The stages-of-grief model has a specific promise built into it: that mourning moves through a sequence, denial toward anger toward bargaining and onward, ending somewhere that looks like acceptance. Plenty of people don’t experience anything like that, and one brain-imaging study offers a clue as to why. For a subset of bereaved people, a photograph of the person they lost doesn’t trigger something that resolves. It lights up the same circuitry that switches on when a person wants something they don’t yet have.
We are writers, not clinicians, and grief that feels stuck, overwhelming or unmanageable is worth raising with a doctor or grief counsellor rather than working through alone with an article. What follows is a careful read of one specific study, not a framework for diagnosing anyone’s mourning.
What the study actually measured
The research, led by Mary-Frances O’Connor and colleagues, was published in NeuroImage in 2008 under the title “Craving love? Enduring grief activates brain’s reward center.” It’s important to be precise about who was studied: 23 women, all of whom had lost either their mother or a sister to breast cancer within the previous five years, split into two groups, eleven with what clinicians term complicated grief and twelve with grief that had followed a more expected course. This is a narrow, specific sample, not a general cross-section of people who’ve lost a parent or sibling, and the study’s findings shouldn’t be read as automatically applying beyond women who’ve lost a close female relative to this particular illness.
The distinction between the two groups matters for reading the result. What the study calls complicated grief is now more often referred to in clinical settings as prolonged grief disorder, added as a formal diagnosis to the DSM-5-TR in 2022: grief that remains intensely disruptive well beyond the timeframe usually expected, interfering with daily functioning rather than gradually easing. That’s a meaningfully different starting point from ordinary, if painful, mourning, and it’s the group in which the reward-circuit finding actually showed up most strongly.
Inside an fMRI scanner, participants viewed sixty composite images, each pairing either a photograph of their deceased relative or a photograph of a stranger with a word, either grief-related or neutral. They were asked to sit with whatever thoughts, feelings or memories the pairing brought up. The researchers were specifically watching activity in the nucleus accumbens, a small structure that sits at the centre of the brain’s reward and craving circuitry, the same system implicated in wanting a specific person, substance or outcome that isn’t currently available.
What lit up, and what it correlated with
Among the women with complicated grief, seeing their deceased relative paired with a grief-related word produced significantly more nucleus accumbens activation than the same pairing did for women whose grief had followed a more typical course. Within the complicated-grief group, that activation tracked closely with how much yearning a woman reported, not with how much time had passed since the death, her age, or how much general negative emotion she was feeling. Both groups of women also showed activity in brain regions associated with physical pain, consistent with grief’s frequent description as something that genuinely hurts, not just something sad.
O’Connor and her co-authors describe the pattern using language borrowed directly from addiction research, proposing that attachment to another person can activate reward pathways with what they call “addiction-like properties,” and that a reminder of someone who’s died can trigger something closer to a craving response than a memory that simply plays and fades. That’s the direct basis for describing this circuit as the same one that lights up when a person wants something, or someone, they don’t have. It isn’t a media gloss layered on afterward; it’s the authors’ own framing of what the data shows.
Where this does and doesn’t connect to the five-stages idea
The paper itself makes no mention of the five-stages model; that connection is worth drawing out here, not something O’Connor’s team argued directly. The five-stage framework comes from Elisabeth Kübler-Ross’s work with terminally ill patients rather than from a study of the bereaved, and grief researchers have separately built a case against treating it as a description of how mourning actually unfolds. One frequently cited analysis, by grief researcher George Bonanno, followed 205 bereaved people and found only about eleven per cent moved through anything resembling the assumed sequential pattern. Grief researcher Robert Neimeyer has put the broader verdict plainly: “scientific studies have failed to support any discernible sequence of emotional phases of adaptation to loss.”
What the nucleus accumbens finding adds to that picture is a plausible mechanism for one specific way grief can fail to resolve on the schedule the stages model implies. If longing for a specific, no-longer-available person engages the same reward circuitry as craving, then persistent yearning years on isn’t necessarily a sign that someone is doing grief wrong, or stuck at some stage they should have moved past. It may be a sign that the brain is running something closer to a craving loop than a memory that simply fades with time, a genuinely different process than the stage model was ever built to describe.
This doesn’t make prolonged, persistent grief less worth taking seriously as something to get support for, particularly where it’s affecting someone’s ability to function months or years on; if that’s the situation, a doctor or a grief-focused therapist is a better resource than any single study. O’Connor has continued this line of research since, including in her own 2022 book on the neuroscience of loss, and later work in the field has generally moved in the same direction as this 2008 finding: away from picturing grief as a fixed sequence with an endpoint, and toward picturing it as an ongoing process of a brain updating its model of the world to account for someone’s absence, a process that can go smoothly, get stuck, or loop back on itself in ways a five-step chart was never built to capture.
What this specific piece of research offers, on its own, is narrower and more useful than a new universal theory of mourning: not everyone grieves along a tidy sequence, and for at least some people, the brain’s own account of why looks less like unfinished mourning and more like wanting.