Implosive therapy

1. Introduction

Version followed: Stampfl's therapist-directed implosive therapy (IT) as set out by Stampfl & Levis (1967) and described by Levis in later chapters (Levis 1980; Levis & Castelda 2005), plus a short note on Levis's 1990s memory-reactivation modification (Levis 1993). The primary chapters are not drawn on in full. Procedural detail therefore comes from opened abstracts, encyclopedia excerpts and secondary summaries, checked against the standard accounts of the method. Claims that rest on unopened text are tagged [uncertain].

Implosive therapy is an imaginal exposure method. Thomas G. Stampfl developed it in the late 1950s, withheld publication for about ten years while it was tested, and later refined it with Donald J. Levis. It was the first approach to systematically expose patients to the very cues their symptoms help them avoid [source: Levis & Castelda 2005, SAGE excerpt].

How change happens: IT rests on a revised version of Mowrer's two-factor theory. Neutral cues paired with pain or deprivation come to elicit fear. Symptoms are avoidance responses, and the relief they bring reinforces them, which blocks extinction. In IT the therapist vividly and repeatedly presents those conditioned cues in imagery without the real aversive event, keeps fear high, and does not allow avoidance. The emotional drive behind the avoidance then extinguishes, and the symptoms lose their function [source: Boudewyns & Shipley 1983 chapter abstract; Levis & Castelda 2005]. The cues include the patient's reported triggers and the therapist's hypotheses about the original conditioning events, which are often psychodynamic in theme. Levis called IT a "learning-theory-based psychodynamic behavioral therapy" (the subtitle of Stampfl & Levis 1967).

IT is the historical forerunner of imaginal flooding and prolonged exposure. Most controlled studies used only the symptom-contingent part of the method [source: Emmelkamp, SAGE 'Flooding' excerpt]. Current exposure guidance differs from classic IT in several ways (see §5).

2. Framework

IT has no published fixed session-by-session manual. The phases below follow the sequence in Stampfl & Levis's accounts. Session counts are as reported, and the phase boundaries are partly reconstructed [uncertain].

Phase A: Diagnostic interviews (typically 1–2 sessions) [source: EBSCO Research Starter; Levis 1980 [uncertain]]

Goal: identify symptom-contingent cues (the situations and stimuli present when the symptom starts), the patient's reportable thoughts, feelings and bodily sensations, and history (childhood relationships, punishments, losses, conflicts). From these the therapist forms hypotheses about the unreported cues lower in the avoidance chain.
Output: a working list of cues ordered into an avoidance serial cue hierarchy (see §4).
Move on when the therapist has enough material for a first scene set and the patient has consented to the procedure.

Phase B: Preparation (part of a session) [source: Stampfl & Levis 1967 as summarised in secondary sources [uncertain]]

Goal: give the rationale and gain informed consent. Train and test the patient's imagery with neutral scenes. Set the rules: imagine as if it is happening now, report and stay with feelings, do not avoid.
Move on when the patient can produce vivid imagery on cue and accepts that sessions will be deliberately distressing.

Phase C: Implosive sessions (reported range 1–15 one-hour sessions, often extended) [source: EBSCO Research Starter; Emmelkamp, SAGE]

Goal: extinguish fear to each cue level, working along the hierarchy. Start with symptom-contingent cues, then reportable internal cues, then hypothesized cues.
Format: individual, therapist-narrated imagery, patient seated with eyes closed. Scenes are dramatic, multi-sensory and often exaggerated beyond realism. Each scene is repeated within the session until anxiety falls on its own.
Session length: clinical flooding sessions of up to several hours have been reported. A single scene may run more than an hour. Sessions are not ended while fear is at its peak [source: Emmelkamp, SAGE; EBSCO].
Move to the next cue level when the current scenes no longer produce observable distress or reported anxiety across repetitions.

Phase D: Homework and generalisation (runs alongside Phase C) [source: EBSCO]

Goal: the patient rehearses the scenes between sessions so that extinction continues and generalises.
End treatment when the scenes, including newly surfaced material, no longer elicit marked anxiety and the target avoidance or symptom has reduced in daily life. Formal termination criteria are [uncertain].

