Association splitting

1. Introduction

Version followed: the Moritz & Jelinek self-help manual Association Splitting: self-help guide for reducing obsessive thoughts (German original 2007, VanHam Campus; free English and other-language editions on the Hamburg UKE Clinical Neuropsychology Unit site). The manual's body is not drawn on. Its foreword, the developers' web page and the 2007 pilot paper are drawn on. Procedural detail beyond those is marked [uncertain] or [synthesis].

What it is. Association splitting (AS; German Assoziationsspaltung) is a brief cognitive self-help technique for reducing obsessive thoughts in OCD. It was developed by Steffen Moritz and Lena Jelinek's OCD research group at University Medical Center Hamburg-Eppendorf (work since 2005, first published 2007). It is a single technique, not a full therapy. The authors present it as a self-help manual and as an add-on to cognitive-behavioural treatment, not a replacement for exposure and response prevention (ERP). [source: Moritz et al. 2007 abstract; UKE manual foreword]

How change happens. It rests on semantic-network and spreading-activation models of memory and on the "fan effect": the more associations a concept has, the weaker and slower each single link becomes. The group's research suggests that in OCD, core concepts are tied narrowly to their obsessive meaning (Wikipedia's example: cancer = death). AS has the person build new neutral or positive associations to these core concepts and strengthen existing ones (cancer = zodiac sign). The obsessive link then loses associative strength, so the core cue is less likely to trigger the obsession, and the thought becomes less intense and less urgent. [source: Moritz et al. 2007 abstract; Wikipedia summarising Jelinek et al. 2009, 2014] The title metaphor "fight fire with fireflies" captures this: many small harmless lights compete with the single threatening one. That reading of the metaphor is an interpretation [synthesis].

2. Framework

AS has no formal phase model in the sources. The sequence below follows the stated logic: identify the obsession, build associations, rehearse, re-evaluate. Session numbers come from the trials. [source: Moritz et al. 2007; UKE foreword; ordering = synthesis]

Session count and format.

• In the pilot, 38 people with obsessions were mailed the manual and reassessed after 3 weeks. About one-third responded (42% of completers). [source: Moritz et al. 2007 abstract]
• Later studies used other formats: an online randomized trial (NCT01035242, likely Moritz & Jelinek 2011 [uncertain]), a Russian-speaking internet sample, therapist-delivered AS vs cognitive remediation (Jelinek et al. 2018), and a brief instruction inside the myMCT self-help programme. Their session structures are not drawn on [uncertain].
• For a clinician, a workable add-on is 2–4 contacts over about 3–4 weeks: one for rationale and targets, one to build associations, and one or two to review and extend, with daily home rehearsal [synthesis].

3. Guiding principles

• Targets obsessions, not compulsions alone. Use it when a specific intrusive thought precedes the ritual; it is not meant for rituals without a preceding obsession, including mental neutralising rituals. [source: UKE foreword]
• Needs at least partial insight. The client must see the thought as at least partly exaggerated; otherwise treat that first. [source: UKE foreword]
• Weaken by adding, not by removing. You do not argue with, suppress or disprove the obsession; you crowd its trigger with other links (fan effect). [source: Moritz et al. 2007]
• New associations must be neutral or positive. Negative, threat-related or reassuring links strengthen the network you want to dilute. [source: Wikipedia, Moritz et al. 2007]
• Build new links and strengthen old neutral ones. Ordinary meanings the word had before the OCD count as much as invented ones. [source: Moritz et al. 2007 abstract]
• Variety and quantity. Many diverse links spread activation more than a few similar ones. This follows from the fan effect; a manual target number was not found [synthesis/uncertain].
• Personal relevance and vividness. Links tied to the client's own memories and senses are easier to recall. This is inferred from memory research; the manual's categories are not drawn on [synthesis].
• Repetition makes it work. One-off brainstorming is not enough; rehearse until the new links come automatically [synthesis consistent with the rationale].
• AS is not a ritual. Do not use it to cancel a thought in the moment the way a neutraliser would. Practise outside OCD episodes [synthesis; see section 5].
• Add-on to evidence-based CBT/ERP. If it does not help, move to cognitive-behavioural techniques and an OCD specialist. [source: UKE foreword; Moritz et al. 2007]
• Low threshold, self-directed. The therapist's role is to coach and troubleshoot, keeping time demands small. [source: format of the manual; framing = synthesis]

4. Intervention inventory

The sources describe one core technique with sub-steps: identify core terms, build associations, rehearse. Items 4.1–4.8 break it into usable steps. Items 4.9–4.11 cover monitoring, integration and the published adaptation. Each step is sourced or marked. All example language is illustrative and is not the manual's wording.

