Mechanisms
Memory Systems · Non-Declarative
Procedural Memory
Example
You write differently to your manager than to your closest colleague, using a different vocabulary, different hedging, and different amount of context supplied. You never learned the rule and couldn't state it now, but the shift happens before the first sentence.
Procedural Memory in Humans
Definition
Acquiring a skill or regularity through repetition without being able to report what was learned. The system is dispositional rather than representational: it is expressed through performance rather than recollection, and it is revealed by reactivating the system in which the learning originally occurred (Squire, 2004). Its scope is far wider than physical skill. Motor learning was initially set aside as a special, less cognitive case, but it turned out to be one member of a broader class of skill-like abilities preserved in amnesia, including perceptual skills such as mirror reading, cognitive skill learning, artificial grammar learning, and category learning; the subtype covering these is labelled skills and habits, and it is separated in the taxonomy from priming, from simple classical conditioning, and from non-associative learning (Squire, 2004). The clearest demonstration that it dissociates from declarative memory comes from a probabilistic classification task in which amnesic patients learned normally despite severely impaired declarative memory for the training episode, while patients with Parkinson's disease failed to learn despite intact memory for the training (Knowlton, Mangels, & Squire, 1996).
Neural Substrate
The neostriatum, comprising the caudate nucleus and putamen, supports the gradual, incremental learning of associations that characterises habit learning. The evidence is a double dissociation in humans: on a probabilistic classification task, amnesic patients with limbic-diencephalic damage learned normally while their declarative memory for the training episode was severely impaired, and patients with Parkinson's disease, whose neostriatum is compromised, failed to learn while their memory for the training remained intact. The authors read this as showing two separate and parallel learning systems (Knowlton, Mangels, & Squire, 1996).
Later human imaging complicates the word parallel. Poldrack et al. (2001) gave healthy volunteers two versions of the same classification task with identical stimuli, one learned trial by trial from feedback and one learned as paired associates, and end-of-training accuracy did not differ between them (86.5% versus 82.7%, F(1,21) = 1.13, p = .30). The feedback version activated the caudate and deactivated the medial temporal lobe; the paired-associate version showed the reverse. Across individuals, signal change in the two regions was negatively correlated, and an event-related follow-up showed the medial temporal lobe active at the very start of learning and then rapidly deactivating as caudate activity rose. The authors interpret this as competition rather than parallel operation, with the two systems arbitrating for control of behaviour, and note that which system is engaged depends on the strategy the task invites rather than on the material itself.
AI at a Glance
Procedural memory acquires skills, habits, and regularities gradually through repeated exposure, expressed through performance rather than recollection, and it survives intact in patients whose declarative memory for the training itself is destroyed (Knowlton, Mangels, & Squire, 1996; Squire, 2004). In AI systems the function is split across three mechanisms that behave differently: procedures installed during training are demonstrably reused as procedures rather than retrieved as answers (Weights), rules supplied in conversation degrade badly under interference and worse still when several must be combined (Context), and bolting on an explicit memory layer helps inconsistently and sometimes actively interferes with immediate rule execution (Memory layer) (Qin et al., 2026; Ruis et al., 2024).
- Non-declarative memory: the umbrella term for the several memory systems that are not accessible to conscious recollection, including procedural memory, priming, and simple classical conditioning (Squire, 2004).
- Memory layer: an external store attached to a model at deployment, into which past interactions or extracted rules are written and from which they are later retrieved, as distinct from anything held in the model's weights or its current conversation.
Procedural Memory in AI
How Far It Has Been Measured
Measurement in AI: Directly Measured
AI Memory Performance: Below Human provisional
Where It Lives in the System
What Has Been Measured
- ImplicitMemBench tests large language models on procedural memory by name: 100 items give a new rule with 1 to 3 examples (e.g., a custom arithmetic operator or reversed tool-call order), then 15 turns of misleading conversation, then score only the model's first attempt on new probes; unlike human procedural tasks, the rule is stated outright and learned in one exposure rather than through practice (Qin et al., 2026).
- Five computer science PhD students completed the same items under the same protocol and scoring rubric and all scored 100%; the best model, Claude-4.1-opus, averaged 76.67% over three runs, about 23 points below the human range, even though the rule was still in the model's context window (what it can see in the current conversation) (Qin et al., 2026).