Levis's modification (1990s) [source: Levis 1993 abstract]: an extension designed to reactivate past traumatic memories presumed to underlie the pathology. It was described as yielding new findings, but its procedure is [uncertain] (see §4.15).

3. Guiding principles

[source: Stampfl & Levis 1967 and Levis & Castelda 2005 as summarised in the opened excerpts; one-line meanings are synthesis]

Symptoms are avoidance. Ask what each symptom lets the patient avoid. That stimulus is your scene material.
Extinction = cue without the aversive outcome. Present the feared cues fully, but never deliver real harm. Imagery is the vehicle.
Maximise anxiety. Aim for the highest arousal the patient can sustain. A mild scene that does not evoke fear does not extinguish anything [source: Emmelkamp, SAGE].
Block avoidance. Notice when the patient breaks the image, switches topic, opens their eyes or tones the scene down, and bring them back into it.
Repeat until it subsides on its own. Stay with a scene, or re-present it, until fear drops without the patient escaping. Ending at the peak may strengthen fear through escape relief [source: Emmelkamp, SAGE].
Work along the cue chain. Start with the accessible symptom-contingent cues. Extinguishing them exposes cues further back in the chain, which you present next (the avoidance serial cue hierarchy) [source: Levis 1980 [uncertain]].
Hypotheses are tested by response. Keep cues that produce visible distress (flushing, sweating, restlessness, protest) and expand them. Drop cues that produce nothing [source: EBSCO].
Realism is not required. Exaggerated or surreal elaboration can reach the emotional core faster than literal accuracy [source: EBSCO].
Therapist-directed. The therapist builds and narrates scenes and decides pacing. The patient's job is to imagine, feel and stay [source: Levis & Castelda 2005 title].
Use what emerges. Memories, images or associations the patient produces spontaneously become new scene material [source: EBSCO].
Practise between sessions. Homework rehearsal extends exposure beyond the room [source: EBSCO].
Hypothesized dynamic content is a stimulus, not a truth claim [synthesis]. Present hypothesized themes as imagery to test. Never tell the patient that they occurred. This guards against suggestion (see §5).

4. Intervention inventory

The original sources name these components. Step-by-step wording is my operationalisation [synthesis] unless a citation says otherwise. All example language is illustrative and is not Stampfl's or Levis's wording.

4.1 Screening, rationale and informed consent

Purpose / when: before any implosive scene. Stampfl & Levis gave patients a rationale. The structured screening and consent steps here are modern exposure practice [synthesis].
Steps:
1. Screen out or postpone IT if any of these apply: current or past psychosis or mania; active suicidal intent or recent serious self-harm; severe dissociation; substance dependence or intoxication; unstable cardiac, respiratory or seizure conditions, or pregnancy (get a medical opinion when unsure); current crisis or ongoing danger. Psychotic reactions, including two in patients with no psychotic history, were reported in a practitioner survey [source: Shipley & Boudewyns 1980]. The contraindication list itself is [synthesis].
2. Explain the learning model in plain terms: avoidance keeps fear alive, and facing the cues in imagery without real harm lets the fear burn out.
3. State the costs honestly. Sessions are meant to be very distressing, may run long, may include imagery that goes beyond what really happened, and symptoms can flare briefly.
4. Offer alternatives, such as graded exposure, prolonged exposure or systematic desensitisation.
5. Agree on a stop signal that ends the session (distinct from ordinary distress, which you will ask them to stay with). Agree on session length with time to settle afterwards. Arrange for no driving or demanding tasks immediately after the first sessions.
6. Get documented consent, including consent to therapist-introduced themes if you will use hypothesized cues (4.7).
Example language (illustrative): "This works by having you stay with the images you've been avoiding until the fear runs out. It will feel bad, maybe very bad, for a while, and it won't harm you. If you need to stop completely, raise your hand. Otherwise I'll ask you to stay in it."
Watch for: consent given under pressure or to please you; a stop signal used as routine avoidance (see §5); medical or psychiatric red flags that emerge later.