4.1 Suitability screening

• Purpose / when: Before starting. Decide whether AS fits this client and this obsession. [source: UKE foreword]
• Steps:
  1. Confirm an OCD formulation. Measure baseline severity with the Y-BOCS (clinician-rated) or a self-report OCD measure. The pilot defined response as a ≥35% Y-BOCS reduction. [source: Moritz et al. 2007]
  2. Check that rituals are driven by identifiable obsessive thoughts (e.g., fear of contaminating others, doubt about having hit a pedestrian, fear of catastrophe if objects are disordered or a prayer is omitted). [source: UKE site examples]
  3. Exclude, or postpone, cases with rituals without preceding thoughts and cases where the main problem is mental neutralising. [source: UKE foreword]
  4. Check insight. The client should see the thought as at least partly exaggerated. [source: UKE foreword]
  5. Agree that AS runs alongside ERP/CBT, or as a bridge while waiting for it [synthesis].
• Example language (illustrative): "Before we try this, I want to check that there's a specific thought that sets off your checking. Is there a word or picture that kicks it off?"
• Watch for: Poor insight, overvalued ideas, or psychotic-level conviction. Treat these first; AS is not appropriate yet [synthesis from the insight criterion]. Also watch for pure "just-right" or incompleteness rituals with no content, where there is no core thought to split [synthesis].

4.2 Psychoeducation: networks and the fan effect

• Purpose / when: Session 1, or the manual's opening. Gives a rationale that makes daily rehearsal meaningful and separates AS from suppression. [source: Moritz et al. 2007; rationale content]
• Steps:
  1. Explain that memory works like a network: a word activates linked words automatically (e.g., "bread" brings up butter, bakery, breakfast).
  2. Explain that in OCD some words have become linked almost only to danger: "germs" → illness → harming my family. Each time the cue appears, activation runs down that one well-worn path. [source: Wikipedia summary of Jelinek et al. biased-network studies]
  3. Introduce the fan effect: when a word has many connections, activation spreads out and each single path carries less. [source: Moritz et al. 2007]
  4. Draw it. Put the core word in a circle with one thick line to the feared meaning. Then add thin lines to many other meanings and say the goal is to thicken those [synthesis: whiteboard illustration].
  5. State clearly that the thought does not have to vanish or be fought. It becomes one option among many [synthesis].
• Example language (illustrative): "Right now the word 'cancer' has one road out of it, and it leads to death. We're going to build lots of side roads, like zodiac signs, crabs on a beach, the Tropic of Cancer. The more roads there are, the less traffic goes down the scary one."
• Watch for: The client hearing "don't think the thought". Correct this at once, since suppression backfires [synthesis]. Also watch for scepticism that it is "too simple" (see section 5).

4.3 Identifying obsessive thoughts and extracting core terms

• Purpose / when: Session 1–2. AS works on concrete cue concepts, so the obsession must be reduced to the words or images that trigger it. [source: Moritz et al. 2007: "core OCD cognitions"]
• Steps:
  1. List the client's main obsessive thoughts in their own words.
  2. For each, ask which single word, object or image sets the thought going. Examples: "AIDS", "knife", "blood", "pedestrian", "cancer", or the name of a feared illness. [examples from UKE site and Wikipedia; specific word list = synthesis]
  3. Write 1–3 core terms per obsession. Choose terms that come up often in daily life, so ordinary encounters give practice chances [synthesis].
  4. Order the terms. Start with one of moderate distress, so building associations is possible without being swamped [synthesis; the manual's ordering rule is not verified, uncertain].
• Example language (illustrative): "When the worry about hurting your daughter shows up, what's the first word or picture in your mind? 'Knife'? Then 'knife' is our first word."
• Watch for: Terms that are whole sentences or abstract ("responsibility"). Turn them into concrete words or images. Writing the term down may itself raise anxiety; that is expected and usually settles (it overlaps with mild exposure) [synthesis].

