The evidence behind the cards
Checked at claim level, against the papers. Including what we got wrong.
Every card in a HELD deck names the idea underneath it, and many name the researcher it comes from. In July 2026 we checked every one of those attributions against the actual papers — not the name, the claim: does the sentence on the card match what that researcher actually found?
Five did not. They are listed below alongside the rest, because a research page that only showed the wins would be marketing. This is the working.
The rule we hold ourselves to: a card may always say less than its source. It may never say more.
What we got wrong, and fixed
A citation that pointed the wrong way
One card offered permission about what to do with a wedding ring and the photographs — keep them, box them, decide later. It cited Vamık Volkan on “linking objects.” In Volkan’s work a linking object is a marker of complicated mourning: an item invested with magical significance that extends grief indefinitely. The card said something close to the opposite of its own citation. The attribution has been struck. The same inversion was then found on three more cards — the grief, pet-loss and SOUL decks' own linking-objects cards — and struck from each.
A citation pointing at the opposite advice
A card about the half-empty bed says that lying in the dark, breathing, still counts. It cited behavioural sleep medicine — a real field, whose central instruction (stimulus control) is to get out of bed when sleep will not come. A reader who followed the citation would land on advice contradicting the card. The citation has been struck; the card stands as what it is, a permission.
A clinical sample borrowed for a general card
A card about the rearranged house cited “derealization in acute grief.” The study behind that phrase measured treatment-seeking patients with complicated grief after a death, at the moment of the loss. Wrong population and wrong construct for a general divorce card. Struck.
A finding stated more strongly than the study
We cited a Cochrane review for the benefit of “attending meetings.” The review tested manualised twelve-step facilitation as a clinical intervention, for alcohol use disorder specifically — not attendance, and not all recovery. The citation was rewritten to say what the review actually tested.
An overstatement, corrected
A pet-loss card called pet bereavement disenfranchised grief’s “most common form.” The literature treats it as an example among many. The wording has been changed to “one of its most familiar forms.” The claim never appeared in that card’s recorded reading, so no re-recording was needed.
The verified record
Cochrane 2020 — AA / twelve-step facilitation
Kelly JF, Humphreys K, Ferri M (2020). Alcoholics Anonymous and other 12-step programs for alcohol use disorder. Cochrane Database of Systematic Reviews, Issue 3, CD012880.
Pagano — helping others in AA
Pagano ME, Friend KB, Tonigan JS, Stout RL (2004). *Helping other alcoholics in Alcoholics Anonymous and drinking outcomes: findings from Project MATCH.
Marlatt & Gordon — relapse taxonomy
Marlatt GA & Gordon JR — relapse taxonomy / cognitive-behavioural model of relapse prevention.
Medication for opioid use disorder — first-line, mortality-reducing
*(was A-004, the highest-stakes claim in the deck — it is printed clinical guidance about mortality and a prescriber will check it)
Exercise in substance use disorder
What it actually found: meta-analyses of exercise during SUD withdrawal — craving reduced across light/moderate/high intensity (SMD −0.71, 95% CI −0.90 to −0.52, no significant difference between intensities); depression during withdrawal reduced (SMD −0.40, 95% CI −0.52 to −0.27, 15 trials, n=1,029).
Doka — disenfranchised grief
Doka KJ (1989). Grief "that persons experience when they incur a loss that is not or cannot be openly acknowledged, socially sanctioned or publicly mourned." Pet loss is a canonical example; pet bereavement research finds grief paralleling human loss while mourners report stigma and absent support.
Stroebe & Schut — dual process model
Stroebe M & Schut H (1999), rationale and description (PubMed 10848151); "a decade on" (2010, PubMed 21058610). Bereaved people oscillate between loss-oriented and restoration-oriented coping; oscillation — including time off from grieving — is the adaptive process, and failure to oscillate flexibly is the risk marker.
Tonkin — growing around grief
Tonkin L (1996). Growing around grief — another way of looking at grief and recovery. Bereavement Care 15(1). The grief does not shrink;
Kaplan & Kaplan — attention restoration theory
Directed attention is finite and fatigues; natural settings evoke soft fascination, which restores it. Well-established construct.
Urge surfing — Marlatt
Introduced by Marlatt (1980s), later formalised in Mindfulness-Based Relapse Prevention. Cravings are treated as waves with a beginning, middle and end; most urges peak and pass within roughly 30 minutes when ridden out. An 8-week MBRP programme reduced craving and improved stress management.
Divorce corpus — the three load-bearing attributions
Slotter, Gardner & Finkel (2010), Who Am I Without You? The Influence of Romantic Breakup on the Self-Concept, PSPB. Breakup produces self-concept content change and reduced self-concept clarity, and reduced clarity uniquely predicts post-breakup emotional distress. Larson & Sbarra (2015) found improvements in self-concept clarity precede gains in wellbeing — the direction runs clarity → wellbeing, not the reverse. VERIFIED. This is a strong, specific, correctly-attributed finding and the divorce deck should lean on it harder than it does.
Bonanno — bereavement trajectories
Bonanno GA — resilience model of grief (2004 onward); Galatzer-Levy, Huang & Bonanno (2018) reviewed 54 studies. Resilience is the modal response to loss and potential trauma. Outcomes resolve into a small set of prototypical trajectories — in one bereavement study: resilience 68.2%, chronic grief 13.2%, depressed-improved 11.2%, pre-existing chronic depression 7.4%. Bonanno's work is an explicit challenge to stage models.