4.2 Diagnostic interview for cue identification

Purpose / when: sessions 1–2, to build scene material [source: EBSCO; Levis 1980 [uncertain]].
Steps:
1. Map each symptom. When does it start, where, with whom? What is seen, heard, smelled or felt just beforehand?
2. Ask what the patient fears would happen if they could not perform the symptom or avoidance. This often points to the next cue back in the chain.
3. Gather the thoughts, images, bodily sensations and feelings they can report in those moments.
4. Take a history of punishments, rejections, losses, injuries, humiliations, family conflict and attitudes to anger and sexuality. This is the raw material for hypothesized cues [source: EBSCO].
5. Note non-verbal reactions as topics come up (flushing, voice change, restlessness). These mark likely high-value cues.
Example language (illustrative): "Take me to the last time it happened. Just before you [checked/left/froze], what was in front of you? What went through your mind? And if you hadn't been able to [avoid], what's the worst thing that would have come next?"
Watch for: vague, intellectualised accounts (ask for concrete sensory detail); topics the patient steers away from (candidate cues); disclosures of current abuse or risk, which take priority over treatment.

4.3 Classifying cues and building the avoidance serial cue hierarchy

Purpose / when: after the interviews and updated throughout. This sets the order of scenes [source: Stampfl & Levis 1967 cue categories via Emmelkamp, SAGE, and EBSCO; hierarchy concept Levis 1980 [uncertain]].
Steps:
1. Sort the cues into three classes. (a) Symptom-contingent: external situations tied to symptom onset. (b) Reportable internal: thoughts, images and sensations the patient can describe. (c) Hypothesized, or unreportable: cues the therapist infers lie further back in the chain. Stampfl's typical themes include aggression, guilt, punishment, rejection and loss of control [source: Emmelkamp, SAGE].
2. Order the cues as a chain from the symptom backwards, with the most accessible cues first and the cues presumed closest to the original aversive events last.
3. Draft 2–4 scenes per level. Each scene should hold sensory detail, the patient's own action, and a feared consequence that unfolds without escape.
4. Revise after every session in light of observed reactions and newly reported material.
Example language: none. This is therapist planning.
Watch for: hypotheses built on theory alone rather than on the patient's history and reactions; a hierarchy that keeps growing without earlier levels being extinguished.

4.4 Imagery training with neutral scenes

Purpose / when: in the preparation phase, to check imagery ability and teach the patient to imagine "as if now." Neutral-scene imagery is documented as the placebo control in IT research [source: Prochaska 1971]. Its use as a training step in the clinical procedure is reported in Levis 1980 [uncertain].
Steps:
1. Eyes closed. Describe an everyday neutral scene in the present tense, covering several senses.
2. Ask the patient to describe what they see and to rate vividness (for example 0–10).
3. Coach them to imagine from inside their own body (first person) and to add detail themselves.
4. Repeat with 1–2 more scenes until vividness is good.
Example language (illustrative): "You're in your kitchen now. Feel the floor under your feet, hear the kettle starting. What colour is the mug in your hand? Tell me what you see."
Watch for: poor imagery (use slower narration, more sensory prompts, or the patient's own descriptions); patients who watch themselves from outside (bring them back to first person).

4.5 Symptom-contingent cue scenes

Purpose / when: the first implosive scenes. They target the external situations that trigger the symptom. Most controlled IT and flooding studies used only this cue class [source: Emmelkamp, SAGE].
Steps:
1. Eyes closed. Narrate in the present tense, second person, with rich sensory detail. Start at the trigger situation.
2. Have the patient stay in the situation without the usual escape, check or ritual. Escalate to the feared outcome unfolding.
3. Narrate with dramatic pace and emotional tone. Keep fear high and avoid soothing language.
4. Hold or loop the scene until signs of fear clearly fall. Then re-present it, and repeat until it no longer evokes much fear (4.11).
Example language (illustrative, for contamination fear): "You're touching the handle. It's sticky. You can't wash. It's on your fingers, spreading up your wrist. You touch your face. Stay with it. Feel the dread rising. Don't push it away."
Watch for: rising arousal followed by a spontaneous drop within the scene (the change process); flat responding (increase vividness or move to the next cue level); subtle neutralising in imagery, such as "I imagined washing afterwards."