4.4 Mapping the current association network

• Purpose / when: Before building new links. Shows the client how narrow the current network is. This links to the developers' finding of biased semantic networks in OCD. [source: Jelinek et al. 2009/2014 via UKE reference list and Wikipedia; using this as an in-session step = synthesis]
• Steps:
  1. Say the core term and ask the client to write the first 5–10 words that come to mind, quickly, without censoring.
  2. Mark each word as OCD-related (threat), neutral, or positive.
  3. Note the proportion of threat words. This becomes a simple baseline you can repeat at review (4.9).
• Example language (illustrative): "I'll say a word, and you write whatever pops up, as fast as you can. No right answers. 'Blood.'"
• Watch for: The exercise turning into reviewing feared scenarios in detail. Keep it to single words [synthesis].

4.5 Generating new neutral and positive associations (core step)

• Purpose / when: The central intervention, in session and as written homework, for each core term in turn. [source: Moritz et al. 2007]
• Steps:
  1. Write the core term in the middle of a page or card.
  2. Generate as many neutral or positive associations as possible. Use several routes; the manual's exact categories are not verified [uncertain]. Routes to prompt:
     • Other meanings of the word. For example, cancer as a zodiac sign or the Tropic of Cancer [source: Wikipedia example].
     • Compound words and phrases containing the term [synthesis].
     • Sound and rhyme [synthesis].
     • Senses: colours, smells, tastes, textures, sounds [synthesis].
     • Personal memories, such as places, people and events from the client's life [synthesis].
     • Images, film, music, books [synthesis].
  3. For each association, add a short personal detail that makes it vivid. Example: not just "red" but "my grandmother's red raincoat" [synthesis].
  4. Screen every entry and strike any that are negative, threat-related or reassuring. Examples: "cancer — but I'm healthy" is reassurance; "knife — I'll lock it away" is avoidance. [principle source: neutral or positive only; screening step = synthesis]
  5. Keep the list. It is the material for rehearsal.
• Example language (illustrative): "Let's go wide with 'blood.' What else is blood? Blood orange, blood moon, a blood donor van outside the supermarket, the colour of the tulips you planted. Which of those has a memory attached for you?"
• Watch for: Drying up after a few items (see section 5). Associations that are really safety statements. Perfectionism about "good enough" associations. If the process is working, the list grows quickly and the client gets curious or even playful with the word.

4.6 Strengthening pre-existing neutral associations

• Purpose / when: Alongside 4.5. Reactivates ordinary meanings the word had before OCD narrowed it. [source: Moritz et al. 2007 abstract: "existing ones strengthened"]
• Steps:
  1. Ask how the word was used before the OCD began, or how other people use it in everyday life.
  2. Recover concrete instances, such as a recipe, a job, a hobby or a joke involving the word.
  3. Add these to the rehearsal list with a vivid detail.
• Example language (illustrative): "Before all this, when did 'knife' come up for you? Cooking? Your dad's pocketknife on camping trips? Let's put those back on the map."
• Watch for: The client discounting old meanings ("that doesn't count any more"). Reframe these as exactly the links being rebuilt [synthesis].