- Across 17 models, procedural memory scores ranged from 46.67% (LLaMA-3.1-8B) to 76.67%, with eight models above 70%; the human baseline is small, unrepresentative and at ceiling, so the size of the gap is uncertain but its direction is not (Qin et al., 2026).
- Saved memory did not close the gap: storing the key rule by hand in an external memory store (Mem0) moved DeepSeek-R1 only from 76.33% to 77.00%, and among memory-augmented agents MemAgent rose from 51% to 60% while MEM1 fell from 50.67% to 27% and MemGPT from 50% to 44% (Qin et al., 2026).
- A 2025 preprint ran gpt-4o and o3-mini on three classic artificial-language experiments with published human data: on a finite-state grammar task (15 people per grammar) both models improved across blocks and found the simpler grammar easier, as people did, but on a morphosyntax task neither model improved over trials, and on a morphology task only o3-mini matched human regularization (Ma & Xu, 2025).
- That comparison is approximate and does not set the rating: models got shorter versions of the tasks (e.g., 120 rather than 480 sentences), could still see the training items in the conversation while people had to remember them, and results were compared as learning patterns rather than on a common accuracy scale (Ma & Xu, 2025).
- Outside comparisons with people, AI agents (systems that act step by step toward a goal) can now save procedural memory, distilling past attempts into step-by-step instructions and scripts; this raised task success on TravelPlanner and ALFWorld, and the saved memory also improved weaker models, but no human baseline was run (Fang et al., 2026).
Where a Mechanism Can Live
04 Product
The whole configuration, and the only thing a benchmark observes directly.
01
Weights
Across deployments
02
Context Window
Within a session
03
Memory Layer
Across sessions
Which of the three a result belongs to is inferred from it, not seen in it.
Read more
Caudate nucleus
Nucleus in the dorsal striatum
A C-shaped nucleus arching alongside the lateral ventricle, the association-connected half of the dorsal striatum.
Head, body and tail follow the curve of the lateral ventricle, with the head bulging into its frontal horn. It receives from prefrontal and parietal association cortex, keeping it largely separate from the sensorimotor input that dominates the putamen. Caudate volume loss is among the earliest and most characteristic imaging findings in Huntington's disease, visible as widened ventricular horns before motor symptoms. Its activation in human studies is usually reported with feedback-based learning and with goal-directed action selection rather than movement execution.
Diencephalon
Division in the forebrain (prosencephalon)
The deep core of the forebrain (thalamus, hypothalamus, epithalamus and subthalamus), wrapped around the third ventricle.
The diencephalon sits at the top of the brainstem and beneath the cerebral hemispheres, its four parts arranged around the third ventricle: thalamus above and lateral, hypothalamus below, epithalamus behind and above, subthalamus tucked between hypothalamus and midbrain. Almost every ascending sensory pathway and most cortical–subcortical loops pass through it, which is why small lesions here produce deficits out of all proportion to their size. It is a developmental term as much as an anatomical one: the diencephalon is one of the vesicles the embryonic forebrain divides into, and the label has stuck.
Limbic system
Division
A loosely bounded set of cortical and subcortical structures ringing the corpus callosum, named before its membership was settled.
As usually drawn, the limbic system includes the hippocampal formation, amygdala, cingulate and parahippocampal gyri, fornix, mammillary bodies and anterior thalamic nuclei (Papez's 1937 circuit as elaborated by MacLean). Its membership has never been agreed: septal nuclei, orbitofrontal cortex, nucleus accumbens and the insula are counted in by some authors and out by others. Many anatomists argue the term should be retired, because the structures it groups differ in connectivity and function and because it carries MacLean's triune-brain framing, which comparative neuroanatomy does not support. It survives as convenient shorthand.
Medial temporal lobe
Cortex in the temporal lobe
The inner fold of the temporal lobe (hippocampus, the cortices wrapped around it, and the amygdala), removed together in the classic amnesia cases.
A regional grouping rather than a single structure: the hippocampal formation, the entorhinal, perirhinal and parahippocampal cortices that feed it, and, on some definitions, the amygdala. It became the canonical "memory region" because the surgeries and infarcts that produce dense anterograde amnesia (H.M. most famously) take out most of it at once. Whether the parts constitute one memory system or several with different jobs is an open argument, and lesions in patients often cross the boundaries that would settle it.