4.6 Reportable internal cue scenes

Purpose / when: once external scenes lose their power. These scenes add the thoughts, images and bodily sensations the patient reports [source: Stampfl & Levis cue classes via secondary sources [uncertain]].
Steps: weave the patient's feared thoughts and bodily sensations into the scene as present experience, for example the racing heart, the thought that they will lose control, or the image of others' contempt. Let these escalate to their feared conclusion inside the scene.
Example language (illustrative): "Your heart is pounding harder. The thought comes: 'I'm going to lose it in front of everyone.' And you do. Everyone turns to look. Feel their eyes on you."
Watch for: internal cues that produce more fear than the external ones, which suggests they sit further back in the chain; spontaneous associations to the past (4.13).

4.7 Hypothesized (dynamic) cue scenes

Purpose / when: the distinctive element of Stampfl's method. Scenes present cues the therapist infers but the patient has not reported, drawn from the history and from themes such as aggression toward significant others, guilt, punishment, rejection and loss of control [source: Emmelkamp, SAGE]. Stampfl and colleagues considered these cues more potent because they are presumed closer to the original conditioning events [source: EBSCO]. Evidence for adding them over symptom cues is mixed [source: Smith et al. 1973; Prochaska 1971].
Steps:
1. Use them only with specific prior consent (4.1), and only after symptom-contingent and internal cue levels are largely extinguished.
2. Base each theme on material from the patient's own history and observed reactions, not on theory alone.
3. Frame the scene explicitly as imagination, not as a claim about what happened [synthesis].
4. Present it like other scenes, with vivid, escalating detail, and watch for the cue-validation signs in 4.10.
5. Keep sexual and abuse-related content at the level of themes and emotions (shame, exposure, punishment). Do not introduce specific assault scenarios the patient has not reported [synthesis; see §5 on suggestion risk].
Example language (illustrative, rejection and punishment theme): "Let yourself imagine this, even if it never happened. You've made your father furious. He turns away and won't look at you. You're shut out. Feel how small and alone you are."
Watch for: strong emotion, sudden silence or protest (expand the theme); no response (drop it); the patient taking the scene as a revealed memory (correct this gently, see §5).

4.8 Exaggeration and surreal elaboration

Purpose / when: when realistic scenes plateau. Implosive scenes deliberately go beyond realism, as in the snake-phobia example of being swallowed by a huge snake [source: EBSCO].
Steps: take the feared consequence past plausibility (magnified size, repetition, bodily invasion, catastrophic social exposure) while keeping it tied to the patient's actual cues. Tell the patient beforehand that scenes will sometimes be unrealistic on purpose.
Example language (illustrative): "The spider is getting bigger, as big as the room. Its legs are closing round you. You can't move."
Watch for: amusement or detachment, meaning the exaggeration has broken immersion (go back toward realism); gratuitous gore unrelated to the patient's fear (avoid it [synthesis]).

4.9 Preventing avoidance during scenes

Purpose / when: throughout every scene. Avoidance responses are not allowed during flooding [source: Emmelkamp, SAGE].
Steps:
1. Look for escape: eyes opening, changing the topic, intellectualising, softening the image, rescue fantasies, joking, falling silent.
2. Redirect briefly and firmly back into the scene, and add a detail to restore vividness.
3. Ask for brief reports ("What's happening now?") to keep the patient engaged without stepping out of the scene.
Example language (illustrative): "You've stepped out of it. Go back. You're still in the room, it's still there. Stay."
Watch for: frequent escapes, which mean the level is set too high (step back down the hierarchy rather than force it [synthesis]).

4.10 Cue validation by observed response

Purpose / when: continuously. This is how the therapist finds which cues matter [source: EBSCO].
Steps:
1. Watch for signs of distress such as sweating, restlessness, flushing, tearfulness, breath changes, and negative or angry comments.
2. Elaborate and repeat cues that produce strong reactions. Drop cues that produce none.
3. Record which elements triggered peaks, and use them to revise the hierarchy (4.3).
4. Add subjective ratings, for example a 0–100 distress scale every few minutes [synthesis, modern exposure practice].
Example language (illustrative): "What's the number right now, zero to a hundred?"
Watch for: a gap between visible distress and the reported rating (trust both and ask about the difference).