4.7 Rehearsal and consolidation (home practice)

• Purpose / when: Daily between contacts, over roughly 3 weeks (the pilot's interval) and longer if helpful. Repetition is what makes the new links compete with the obsessive one. [source: Moritz et al. 2007 for the 3-week interval; dose and methods = uncertain/synthesis]
• Steps:
  1. Once or twice a day, at a calm time, take the card. Read the core term, then go through the associations one by one, imagining each vividly for a few seconds [synthesis; manual dose not verified].
  2. Vary the format: say them aloud, write them out again from memory, look at photos that stand for them, or add new ones as they occur [synthesis].
  3. When the core word turns up in daily life (in a newspaper, on a shop sign, in a conversation), briefly note one of the new associations. This uses natural cue exposure. The aim is practice, not relief [synthesis].
  4. Log practice in a simple diary: date, minutes, new items added, and a 0–10 rating of how strongly the word pulled toward the obsession.
• Example language (illustrative): "Think of it like a gym for this word. A few minutes a day, run through the list and really picture each one. Don't use it to push the worry away. Just practise."
• Watch for: Rehearsal drifting into a ritual: compulsive counting, needing to finish the list "properly", or reciting it to cancel a thought (see section 5). Signs it is working: the new associations come more readily when the word is heard, and the 0–10 pull ratings drop.

4.8 Handling the obsession when it occurs

• Purpose / when: During triggers. The sources give no instruction for in-the-moment use [uncertain]. The guidance below is synthesis built to protect against neutralisation.
• Steps (synthesis):
  1. Notice the obsession and label it ("that's the OCD link again").
  2. Do not use the association list to cancel or undo the thought. Let the thought be there, and continue with ERP response-prevention if that is part of the plan.
  3. Later, at a set practice time, add any new neutral association that came up.
• Example language (illustrative): "When the thought shows up, you don't have to answer it with your list. The list is for practice time. In the moment, just let the thought be there and carry on."
• Watch for: Clients who feel the list "makes it safe". This is a sign AS has become a safety behaviour; separate practice from triggers.

4.9 Progress monitoring and review

• Purpose / when: At about 3 weeks, then periodically. Decide whether to extend, continue or switch. [source: Moritz et al. 2007 used Y-BOCS at 3 weeks; idiographic measures = synthesis]
• Steps:
  1. Re-administer the Y-BOCS (or the self-report measure used at baseline). A ≥35% reduction matches the pilot's response threshold. [source: Moritz et al. 2007]
  2. Re-rate the intensity of each targeted obsession (0–100) and how often it occurred [synthesis].
  3. Repeat the free-association test (4.4) and compare the share of threat-related words [synthesis].
  4. Review practice logs.
  5. Decide:
     • Good change: extend to the next core term.
     • Partial change: continue, increase variety and vividness, and fix rituals in practice.
     • No change despite adequate practice: stop AS and prioritise ERP/CBT or refer to an OCD specialist. [source: UKE foreword for the switch advice]
• Example language (illustrative): "Let's see what's shifted. When I say 'knife' now, what comes up first?"
• Watch for: The pilot found responders were a subgroup (about one-third). Non-response is common and is not the client's failure. [source: Moritz et al. 2007]

4.10 Extending to further core terms and integrating with CBT/ERP

• Purpose / when: After the first term shows change, or in parallel with ERP. [source: UKE foreword: add-on to CBT; sequencing = synthesis]
• Steps (synthesis):
  1. Work through the remaining core terms one at a time, repeating 4.4–4.9.
  2. In ERP, AS can lower the starting intensity of a cue word before exposure, or follow exposure to broaden the network. Keep ERP response-prevention rules intact.
  3. Metacognitive work from myMCT can sit alongside AS. myMCT contains a short AS instruction and a related exercise, "attention splitting", aimed at over-focus on external stimuli. Its procedure is not drawn on [uncertain]. [source: Wikipedia citing Moritz et al. 2010]
• Watch for: AS replacing exposure the client finds hard. Keep both in the plan.

4.11 Adaptation: weight and shape concerns

• Purpose / when: The developers' group tested AS for intrusive weight- and shape-related thoughts in students with high concerns (feasibility, n=13, 10 completers; about 70% rated it helpful). [source: Musiat et al. 2014 abstract]
• Steps: Same structure as 4.3–4.9, with body, shape or food words as core terms. The study's specific materials are not drawn on [uncertain]. Do not set weight, calorie or exercise targets. Associations must be neutral or positive, not dieting- or appearance-evaluative [synthesis].
• Example language (illustrative): "The word 'thighs' only seems to lead to one place for you right now. What else do thighs do? Carry you up hills, kick a football with your nephew..."
• Watch for: Associations that sneak in comparison or restriction. Use only as an add-on to eating-disorder treatment, and assess medical risk through standard routes [synthesis].