Putamen
Nucleus in the dorsal striatum
The lateral striatal nucleus, separated from caudate by the internal capsule and dominated by sensorimotor cortical input.
Lies lateral to the globus pallidus and medial to the external capsule and claustrum, forming the outer part of the lentiform nucleus. Its cortical input comes mostly from motor, premotor and somatosensory areas, arranged somatotopically, and it is the striatal territory where dopamine loss in Parkinson's disease is most severe: posterior putamen first, which fits the early motor signs, though the cognitive decline of the disease tracks cortical Lewy pathology rather than this striatal gradient. Its involvement in stimulus-response learning has been tested directly in people, where focal putaminal lesions impair rule-based category learning while sparing information-integration learning.
Striatum
Subcortical in the basal ganglia
The input structure of the basal ganglia: caudate, putamen and nucleus accumbens, receiving from nearly all of cortex.
Ninety-five percent of its neurons are GABAergic medium spiny cells, held at a hyperpolarized resting potential and firing only when converging excitatory input depolarizes them, a pattern described most clearly under anesthesia and in slow-wave sleep. They come in two classes, distinguished by dopamine receptor type and by whether they project straight to the output nuclei or by way of the external pallidum. Dopamine from the midbrain reaches all of it, and acetylcholine from large local interneurons modulates it. The striatum is the usual localization for habit formation and reinforcement-based learning, though how those relate to its role in ordinary movement initiation is unresolved.
Fang et al. (2026)
Memp: Exploring Agent Procedural Memory
In
Findings of the Association for Computational Linguistics: ACL 2026, 17490-17502
Find it
Knowlton et al. (1996)
A Neostriatal Habit Learning System in Humans
In
Science, 273(5280), 1399-1402
Abstract
Amnesic patients and nondemented patients with Parkinson's disease were given a probabilistic classification task in which they learned which of two outcomes would occur on each trial, given the particular combination of cues that appeared. Amnesic patients exhibited normal learning of the task but had severely impaired declarative memory for the training episode. In contrast, patients with Parkinson's disease failed to learn the probabilistic classification task, despite having intact memory for the training episode. This double dissociation shows that the limbic-diencephalic regions damaged in amnesia and the neostriatum damaged in Parkinson's disease support separate and parallel...
Find it
Ma & Xu (2025)
Implicit In-Context Learning: Evidence from Artificial Language Experiments
In
arXiv
In summary
Humans acquire language implicitly, absorbing complex patterns without explicit awareness, but whether LLMs show human-like pattern extraction during in-context learning at inference remains unclear. The authors adapt three classic artificial language learning experiments spanning morphology, morphosyntax, and syntax to evaluate implicit learning in gpt-4o and o3-mini. Results show alignment between models and human behaviour that is specific to linguistic domain: o3-mini aligns better in morphology, and both models align in syntax.
Find it
Qin et al. (2026)
ImplicitMemBench: Measuring Unconscious Behavioral Adaptation in Large Language Models
In
Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), 28232-28261
Abstract
Existing memory benchmarks for LLM agents evaluate explicit recall of facts, yet overlook implicit memory where experience becomes automated behavior without conscious retrieval. This gap is critical: effective assistants must automatically apply learned procedures or avoid failed actions without explicit reminders. We introduce ImplicitMemBench, the first systematic benchmark evaluating implicit memory through three cognitively grounded constructs drawn from standard cognitive-science accounts of non-declarative memory: Procedural Memory (one-shot skill acquisition after interference), Priming (theme-driven bias via paired experimental/control instances), and Classical Conditioning...
Find it
Squire (2004)
Memory systems of the brain: A brief history and current perspective
In
Neurobiology of Learning and Memory, 82(3), 171-177
In summary
Traces the history of the idea that memory comprises distinct systems, from nineteenth-century philosophical precursors through the mid-century experimental turn to the position, settled by about 1980, that a fundamental division separates memory available to conscious recollection from memory that is not. Describes the subsequent move beyond that dichotomy to a biologically grounded framework of multiple systems supported by the hippocampus and related structures, the amygdala, the neostriatum, and the cerebellum, and offers a current account of how these systems operate together to support behaviour.
Find it