Does Anorexia Cause Cognitive Difficulties or Vice Versa?

OK, so we have some strong evidence for correlations between anorexia and cognitive impairment. (See part two.) Now we can start asking causal questions. It might seem obvious that if cognitive deficits and anorexia are often found together, anorexia must cause the deficits. Still, it could be the other way around, or both could be caused by something else.

Do cognitive deficits predate or cause anorexia?

First, it’s possible that people with poor cognitive flexibility, for example, are more likely to develop anorexia. If we found this, it might suggest that the cognitive issues cause the anorexia or that both result from other problems.

One way to test this possibility is to ask whether a cognitive feature like inflexibility may be a trait marker (a persisting personality trait), not a state marker (a temporary state induced, for example, by anorexia or starvation), and then whether having this trait makes anorexia more likely. The only way to do this properly is to conduct prospective research: to run studies asking whether people who don’t yet have anorexia have cognitive deficits or patterns that then remain present once they are ill—and to include enough individuals that you can confidently label one a risk factor for the other. This type of study is difficult to conduct, and I couldn’t find any. Some prospective research exists (e.g., Stice et al., 2021), but here “cognitive symptoms” are defined as the classic risk factors in cognitive-behavioural models of eating disorder etiology (overvaluation of weight and shape, fear of weight gain, feeling fat), rather than the features we’re thinking about here.

Another way to go is to ask whether cognitive inflexibility has causes other than anorexia, and one way to do this is to see whether there are genetic influences. There’s some evidence in favour. An early study on set-shifting (moving between tasks or perspectives; see part two) found that

Both affected and unaffected sisters had more set-shifting difficulties than unrelated healthy women. This finding, together with the replicated finding that set-shifting difficulties persist after recovery, suggests that set-shifting difficulties are trait characteristics and may inform the search for the endophenotype in anorexia nervosa (Holliday et al., 2005).

(As for the “persist after recovery” conclusion, note that “normal weight” here meant a BMI of over 17.5!). Another study found impaired set-shifting in the healthy sisters of patients with anorexia, with some possible neurological markers (Friederich & Herzog, 2011), and a third found evidence for poor executive functioning (especially poor decision-making and set-shifting) in women with anorexia and their unaffected relatives, and suggested that the decision-making impairment is moderately genetically heritable (Galimberti et al., 2012). None of these results ruled out shared environmental influences, but they are compatible with the idea that cognitive inflexibility may be a trait rather than a state, and that anorexia is made more likely by having low flexibility—or indeed that both are made more likely by having some other characteristic.

Do cognitive deficits end when anorexia does?

Then there’s the perhaps more obvious causal question at the other end: Does the cognitive impairment persist after the anorexia is over? If not, we can confidently conclude that the anorexia caused the cognitive problems.

To answer this, we can measure how individuals perform before and after an intervention designed to help them recover, e.g. a clinical trial. This is where methodological failings often cloud the issue, because as I (for example, here) and others have pointed out, criteria for “recovered” in clinical studies are often laughably minimal. There’s been some improvement in recent years, including in differentiating better amongst different levels of progress towards recovery—for example, between weight-restored and recovered, or new and sustained weight restoration (Tomba et al., 2019). But the overall tendency continues of recruiting individuals to play the “recovered” role in the research design who are probably nowhere near recovered.

This habit causes a deeply unfortunate feedback loop between clinical trials and other research studies that involve no treatment: The trials call people recovered when they’re not, and then few truly recovered people are available for research on what recovery means, so the research finds that recovery doesn’t help much, and so researchers, clinicians, and participants adjust down further their definitions of what recovery means and makes possible. It’s a fairly depressing vicious circle driven by the boringly predictable pressures of funding and career progression (it’s cheaper to get someone to a BMI of 18.5 than to where she needs to be; and the more participants you can say your trial got recovered, the easier it is to publish papers in good journals) and the human status games that swirl around both.

Despite these problems, the short story is: Cognitive capacity does seem to bounce back in recovery. This seems to happen even potentially quite early in treatment, and to be linked with other types of improvement. For example, Duriez et al. (2021) found that cognitive flexibility improved during treatment, and that enhanced flexibility correlated with a significant amount of improvement in daily life functioning and reduction of eating disorder and depressive symptoms.

Looking at short-term memory and visuospatial and organizational tasks, Lozano-Serra et al. (2014) found that “After weight recovery, AN patients presented significant improvements in all tests, and differences between patients and controls disappeared.” They found more improvement on visuospatial, memory, and attentional flexibility in patients who had recovered at least one menstrual cycle, suggesting that “The normalization of hormonal function seems to be essential for the normalization of cognitive performance, even in adolescents with a very short recovery time.”

This finding about direct links with biological status aligns with the observation by Hatch et al. (2010) that when underweight, participants with anorexia performed worse than healthy controls on sensorimotor speed tasks and were more susceptible to task interference, though they had better working memory than controls. Once the participants were “weight-restored” the researchers found that on a range of tasks, individuals significantly improved relative to their own performance and relative to controls.

Eating Disorders Essential Reads

How Children with Eating Disorders Are Groomed Online

Reevaluating Military Fitness Standards

What about when cognitive improvement is the specific goal of a therapeutic intervention? In a trial of cognitive remediation therapy (CRT), (Leppanen et al., 2018), participants showed significant improvement in task-based measures of neurocognition. Overall (Tchanturia et al., 2014), CRT studies for anorexia report promising findings, and CRT appears to improve set-shifting, central coherence, executive functioning, and information processing. Unfortunately, cognitive functioning doesn’t tend to get measured except in trials of cognition-focused treatment methods, so we don’t know whether the improvements would be generated more, less, or equally efficiently by another method—one that focused on restoration of normal eating patterns, for example. (My money would be confidently on bigger and more lasting, if not quicker, wins that way!)

So without more prospective evidence, we can’t rule out causal links the other way around: Anorexia may be more likely in individuals who have specific cognitive deficits or traits, and be caused by them or by something else that causes both. But the upshot seems to be: If you have anorexia and you want to get smarter on pretty much any measure, your best bet is to get yourself weight-restored (ideally properly!) and hormonally back to normal.

In the next installment, we’ll consider the counterevidence and ask whether it poses a serious challenge to this conclusion.

To find a therapist, visit the Psychology Today Therapy Directory.