Not included. Group or classroom AS protocols, child or adolescent versions, and app versions are not described in the sources. Rodríguez-Martín et al. (2013) tested AS in a nonclinical sample with unwanted intrusive thoughts, but their materials are not drawn on [uncertain].

5. Handling problems

The sources give two problem rules: who AS does not suit, and what to do on non-response. Everything else here is synthesis from OCD/CBT practice and is marked so.

• Client cannot generate associations (stuck). [synthesis]
  • Switch from free recall to prompted routes: other meanings, compounds, senses, places, people, songs, films.
  • Use a dictionary or thesaurus entry for the word together, or brainstorm aloud and let the client pick.
  • Accept silly or playful items; quantity first, then choose the vivid ones.
  • Set homework to collect three new items a day from real life.
• Every association slides back to the threat. For example, "blood orange" leads to "blood", which leads to infection. [synthesis]
  • Normalise this as the old pathway pulling. Strike the item and choose items further from the core meaning, such as sensory and autobiographical ones.
  • If this persists, the term may be too "hot". Start with a milder core term or do ERP work on the cue first.
• AS becomes a ritual or neutraliser. Signs: reciting the list to undo a thought, needing the "right" number or order, distress if practice is missed. [synthesis, building on the UKE exclusion of mental-neutraliser presentations]
  • Separate practice time from trigger time (4.8).
  • Deliberately vary order and number.
  • Drop any fixed count.
  • Treat the urge to recite as a compulsion to resist within ERP.
  • If this persists, stop AS.
• Using AS as thought suppression ("if I've done it right, the thought shouldn't come"). Re-explain that the goal is a wider network, not an absent thought. Expect the thought to keep occurring for a while [synthesis].
• Reassurance-seeking through AS. The client asks the therapist to confirm an association is "safe", or builds reassuring links ("I'm healthy"). Decline to reassure, and strike reassurance items from the list. [synthesis; neutral/positive-only rule sourced]
• Scepticism ("word games can't fix this"). [synthesis]
  • Agree it looks simple.
  • Show the network drawing and the free-association baseline.
  • Frame it as a low-cost 3-week experiment with a clear review point. The developers themselves present it as helping a subgroup. [source: Moritz et al. 2007]
• Non-adherence to rehearsal. [synthesis]
  • Shrink the dose to 2–3 minutes.
  • Attach practice to an existing routine, such as after brushing teeth.
  • Keep the card in the wallet or phone notes.
  • Review the log at every contact; problem-solve rather than criticise.
• Overwhelm or high anxiety when writing the core term. [synthesis]
  • This is a brief exposure effect. Slow down, start with a less distressing term, and let anxiety settle before generating.
  • If anxiety is severe or dissociation appears, pause AS and stabilise. Do not push a self-help add-on through acute distress.
• Ruptures: client feels trivialised ("you think my fear is a word game"). [synthesis]
  • Acknowledge the seriousness and the suffering.
  • Repair by explaining that AS addresses one mechanism (automatic links) and sits beside the rest of treatment.
  • Invite the client to choose whether to continue.
• Poor insight or overvalued ideas. AS is not indicated until insight is at least partial. Work on insight with cognitive methods first. [source: UKE foreword]
• Compulsions without obsessions, or mainly mental rituals. These are outside the stated target. Use ERP, including response prevention for mental rituals. [source: UKE foreword]
• No response at about 3 weeks. This is common (about one-third responded in the pilot). Stop or park AS and move to CBT/ERP or an OCD specialist or self-help organisation. [source: UKE foreword; Moritz et al. 2007]
• Harm, sexual or religious obsessions. [synthesis]
  • Assess carefully that the thoughts are ego-dystonic intrusions, as in OCD, before starting.
  • Choose neutral core terms (e.g., "knife", "child") and wholesome associations.
  • Never generate associations that elaborate the feared act.
  • If there is any indicator of actual risk to self or others, follow standard risk assessment before any OCD technique.
• Comorbid depression. Low energy can sink rehearsal. Reduce the dose and schedule practice as a behavioural-activation task [synthesis]. Screen for suicidality by standard practice.
• Eating-disorder adaptation pitfalls. Associations can drift into body comparison or restriction. Keep them neutral or functional (what the body does). Keep AS subordinate to eating-disorder treatment and medical monitoring [synthesis; adaptation source: Musiat et al. 2014].

