Working Memory Taxation and EMDR 2.0: The Mechanism Behind the Modification

Working memory is the small, short-term system that holds and manipulates information you're actively using — a phone number you're about to dial, a sentence you're composing, an image you're picturing. Baddeley's classic model breaks it into a visuospatial sketchpad (images, locations), a phonological loop (verbal material), and a central executive that allocates attention between them. The system has a very limited capacity: roughly a handful of items at once, and only one attention-demanding task can be fully held at a time.

That capacity limit is the whole engine of the bilateral stimulation mechanism. When a client holds a trauma image in the visuospatial sketchpad and simultaneously performs a task that also uses the sketchpad (tracking a moving dot, following taps), the two compete for the same limited slot. The image loses fidelity — colors dull, edges soften, sound and detail drop out — and the emotional charge that rode on that vividness drops with it (van den Hout & Engelhard, 2012).

Traumatic memories are stored with unusually vivid sensory detail. Recalling them re-activates that detail. If, during recall, working memory is simultaneously loaded by a second attention-demanding task, the memory image is degraded during re-encoding. When it is stored back down, it is stored in this lower-vividness, lower-emotion form. Repeat this across sets and the memory progressively loses its charge. This is the working memory taxation model, and it is the leading mechanism-of-action account of EMDR (Baddeley, 1986; Andrade, Kavanagh & Baddeley, 1997; van den Hout & Engelhard, 2012).

Three findings from the taxation literature matter clinically.

1. Any sufficiently demanding dual task reduces vividness and emotionality. Eye movements, tapping, mental arithmetic, playing Tetris, complex drawing tasks — all produce the effect, so long as they draw on the same working-memory sub-system as the image (Andrade, Kavanagh & Baddeley, 1997; Engelhard et al., 2010).

2. There is a load sweet spot. Too little load and the image is unchanged; too much and the client loses the image entirely and no reprocessing happens (Engelhard, van den Hout & Smeets, 2011). This inverted-U pattern is why titration matters.

3. Meta-analytic support is solid. Mertens et al. (2019) reviewed the taxation literature and concluded the effect is robust across modalities and populations. This doesn't rule out complementary mechanisms (REM-like consolidation, orienting response, parasympathetic shift) — it just gives us the clearest, most testable account so far.

EMDR 2.0, developed by Ad de Jongh and Suzy Matthijssen and colleagues in the Netherlands, takes the working memory model seriously and asks: if taxation is the mechanism, can we tax harder, more variably, and more deliberately — and get faster, more efficient reprocessing? Their empirical answer, refined across a series of studies (Matthijssen et al., 2021; Matthijssen, Heitland, Verhoeven & van den Hout, 2019), is yes.

EMDR 2.0 is a modification of Phase 4 desensitization. Phases 1, 2, 3, and 5–8 are unchanged. What changes in Phase 4:

– Higher-taxation dual tasks. Instead of only horizontal eye movements at moderate speed, EMDR 2.0 uses harder or more variable tasks: fast eye movements in multiple directions, mental arithmetic, backwards counting, category generation, dual-modality combinations.

– Shorter sets. Because load is higher, sets are shorter — often 20–30 seconds — before checking in.

– Variable modalities. Tasks are rotated deliberately across sets so the brain doesn't habituate to any one demand.

– Explicit titration. The therapist actively grades load up or down based on the client's SUD, tolerability, and whether the image stays available.

See the EMDR 2.0 overview for the full protocol, or the EMDR 2.0 tools page for the specific dual-task menu.

Engelhard, van den Hout & Smeets (2011) showed the effect follows an inverted U: too little load and vividness doesn't drop; optimal load and it drops sharply; excessive load and the client stops being able to hold the image at all, which halts reprocessing. Practical titration signals in session:

– Client can hold the image, SUD is dropping across sets → load is in the sweet spot. Continue.

– Client can hold the image but SUD is flat across 2–3 sets → load may be too low. Increase task difficulty or add a modality.

