cd /news/machine-learning/stability-does-not-mirror-plasticity… · home › topics › machine-learning › article
[ARTICLE · art-172339] src=arxiv.org ↗ pub= topic=machine-learning verified=true sentiment=· neutral

Stability Does Not Mirror Plasticity: A Module-Level View of Declarative and Procedural Learning in Transformers

A new arXiv paper (2611.23427v1) reports a double dissociation in how transformer modules handle continual learning: query-key (QK) and value-output (VO) projections are the most plastic on procedural tasks, while MLPs are the most plastic on declarative memorization. Stability does not mirror plasticity — when learning a new procedure, updating QK retains over half of stored declarative memory, updating MLPs retains far less, and updating VO loses roughly four-fifths. The authors attribute the asymmetry to plasticity tracking which module best suits a task's primary computation or storage, while stability depends on all the ways an existing capability relies on that module, including access to information stored elsewhere.

recommendation

For continual learning on procedural tasks, prefer updating QK projections over MLPs or VO: the paper measured QK retaining over half of declarative memory while VO updates lost roughly four-fifths.

more in playbook: Fine-tuning and data
by read1 min views2 publishedNov 19, 2026

arXiv:2611.23427v1 Announce Type: new Abstract: Plasticity and stability in continual learning are typically studied at the entire-model level and in a task-agnostic manner, yet prior work has reported divergent conclusions across settings. Inspired by cognitive distinctions between declarative and procedural memory, we instead study how plasticity and stability vary across transformer modules and task types. In a controlled synthetic environment with clean measurement, we show a double dissociation: query-key (QK) and value-output (VO) projections are the most plastic on procedural tasks, whereas MLPs are the most plastic on declarative memorization. Stability, however, does not simply mirror plasticity: when learning a new procedure, updating QK retains over half of the stored declarative memory, while updating MLPs retains far less and updating VO loses roughly four-fifths. Targeted module reset-and-recovery experiments explain this asymmetry: plasticity tracks which module is best suited to a task's primary computation or storage, whereas stability depends on all ways the existing capability relies on that module, including access to information stored in other modules. Experiments on a realistic model and tasks broadly echo these asymmetric dynamics. Overall, our work shows that plasticity and stability depend on different aspects of a module's functional role, offering a task-dependent, module-level perspective on continual learning.

── more in #machine-learning 4 stories · sorted by recency
── more on @arxiv 3 stories trending now
sponsored brought to you by zahid.host 4,200+ EU-deployed projects
reading about agents? ship yours in a single git push.

Run your AI side-project on zahid.host

EU-based hosting, git-push deploys, automatic HTTPS, no cold starts. Free tier with a custom domain — perfect for shipping the agent you just read about.

$git push zahid main
→ Live at https://your-agent.zahid.host ✓
Get free account → Pricing
from €0/mo · no card required
LIVE [news/stability-does-not-m…] indexed:0 read:1min 2026-11-19 · —