4.11 Within-session repetition to extinction, and session closure

Purpose / when: for every scene. Exposure continues until fear subsides on its own, and stopping early may strengthen fear [source: Emmelkamp, SAGE; EBSCO].
Steps:
1. Plan for long sessions (book 90 minutes or more where possible [synthesis]).
2. Repeat or extend the scene until distress clearly falls from its peak, for example below half of the peak rating [synthesis].
3. Do not end at the peak. If time runs out, continue the same scene at a level that brings descent before stopping [synthesis].
4. Close with eyes open, orientation to the room, a brief review of what changed, and the homework plan. Before the patient leaves, check that they are settled and that any risk remains unchanged [synthesis].
Example language (illustrative): "Notice it's lower than ten minutes ago, though nothing changed in the scene. That's the fear running out."
Watch for: fear that falls from session to session (the expected pattern); fear that stays high over repeated sessions (see §5).

4.12 Patient verbalisation and role enactment within the scene

Purpose / when: to deepen involvement once the patient can sustain imagery. Secondary accounts describe patients visualising and acting out the scenes, and being told to "lose" themselves in them [source: EBSCO]. The exact original instructions were not verified [uncertain].
Steps:
1. Invite the patient to say aloud what they feel and what they would say to figures in the scene, staying seated with eyes closed.
2. Feed their words back into the narration so the scene follows their reactions.
3. Keep the therapist in charge of direction, so that the patient's additions do not turn into escape routes.
Example language (illustrative): "Say it to her, out loud, what you want to say." Then: "And she just stares back at you."
Watch for: the patient's enactment turning into rescue or revenge fantasies that cut fear short (return to the feared outcome).

4.13 Incorporating spontaneously recalled material

Purpose / when: whenever new memories, images or associations surface. Secondary accounts report a patient recalling a childhood humiliation mid-treatment, which was then built into later scenes [source: EBSCO].
Steps:
1. Note the material without interpreting or validating it as historical fact.
2. Ask for the patient's own sensory description, and build the next scene from their words.
3. Place it in the hierarchy where the patient's reaction suggests.
Example language (illustrative): "You mentioned the classroom. Tell me what you see there, in your own words, and we'll work with that next time."
Watch for: material shaped by your own earlier hypothesized scenes (see suggestion risk in §5).

4.14 Homework rehearsal

Purpose / when: between sessions, from the first implosive session onward. Patients rehearse the disturbing scenes outside therapy [source: EBSCO]. Research versions delivered IT scenes by tape recording [source: Prochaska 1971].
Steps:
1. Assign one scene the patient has already worked through in session, ideally one partly extinguished. Do not assign new or top-level scenes [synthesis].
2. Agree on a daily slot (for example 30–45 minutes [synthesis]) in a private place, and have them stay with the scene until distress falls.
3. A recording of your in-session narration can be used, with the patient's consent and stored securely [synthesis, based on taped delivery in the research].
4. Have them log distress at start, peak and end, and review the log next session.
Example language (illustrative): "Each day, play through the scene we did today and stay with it until it drops by about half. Note the numbers for me."
Watch for: homework turning into avoidance, such as skipping, half-imagining or stopping at the peak (see §5); distress that does not fall when practising alone (go back to in-session work).

4.15 Levis's memory-reactivation modification (described, not operationalised)

Purpose / when: Levis reported a modification that reactivates past traumatic memories presumed to underlie the symptoms. He described it as producing findings that challenge prevailing beliefs [source: Levis 1993 abstract].
What is known: the procedural detail is in sources not drawn on here (Levis 1993 full text and later chapters) [uncertain]. No steps are given, because none could be verified.
Watch for: any method aimed at "recovering" memories carries a documented risk of creating false memories. If you pursue this, get the primary texts and specialist supervision. Current trauma-focused practice works only with memories the patient already reports [synthesis].

5. Handling problems

The sources say little about troubleshooting. Beyond the points cited, everything here is [synthesis] drawn from general exposure practice.

