these preferences, finding that both anorexics and controls showed a similar response to fullness, in that taste preferences for both fat and sugar were temporarily reduced afterwards.

Fear of weight gain may be a factor keeping the preferences for sweetness lower than they might otherwise be. Eiber et al. (2002) found that participants with a range of eating disorders (bulimia, anorexia, binge-eating disorder) all reported (again on a 9-point scale from ‘extremely unpleasant’ to ‘extremely pleasant’) more pleasantness when they were allowed to spit out a sugar solution than when instructed to swallow it, suggesting that fear of weight gain rather than an inability to experience pleasure is partially responsible for anorexic responses to sweetness. A weakness of this study, however, was the absence of a healthy control group, which makes it impossible to tell how this preference may be interacting with the specific physiological and psychological factors involved in the eating disorders.

The fat-versus-sugar question was investigated in a study with a control group (Stoner et al. 1996), which used a list of 50 common foods varying in sugar and fat content and overall calorie content, and asked participants to indicate ‘Do you like this food?’ and ‘Would you like to eat this food?’ from ‘not at all’ to ‘extremely’ by bisecting a line at the appropriate point. All eating-disorder patients rated their desire to eat high-calorie foods lower than their desire to eat low-calorie foods, whereas healthy controls rated the two desires equally; anorexics rated their desire to eat high-calorie foods lower than controls, whereas anorexic-bulimic and bulimic groups didn’t. In anorexics, the correlation between liking and desire to eat the high-calorie foods was significant by the end of treatment. On the liking as opposed to the wanting dimension, anorexics liked high-fat foods less than controls, but interestingly, they liked low-fat foods the same as controls; and the liking ratings for high-fat-high-carb foods (unlike other combinations) also increased during treatment. As for the interaction of liking and wanting, although in general they were positively correlated, there were some interesting exceptions in the anorexic groups: in lower-weight anorexics (below 70% of their ideal bodyweight), liking for and desire to eat high-fat-low-carb and high-fat-high-carb foods were not correlated, and for higher-weight anorexics (more than 70% of ideal bodyweight), the low-fat-high-carb ratings showed no correlation. Again, these differences were diminished over the course of treatment. Overall, then, this study supports the theory that food avoidance in anorexia is not necessarily due to diminished liking—only the desire to eat was, in general, rated lower than that of controls. There also seems to be a greater reluctance in anorexics to eat foods that are liked, especially if they’re high in fat—they rated their desire to eat high-fat foods lower than their desire to eat low-fat foods, but rated their liking as the same.

And at the neural level, while it used to be thought that dopamine was the neurotransmitter involved in ‘hedonic pleasure,’ or ‘liking,’ it now seems that dopamine signaling is associated specifically with the ‘wanting’ aspect of reward (Berridge and Robinson 1998, Leyton et al. 2002). In anorexia, increased release of dopamine into a brain area called the nucleus accumbens during punishing starvation- or exercise-related activities seems to induce an experience of partial reward (wanting without liking), a pathway which when reinforced over time may help maintain the destructive patterns of undereating. There’s also evidence (Cowdrey et al. 2011) that recovered anorexics (again, see my caveats below) show an increased neural response to both pleasant and aversive food stimuli despite there being no differences from controls in the subjective experience of the stimuli, on either the wanting or the liking dimension, or for intensity of the stimuli (all measured by line bisection). Specifically, the pattern of responses to pictures of chocolate and to a pleasant chocolate taste suggested that a hypersensitive neural response to food, irrespective of positive or negative valence, along with some degree of dysfunction in early stages of emotion processing, may underlie anorexia and be linked to the restrictive behaviours that characterise it. When processing a negative stimulus (a picture of mouldy strawberries followed by an unpleasant strawberry taste), comparing activation patterns in recovered anorexics and controls also indicated that some kind of reward/punishment contamination may be at play.

One key question is whether all this goes back to normal after weight restoration and recovery—the last cited study suggests not, and that lasting differences in neural activation may serve as a biomarker for anorexia independent of subjective report. One factor which does seem to normalise is the unusually elevated levels of neuropeptide Y and peptide YY (11 and 14), which are important regulators of feeding behaviour (Kaye et al. 1990), and the reduced levels of neuropeptide leptin (Eckert et al. 1998); these appear to be a result of the physically malnourished state. Overall the picture is very mixed, and there’s an urgent need in this research area, perhaps in the brain-imaging studies more than anywhere, for a more coherent and well-founded approach to clarifying the criteria for inclusion in a particular experimental group, given that the physiological effects of starvation are so significant.

There are real problems with taking findings about post-recovery responses in these studies too seriously, simply because the criteria for ‘full recovery’ and ‘weight restoration’ are so lax. In Cowdrey and colleagues’ 2013 study, for instance, ‘weight-restored’ means a BMI of over 18. At 18.1, for the majority of Caucasian patients, the majority of the cognitive and physiological impairments associated with anorexia will still be in place, and indeed the other criterion for inclusion in this group is that ‘significant ED symptoms’ have to be reported—so why this condition is labeled this way is hard to fathom. Weight has not been restored in anything but the most partial sense. (Especially not if we take into account the temporary overshoot factor I discuss in my post on ‘not stopping halfway.’) ‘Fully recovered’ in this study means that the participants’ BMI has been between 18.5 and 25 for at least 12 months—again, a category encompassing potentially very ill people—and with EDE-Q scores comparable to global mean scores. In Cowdrey et al. (2011) there are additional requirements, namely that menstruation has also been normal and no psychoactive medication has been taken for 12 months. In Drewnowski et al. (1987) defined ‘return to target body weight’ as reaching the 50^th weight percentile for the return of menses (a measure of bodyfat relative to height; Frisch and McArthur 1974) and maintaining it for at least three weeks. BMI was used only as a supplementary measure, but the mean post-treatment BMI for the anorexic participants was only 18.9 (at an average age of 16.3). Brain imaging in general has substantial problems: the seduction of the pictures of ‘lit up’ brain areas it produces conceals a raft of analytical and statistical weaknesses. Other methodological decisions in these studies are questionable too: the ‘liking’ prompt in Cowdrey et al. 2013—‘How pleasant would it be to experience the taste of this food now?’—is phrased in a way that makes it much more cognitively connected to the ‘wanting’ prompt—‘How much do you want some of this food now?’—than it should be; why not simply ask ‘how much do you like this food?’ rather than bringing issues of imminent eating into the picture?

Despite these problems, though, this research makes clear that liking food isn’t the same as wanting to eat it. This is true for the general population too (Finlayson et al. 2007), but the divergence between the two may be particularly significant in anorexia. This insight offers us ways of thinking more clearly about how hunger, appetite, cravings, and profound preoccupation with and often delight in food coexist with clinically low energy intake and body weight. When someone with anorexia tells you ‘I’m not really hungry right now’ or ‘I’m fine for now, thanks’ or ‘No, thanks, I’m not too keen on that,’ remember that it probably doesn’t mean they’re not hungry or don’t like whatever it is—it may mean no more than simply: I am not willing to eat this now. That’s harder to say outright, though, so not wanting gets clothed in the softer, more acceptable language of not liking or not being hungry.

This post came into being thanks to Charlotte Baker, a fourth-year medical student who kindly shared with me an excellent review essay that introduced me to this research area and guided my subsequent research and writing.