A sad song and a bad memory let taste get to the decision first, most of all in women with bulimia

"Comfort food" is such an everyday idea that it is easy to forget how little is known about what actually happens in the moment a bad mood tips someone toward the biscuit tin. For people with bulimia nervosa, an eating disorder marked by stretches of strict dieting broken by episodes of out-of-control eating, often followed by purging, the question is more than academic: binges are far more likely when negative feelings are running high. A study published in its final form in eLife on 1 October 2026 by researchers at the Icahn School of Medicine at Mount Sinai, Columbia University and colleagues looked inside the decision itself.

The team went back to an experiment first reported in 2019. Twenty-five women who met the diagnostic criteria for bulimia and 21 women without an eating disorder, aged 18 to 40 and matched for age and BMI, came in on two separate days after a standard meal. On one day they spent eight minutes writing about a recent upsetting experience while listening to Samuel Barber's Adagio for Strings; on the other, they wrote about their route to the lab to a neutral piece of music. Questionnaires confirmed the sad version worked. Each time they then rated 43 foods, some low in fat and some high, for healthiness and for tastiness, and chose between each food and a reference item they had rated neutral on both. One choice was picked at random and they got to eat it.

The original analysis found that mood did not change what people chose: women with bulimia picked the low-fat options more often in both moods, steered strongly by how healthy they thought each food was. The new work used a computer model that reads both the choices and the split-second timing of each response, and asks two questions: how much weight does a person give to taste and to health, and which of the two enters the decision first? In a neutral mood, the women with bulimia gave more weight to healthiness, especially for high-fat foods, and their weight on taste for those foods was actually negative, as if tastiness counted against a food. After the sad induction, both groups started to weigh up taste earlier relative to health, leaving a longer window in which taste alone was pushing the decision, and that shift was larger in the women with bulimia. Mood made the healthy women put less weight on taste; it did not do that for the women with bulimia.

The most clinically interesting result came from linking the model to symptoms. Among the women with bulimia, those whose low mood gave taste the biggest head start on high-fat foods reported more "subjective" binges over the previous three months, episodes that feel out of control even if the amount eaten isn't objectively large. There was no link with objectively large binges. The authors suggest that in a low mood, someone could decide to eat a tasty food before its healthiness is even considered, and that therapies which train attention could one day target this timing. The limits are plain: 46 women, all female, one lab session per mood, a mild mood induction, a snack-sized reward, final choices that did not change, and a symptom link that was exploratory and based on recall. It is a clue about how mood may bend food decisions, not a treatment.