New Theory of Disuse

In the AI Story

The theory's most provocative claim reframes forgetting as adaptive rather than pathological. Forgetting is not the decay of stored information but the reduction of retrieval strength—a resource-management operation that suppresses currently irrelevant items to improve access to currently relevant ones. A memory system that never forgot would drown in interference, unable to prioritize. The brain's 'forgetting' is analogous to a search engine's ranking algorithm: both suppress some results to make others more accessible. The analogy illuminates AI's cognitive impact—when external systems maintain permanent retrieval strength for all information, the brain's adaptive suppression mechanism loses its function.

Bjork and Bjork's framework predicted, with uncomfortable precision, what large-scale AI adoption would do to human memory architecture. When AI maintains external retrieval strength for any information the user might need, the conditions that build storage strength—effortful retrieval after partial forgetting—never occur. The user develops a specific cognitive profile: permanently high retrieval strength (via the tool) and permanently low storage strength (because the conditions for building it have been systematically eliminated). The profile feels like competence—the user always has the answer—but lacks the internal architecture that genuine understanding requires.

The theory distinguished three routes to building storage strength, each corresponding to a desirable difficulty. First, spaced retrieval after partial forgetting—the effortful reconstruction from a degraded trace deposits deeper encoding than maintenance from a fresh trace. Second, generative processing—producing an answer activates associative networks that receiving an answer does not engage. Third, varied encoding contexts—learning the same material under different conditions builds flexible representations that single-context learning cannot produce. AI tools, in their default design, bypass all three routes.

Origin

The New Theory of Disuse appeared in 1992 as a chapter in the festschrift for William K. Estes, Bjork's doctoral mentor at Stanford. The theory emerged from Bjork's three-decade engagement with retrieval inhibition, interference, and the mechanisms by which memory access is regulated. The framework built on earlier two-process theories (short-term versus long-term memory, familiarity versus recollection) but rejected their anatomical commitments in favor of a purely functional distinction grounded in retrieval dynamics.

Elizabeth Ligon Bjork's collaboration was decisive—her expertise in metacognition and judgments of learning provided the empirical foundation for understanding how learners assess their own memory strength and why those assessments systematically diverge from actual retention. The joint research program demonstrated that learners' confidence tracks retrieval strength, not storage strength, explaining why massed practice (which maximizes current retrieval strength) feels more effective than spaced practice (which builds storage strength at the cost of current retrieval ease).

Key Ideas

Two strengths, one memory. Every item in memory has storage strength (depth of encoding, associative richness) and retrieval strength (current accessibility)—independent dimensions that common intuition and most educational practice conflate into a single 'memory strength.'

Forgetting is suppression, not decay. Retrieval strength declines not because stored information erodes but because competing items interfere, and the brain adaptively reduces retrieval strength for currently irrelevant information—a curation mechanism AI disrupts by maintaining external permanent access.

Desirable difficulties build storage strength. Spacing, generation, and interleaving reduce current retrieval strength (making practice feel harder) while building storage strength (making retention stronger)—the mechanism resolving the paradox that difficulty during learning predicts success during application.

External retrieval prevents internal encoding. AI tools that maintain permanent external retrieval strength eliminate the drop in retrieval strength that triggers effortful re-learning—preventing the cycle of forgetting and re-retrieval that builds storage strength, leaving users with access but not understanding.