Gross — expressive suppression
Gross JJ. Suppression comes late in the emotion-generative process and inhibits the outward signs of emotion. Findings: suppression decreases behavioural expression but FAILS to decrease the experience of emotion, impairs memory, and increases physiological responding — in the suppressor AND in their social partners. Reappraisal, by contrast, reduces experience and expression with no memory cost.
Fisher — romantic rejection
Fisher HE, Brown LL, Aron A, Strong G, Mashek D (2010), Reward, addiction and emotion regulation systems associated with rejection in love, J Neurophysiology (PubMed 20445032). fMRI of recently-rejected participants still "in love": images of the beloved activated the ventral tegmental area and regions associated with craving, reward and addiction — the authors link this to the obsessive behaviours that follow rejection.
Eisenberger — social pain
Eisenberger NI, Lieberman MD, Williams KD (2003), Does rejection hurt? An fMRI study of social exclusion, Science. Foundational work on social rejection recruiting neural substrates overlapping physical pain.
Sbarra & Neff — self-compassion after divorce
Sbarra DA, Smith HL, Mehl MR (2012), When leaving your ex, love yourself: observational ratings of self-compassion predict the course of emotional recovery following marital separation. Psychological Science 23(3):261–269.
Bowlby — separation distress
Bowlby & Robertson: separation response runs protest → despair → detachment; protest involves crying, searching, resistance to comfort, often worst at night and around sleep, with real physical distress. Attachment theory holds that separation arouses this at every age, not only in childhood.
Doka & Martin — grieving styles
Intuitive (affective; processes by feeling and talking) vs instrumental (cognitive/behavioural; processes by doing and thinking).
Rando — STUG
Rando TA, early 1990s: Subsequent Temporary Upsurges of Grief — intense, unexpected waves triggered long after the loss by a song, a scent, an image, an anniversary. Rando's own metaphor is oceanic: grief usually laps at the ankles, then a wave takes your feet out.
Neimeyer — meaning reconstruction
Neimeyer RA — meaning reconstruction is the central process in grieving; mourners face twin needs to process the event story and to access the back story of the shared life, in a way that reaffirms attachment security. The strongest available framing for that card is the event-story/back-story pair, which it does not currently use.
Bruce & Schultz — nonfinite loss
Bruce EJ & Schultz CL (2001), Nonfinite Loss and Grief: A Psychoeducational Approach. Nonfinite grief is mourning for what was never realised, rather than for something possessed and lost — a continuing rather than resolving presence.
McEwen — allostatic load
McEwen BS & Stellar E (1993). The multi-systemic "wear and tear" the body sustains when stress responses are activated too often, too long, or never shut off — cumulative across metabolic, endocrine, cardiovascular, immune, GI and nervous systems.
Pennebaker — expressive writing
What the literature actually shows — and it is not clean. Smyth's 1998 meta-analysis: d ≈ 0.47. Across 100+ studies since: d ≈ 0.16. At least one recent meta-analysis found no significant effect averaging over all outcome categories. Results are difficult to replicate and the necessary conditions are unclear. A 2025 grief-specific meta-analysis found writing reduces grief but with small effect sizes, moderated by more sessions and therapist feedback.
Volkan — linking objects
Volkan VD, Linking Objects and Linking Phenomena: A Study of the Forms, Symptoms, Metapsychology, and Therapy of Complicated Mourning.
Boss — ambiguous loss
Boss P (1999). Ambiguous Loss: Learning to Live with Unresolved Grief. Harvard University Press. Ambiguous loss is loss that remains unclear, unresolved and without closure — including the case where a person is physically absent but psychologically present. Boss's own catalogue of ambiguous losses names divorce explicitly, alongside missing persons, immigration, adoption and dementia.
Terhell, Broese van Groenou & van Tilburg — network change after divorce
Terhell EL, Broese van Groenou MI, van Tilburg TG (2004). Network dynamics in the long-term period after divorce. Journal of Social and Personal Relationships, 21, 719–738. DOI 10.1177/0265407504047833. A 12-year, three-wave longitudinal study of 104 adults (40 men, 64 women) who divorced in 1987 or 1988. Most divorcees experienced network losses shortly after the divorce, and in about half of those cases the losses were not compensated for in the later years. For some, divorce brought network gains — but only in the longer term. Differences were partly explained by attitude toward the divorce and by conflict with the ex-partner afterwards, not by personal capacities or structural conditions.
Where the evidence is thin, we say so
Expressive writing is the clearest example. The early meta-analysis reported a moderate effect; across a hundred-plus studies since, the average has fallen to small, and at least one recent review found no significant effect overall. Our card invites writing. It does not promise that writing will help, because the evidence no longer supports that promise.
Four cards currently cite a body of literature rather than a specific finding. That is on our list to fix: a literature name reads like evidence while being impossible for you to check.
These cards are companionship, not therapy
Nothing here is treatment, and no card should change anyone’s care. Where a card touches something clinical — medication for opioid use disorder is first-line and reduces mortality; grief that stays very heavy for very long deserves a person, not a card — we put that on the card itself, not in a footnote.