MENTAL MODEL · M0019

Catastrophic Forgetting

Learn & MetacognitionHigh supportCognitive Science

Included

Version 1.0.0 · Updated 2026-07-28

CORE DEFINITION

Catastrophic interference, also known as catastrophic forgetting, is the tendency of an artificial neural network to abruptly and drastically forget previously learned information upon learning new information.

SCAFFOLDING EFFECT

psychology

Reduce cognitive load

Evaluate old and new tasks during continual training so improvements on the new task do not conceal lost capabilities. Business transformation is an analogy, and EWC is not the only remedy.

anchor

Anchor fast decisions

In neural networks (especially in continual learning), the phenomenon where knowledge of old tasks is drastically forgotten after learning new tasks, because weights are overwritten by new data, destroying original representations; it is a core challenge in continual learning.

MINIMUM ACTION

In progress 0/3

Practice this model in one real situation:

In continual training, use regularization (e.g., EWC), replay old samples, or reserve sub-networks to protect old knowledge
evaluate the trade-off between new and old task performance
avoid direct overwriting fine-tuning.

Check to track your progress (stored locally)

Learning progress0%

account_treeGenealogyexpand_more

menu_bookReferencesexpand_more

Source support: Explicit

link
en.wikipedia.orghttps://en.wikipedia.org/wiki/Catastrophic_interferenceEN · Explicit
verified

RELATED MODELS