Fear does not drop within a session.
- Check for hidden avoidance, such as neutralising images, dissociating, or "watching from outside." Fix it with 4.9 and first-person prompts.
- Extend the session and keep the scene going. Ending at the peak risks strengthening fear [source: Emmelkamp, SAGE].
- If fear still will not fall over 2–3 sessions, the cue level is probably wrong. Step back a level, or re-examine the hierarchy.
Stuck: scenes produce little emotion.
- Raise vividness with present tense, more senses and slower pacing.
- Use the patient's own words and images.
- Try the next cue class, from symptom-contingent to internal to hypothesized (4.5–4.7).
- Test new hypotheses and drop those that produce nothing (4.10).
- Check for intoxication, sedation or over-control.
Resistance or refusal to continue.
- Revisit the rationale and the patient's goals, and acknowledge the cost.
- Offer a lower starting level or a gentler format. Systematic desensitisation and graded or prolonged exposure are legitimate alternatives, and switching is not failure.
- Never coerce. Distress during a scene is expected, but the patient's stop signal must always be honoured.
Overwhelm, panic or dissociation.
- Brief panic reactions were rare but did occur in practitioner survey data [source: Shipley & Boudewyns 1980].
- Distinguish intense but engaged fear (continue) from loss of contact (eyes glazed, unresponsive, disoriented).
- For loss of contact: open the eyes, ground the patient (orientation, name the date, feel the feet), then resume at a lower intensity once they are present. Stopping entirely risks reinforcing escape, so where possible end with a manageable scene that brings fear down.
Psychotic reaction or marked deterioration.
- Acute psychotic reactions were reported, including in patients without a known psychotic history [source: Shipley & Boudewyns 1980].
- Stop IT, assess mental state and risk, arrange psychiatric review, and do not resume flooding without specialist advice.
Suicidal ideation emerging.
- Pause exposure. Do a standard risk assessment covering ideation, intent, plan, history and protective factors, and build a written safety plan.
- Where relevant, discuss lethal-means safety: storage away from the home, removal, or holding by a trusted person.
- Resume only when risk is stable, and consider a less intense method.
Non-adherence to homework.
- Find out what happened: skipped, half-done, or stopped at the peak.
- Reduce the dose (shorter, an easier scene), provide a recording, link practice to a daily routine, and do the first rehearsal together at the end of the session.
- Treat avoidance of homework as data about the cue hierarchy.
Rupture: the patient feels attacked or that the therapist is cruel.
- IT casts the therapist as the narrator of the feared material, so ruptures are predictable.
- Step out of the scene and name it: "It makes sense you're angry with me; I've been putting you through this on purpose."
- Re-link the method to the patient's goals, and give the patient more say over the next scenes.
- Repair before you resume.
Distress over hypothesized content ("Why would you think that about my father?").
- Clarify that hypothesized scenes are imagined stimuli, not interpretations or accusations.
- Drop themes that are offensive or culturally inappropriate and did not produce an emotional response.
- Mixed evidence that dynamic cues add value [source: Smith et al. 1973] makes symptom-contingent work a defensible default.
Suggestion and false-memory risk.
- Therapist-introduced scenes about childhood sexual or physical abuse, and memory "reactivation" methods (4.15), can create beliefs or images that later feel like memories.
- Never present unreported abuse as fact. Do not introduce specific abuse scenarios. Document that hypothesized content was framed as imagination. Avoid asking repeatedly "what else might have happened."
- If the patient begins to treat scenes as recovered memories, discuss the uncertainty openly and get consultation.
Patient with PTSD or a known trauma.
- Use current trauma-focused protocols, such as prolonged exposure or CPT, which work with the patient's own reported memory rather than therapist-hypothesized content.
- IT-style exaggeration and invented content are not recommended for trauma memories.
Symptoms return after a gap.
- Expect some return of fear. Run booster scenes, rehearse the original scenes and extend practice to new contexts.
Medical concerns in session (chest pain, fainting).
- Stop and assess. Seek medical help when needed. Check the original screening (4.1).

Sources

Sources:

Further reading (not drawn on) [uncertain where relied on]:

Stampfl, T. G., & Levis, D. J. (1967). Essentials of implosive therapy: A learning-theory-based psychodynamic behavioral therapy. Journal of Abnormal Psychology, 72, 496–503.
Levis, D. J. (1980). Implementing the technique of implosive therapy. In A. Goldstein & E. B. Foa (Eds.), Handbook of Behavioral Interventions. Wiley.
Stampfl, T. G., & Levis, D. J. (1973; 1976). Chapters on implosive therapy (as listed in Levis 1993 references).
Boudewyns, P. A., & Shipley, R. H. (1983). Flooding and Implosive Therapy: Direct Therapeutic Exposure in Clinical Practice. Plenum.

← Previous · II-014Flooding Next · II-016 →Inhibitory-learning exposure