PSYREFLECT
RESEARCHAugust 3, 20264 min read

Prolonged Grief in the Scanner: When Salience and the Default Mode Run Too Close

Key Findings
  • Cross-sectional resting-state fMRI in 87 older bereaved adults: 42 with probable prolonged grief disorder versus 45 with integrated grief, equated on demographics and on time since loss. Connectivity between the salience network and the default mode network was higher in the prolonged grief group (p corrected = 0.014).
  • Symptom associations were exploratory and were examined within the prolonged grief and integrated grief groups separately. On that analysis, salience-to-default-mode connectivity correlated positively with grief severity (p corrected = 0.04) and with grief-related avoidance (p corrected = 0.04).
  • Within-network signal amplitude (fractional amplitude of low-frequency fluctuations) was higher in the default mode network of the prolonged grief group but not in the salience network. This result is reported at an uncorrected threshold only (p uncorrected = 0.04).
  • Group independent component analysis carried 8 networks into the pairwise comparison. Principal component analysis reduced the symptom picture to 4 dimensions, and connectivity across several network pairs beyond the salience-default-mode link was associated with a single component loading intrusive thoughts, yearning and avoidance together.

The neurobiology of grief has one famous image attached to it: in prolonged grief, reminders of the deceased recruit reward circuitry, and yearning starts to look like craving. It is a compelling picture and it has carried a decade of talk about grief as an addiction to the lost person. A resting-state study from the Medical College of Wisconsin lands somewhere less quotable and, for the consulting room, considerably more useful.

What the scan actually compared

Blair and colleagues scanned 87 older bereaved adults at rest: 42 who met criteria for probable prolonged grief disorder and 45 with integrated grief. The two groups were equated on demographics and on time since loss, and that is the design decision that matters most. This is not the bereaved against the never-bereaved. Both groups had lost someone; both had been living with it for a comparable stretch. What separates them is whether the grief settled.

Group independent component analysis decomposed the resting signal, and eight networks were carried forward: salience, default mode, left and right executive control, ventral and dorsal attention, sensorimotor, visual. Connectivity was then computed between every pair. The group difference reported is on one of those pairs: connectivity between the salience network and the default mode network was higher in prolonged grief (p corrected = 0.014). Explored separately within each group, the same connection tracked how bad things were: it correlated positively with overall grief severity (p corrected = 0.04) and with grief-related avoidance (p corrected = 0.04). Those symptom correlations are exploratory and within-group, not a whole-sample result.

Two further results are worth holding at different confidence levels. Within-network amplitude was higher in the default mode network of the prolonged grief group and unremarkable in the salience network, but that comparison is reported at an uncorrected threshold (p uncorrected = 0.04) and should be treated as a hint rather than a finding. More interesting is the symptom analysis: principal component analysis reduced the clinical picture to four dimensions, and connectivity across several network pairs beyond the salience-default-mode link was associated with a component on which intrusive thoughts, yearning and avoidance loaded together.

What this looks like in the room

Take the two networks seriously for a moment. The default mode network is what runs when nothing external is demanding attention: autobiographical memory, self-referential thought, mental time travel, the internal narrator. The salience network arbitrates what deserves the spotlight, mediating between internally generated content and the outside world. In prolonged grief the coupling between them is stronger, and within each group the stronger it was, the more severe the grief and the more the person avoided.

That is a mechanistic sentence a clinician can use. It does not say the patient is addicted to the deceased. It says the machinery that tags internal, self-referential, memory-laden material as urgent is running hot, and that grief-related avoidance is one of the behaviours moving with it. It does not set avoidance against yearning: the component analysis puts intrusive thoughts, yearning and avoidance on one dimension rather than separating them out, and the authors relate the wider pattern of between-network connectivity to all three. Patients present these as three separate complaints. Here they behave as one process.

The practical implication is a matter of emphasis, not a new protocol. When bereavement has not integrated, avoidance – of the cemetery, of the photographs, of the sentence "he is dead" – is the tractable handle on a cluster the patient experiences as three separate problems, and it is the one a clinician is least tempted to simply soothe. Grief-focused CBT and complicated grief treatment already work this way. What the study adds is a reason to hold the line when a patient insists that the avoidance is incidental and the pain is the real problem.

What it cannot support is anything diagnostic or predictive. The design is cross-sectional, so whether this connectivity pattern preceded the stalled grief or developed while living with it is simply not addressable. The sample is modest for imaging, and two of the three headline p values sit at 0.04. Nobody should be scanning a bereaved patient, and nobody should tell one that their brain shows prolonged grief.

Within each grief group, the tighter the salience and default mode pair ran, the more severe the grief and the more the person avoided – and avoidance is precisely the part of grief a patient will argue is incidental.

Limitations

Cross-sectional design in 87 older adults, so the connectivity difference cannot be dated relative to the onset of prolonged grief and no causal or predictive claim follows. The symptom correlations were exploratory and computed within groups. At the time of writing the full text was not openly accessible, so this analysis rests on the published abstract and record: effect sizes, the demographic detail and the specific grief instrument and cutoff could not be checked.

Source
Biological Psychiatry: Cognitive Neuroscience and Neuroimaging
Disrupted Large-Scale Brain Network Connectivity in Prolonged Grief Disorder: Relationship With Grief-Related Avoidance, Yearning, and Intrusive Thoughts
2025-08-05·View original
Tags
prolonged grief disorderneuroimagingdefault mode networksalience networkbereavement
Related
Resource
Porto, September: a grief conference with a level reserved for diagnosis and treatment
SPTFE and Universidade de Maia, with Bereavement Network Europe, the Danish National Center for Grief and the Irish Hospice FoundationRead →
Research
Chronic Depression Has Its Own Brain: Inflammation Carves a Distinct Circuit Signature
Biological PsychiatryRead →
Research
Where the Threshold Sits: Prolonged Grief in Kenya, Namibia and South Africa
Global Mental Health (Cambridge)Read →
PsyReflect · Free · Mon & Thu
Get analyses like this every Monday and Thursday.
Only what matters for practice. Curated by a clinical psychologist. 5 minutes instead of 4 hours of monitoring.
← Previous
The Alliance, Graded: A Free Entry Point to the Relationship Evidence