Sources

Sources

• Fight fire with fireflies! Association splitting: a novel cognitive technique to reduce obsessive thoughts (Macquarie University research portal record and abstract)
• Manual Association Splitting (English): foreword page, UKE Clinical Neuropsychology Unit
• Association Splitting: a technique for reducing obsessive thoughts, UKE Clinical Neuropsychology Unit
• Association Splitting: feasibility study of a novel technique to reduce weight and shape concerns (Springer abstract page)
• Association splitting (Wikipedia; secondary)

Further reading (not drawn on) [uncertain]

• The ClinicalTrials.gov record NCT01035242.
• The English and French manual PDFs on clinical-neuropsychology.de, and the full 2007 article PDF there.
• Moritz et al. 2010 in Dialogues in Clinical Neuroscience: not drawn on.
• PubMed 21681869: not drawn on.
• Europe PMC: not drawn on.

Books and papers cited (not all drawn on)

• Moritz, S., & Jelinek, L. (2007). Association splitting: self-help guide for reducing obsessive thoughts (German original Assoziationsspaltung). VanHam Campus Verlag. The manual followed; the body is not drawn on.
• Moritz, S., Jelinek, L., Klinge, R., & Naber, D. (2007). Fight fire with fireflies! Association splitting: a novel cognitive technique to reduce obsessive thoughts. Behavioural and Cognitive Psychotherapy, 35(5), 631–635. doi:10.1017/S1352465807003931. Abstract only.
• Moritz, S., & Jelinek, L. (2011). Further evidence for the efficacy of association splitting as a self-help technique for reducing obsessive thoughts. Depression and Anxiety. Not drawn on; its link to NCT01035242 is not confirmed.
• Jelinek, L., Hottenrott, B., & Moritz, S. (2009). Papers on biased semantic networks in OCD and an introduction of Assoziationsspaltung (German). Cited via the UKE site; not drawn on.
• Jelinek, L., et al. (2014). Further evidence for biased semantic networks in obsessive-compulsive disorder. Journal of Behavior Therapy and Experimental Psychiatry, 45(4), 427–434. Not drawn on.
• Hottenrott, B., et al. (2011). Assoziationsspaltung als Zusatzintervention für Patienten mit Zwangsstörung: eine Fallbeschreibung. Not drawn on.
• Jelinek, L., et al. (2018). "Association splitting" versus cognitive remediation in obsessive-compulsive disorder: a randomized controlled trial. Journal of Anxiety Disorders. Not drawn on.
• Rodríguez-Martín, B. C., Moritz, S., Molerio-Pérez, O., & Gil-Pérez, P. (2013). Effectiveness of association splitting in reducing unwanted intrusive thoughts in a nonclinical sample. Behavioural and Cognitive Psychotherapy. Not drawn on.
• Musiat, P., Moritz, S., Jacobi, C., & Schmidt, U. (2014). Association splitting: feasibility study of a novel technique to reduce weight and shape concerns. Eating and Weight Disorders, 19(2), 153–158. Abstract only.
• Moritz, S., Jelinek, L., Hauschildt, M., & Naber, D. (2010). How to treat the untreated: effectiveness of a self-help metacognitive training program (myMCT) for obsessive-compulsive disorder. Dialogues in Clinical Neuroscience, 12(2), 209–220. Not drawn on.
• Ching, T., Jelinek, L., Hauschildt, M., & Williams, M. (2020). Association splitting for obsessive-compulsive disorder: a systematic review. Current Psychiatry Research and Reviews, 15(4), 248–260. Not drawn on.