– Client loses the image, reports confusion, or cannot complete the dual task → load is too high. Reduce task difficulty, simplify, or drop back to standard BLS.

– Client shows dysregulation, dissociation, or window-of-tolerance breach → stop, ground, and revisit Phase 2 preparation. This is the same rule as in standard EMDR — higher taxation doesn't override it.

The EMDR 2.0 titration worksheet formalizes this decision loop.

Both protocols work through the same working-memory-taxation mechanism. The difference is the intensity and variability of the dual task in Phase 4.

For most clients the two are interchangeable. EMDR 2.0's edge shows up with clients who habituate to standard BLS or plateau in SUD.

Use these signals to grade working-memory load up or down inside a Phase 4 set. Check every 2–3 sets, not every pass.

• Increase load when SUD is flat across 2–3 sets

  Add speed, change direction, add a second modality (tactile + eye movements), or introduce a light verbal task.

• Decrease load when the client loses the image

  Simplify the task, slow it down, or drop back to standard horizontal BLS until the image is available again.

• Stop and stabilize on window-of-tolerance breach

  Ground, orient, revisit Phase 2 resources. Higher taxation never overrides safety.

• Don't push harder just because SUD is high

  High SUD is not the trigger for more load — flat SUD is. Aggressive taxation on a dysregulated client causes dissociation, not processing.

• Don't rotate modalities mid-set

  Change tasks between sets so the client's working memory can reallocate cleanly. Mid-set rotation just adds noise.

• Don't assume plateau means the target is stuck

  First rule out insufficient load. Only then consider blocking beliefs, feeder memories, or an interweave.

The mechanism is the same in both — taxation of working memory during memory recall. What differs is how aggressively that mechanism is worked.

– Standard EMDR uses moderate, mostly horizontal eye movements at consistent speed, with sets of roughly 24–30 passes and periodic check-ins. It's well-tolerated, easy to titrate down, and clinically validated across decades.

– EMDR 2.0 uses higher-load, more variable dual tasks in shorter sets, with more frequent titration checks. In head-to-head studies it has shown comparable or superior outcomes with fewer sets required to reach a low SUD (Matthijssen et al., 2021). See EMDR 2.0 vs standard EMDR for the trial data.

For most clients the two are interchangeable. EMDR 2.0's advantages are most visible with clients who habituate to standard BLS, plateau in SUD, or benefit from a higher-cognitive-load frame. Its risks are the same as standard EMDR plus one additional consideration: higher load can breach window of tolerance faster, so titration discipline matters more.

Even if you never formally adopt EMDR 2.0, the taxation model changes how you think about Phase 4:

– If a client plateaus in SUD across several sets, the first question is 'is the working memory system actually being taxed?' Increase load (speed, direction change, add tactile, add a light verbal task) before you assume the target itself is stuck or needs an interweave.

– If a client dissociates or loses the image, the first question is 'is load too high?' Reduce load before assuming the client is avoidant.

– Modality is not sacred. Eye movements, taps, and tones are all valid taxation delivery systems. Switch if one isn't working.

– The Butterfly Hug works because it taxes working memory too — the taxation model explains why self-administered BLS is useful for regulation but not for full reprocessing (too little load, no clinical structure).

Taxation is the best-supported mechanism, not the only one. It doesn't fully account for: (a) the affect and body-based shifts clients report during and after reprocessing, which point to autonomic and interoceptive changes beyond image degradation; (b) the network-level integration that seems to happen between the target and adaptive memory networks; (c) the effect of BLS on positive resource installation in Phase 5, where the goal is strengthening a positive image, not degrading it. Complementary theories — REM-like consolidation, orienting response, adaptive information processing — likely fill these gaps. Taxation is the foundation, not the whole building.

Every set in Phase 4 lands the client in one of three zones. Reading which zone you're in is a core clinical skill in EMDR 2.0.