Learning¶
Core Idea¶
Learning is the process by which an agent acquires or modifies an internal capability — knowledge, skill, model, or behavior — as a durable result of experience or information, such that the agent's future performance or predictions change. It is the learner-side counterpart to pedagogy: where pedagogy is the deliberate teaching aimed at causing it, learning is the agent's own update. The essential commitment is a durable, experience-driven self-update of an agent's internal state that carries forward to alter later behavior — distinguishing it from a one-off response that leaves the agent unchanged.
How would you explain it like I'm…
Getting Smarter from Experience
Learning from Experience
Learning
Broad Use¶
- Psychology / behaviour: conditioning, skill acquisition, practice effects, observational learning.
- Machine learning: updating model parameters from data so future predictions improve — the same experience-to-capability structure in a computational substrate.
- Biology: adaptive immunity (immune memory) is a biological learning system — exposure durably changes future response.
- Organizations: organizational/collective learning — routines and knowledge updated from accumulated experience.
- Animal behaviour: classical and operant conditioning across species — the same experience-driven update pattern in non-human agents with no language and no instruction.
Clarity¶
Learning names a specific, narrow thing inside the broader cluster of change: a durable, experience-driven update to an agent's internal capability that carries forward into altered future behavior or prediction. It is not the input (experience), not the storage (memory), not the loop that supplies error signals (feedback), not the teacher's effort to cause it (pedagogy), and not every kind of system-level change under pressure (adaptation). What learning adds is the conjunction of three commitments that no neighboring prime carries together: an agent with modifiable internal state, an experiential cause of the change, and durability such that the change persists and shows up later. Naming that conjunction prevents the common slippage where "the system changed" gets reported as learning when in fact the change was random drift, a one-off reaction, or a structural redesign with no role for experience.
Manages Complexity¶
Learning decomposes an opaque "the agent got better" into four named roles whose presence can be checked one at a time: (1) an agent with a modifiable internal state (memory, weights, schema, routine — whatever holds the change); (2) experiential input — observations, feedback, practice, instruction, or exposure that supplies the cause; (3) an update mechanism that modifies the internal state in light of that input together with a performance or error signal; and (4) durability — the change persists and manifests as altered future behavior or prediction. Once those four roles are named, an analyst can convert a vague "we learned this" into a structured problem: which state is modifiable, what experiential signal is reaching it, what update rule is operating, and does the change actually persist past the moment? This makes the same diagnostic apply equally to a child learning to ride a bike, a neural network training on examples, and an organization adjusting its routines after a postmortem.
Abstract Reasoning¶
Learning supports a sharp counterfactual: if the experiential input were withheld, removed, or scrambled, the change would not happen — and if the agent had no modifiable state, the input would pass through without leaving a trace. That move lets the analyst predict where learning will and will not occur in any new domain: it requires an asymmetric, directed coupling from experience to internal state to future behavior, and the absence of any of those three breaks the pattern. The reasoning also enables clean cross-cutting comparisons across substrates — supervised vs. unsupervised updates, single-trial vs. multi-trial regimes, sample-efficiency, transfer between contexts, forgetting and decay, the gap between performance gain and underlying capability change. These operations are substrate-neutral because they ride on the structural topology (agent → experience → update → durable capability) rather than on the substrate's specific machinery.
Knowledge Transfer¶
The four-role structure travels intact across substrates with no metaphorical stretching. A machine-learning researcher reading about adaptive immunity recognizes a parameter-update loop driven by exposure samples; an immunologist reading about supervised learning recognizes antigen-driven memory cell selection; an organizational-learning theorist recognizes the same structure in postmortem-driven routine revision; a psychologist recognizes it in operant conditioning. The biological case — adaptive immunity — is especially clean because there is no language, no instruction, no consciousness, and yet the four roles are all clearly present: modifiable internal state (the immune memory cell repertoire), experiential input (pathogen exposure), an update mechanism (clonal selection and affinity maturation), and durability (life-long altered response). That single case rules out the suspicion that "learning" is a specialty of cognitive psychology — it forces the prime to live at the structural level where the same pattern crosses cleanly into machine learning and organizational routines.
Example¶
Consider a chess player who repeatedly plays a particular opening, loses several games to a specific counter, studies those losses, and over the next month stops falling for the counter. Walk the four roles: the modifiable internal state is the player's repertoire of move-evaluations and pattern recognition; the experiential input is the sequence of losses plus the post-game review; the update mechanism is whatever cognitive process re-weights the player's response to the position the counter creates (combining the error signal — losing — with the corrective information); and the durability is the persisting altered behavior — the next time the position appears, the player plays differently. This is learning, not adaptation in general: the change is experience-driven (no losses, no review, no change), agent-internal (the chess world did not change), and capability-altering (future performance is different). The same pattern recurs in a neural network trained on labeled examples (parameters as state, examples as input, gradient descent as update, persistent loss reduction as durability), in adaptive immunity (memory cell repertoire, antigen exposure, clonal selection, life-long altered response), and in an engineering team revising its deployment runbook after an outage (the runbook as state, the outage as input, the postmortem as update, fewer future outages as durability).
Relationships to Other Abstractions¶
Current abstraction Learning Prime
Parents (2) — outgoing typed relations
-
Learning is a kind of Adaptation Prime
Learning is a specialization of adaptation in which an agent's internal capability is the structure modified in response to experience.
-
Learning includes, conditional Memory Consolidation Prime
Memory Consolidation is an internal stabilization phase of Learning when durable change depends on converting a fragile trace into stable memory.
Condition / exception The relation holds only where learning durability uses a distinct fragile-to-stable trace stage; learning without such a stage is excluded.
Children (62) — abstractions linked here by typed relations
-
Active learning Domain-specific is a kind of Learning
It remains its own entry because its identity is fixed by the learners and learning objective, preparatory knowledge, meaningful task beyond passive listening, individual or collaborative reasoning, produced response artifact or decision, feedback and reflection, instructor facilitation, participation and accessibility design and outcome assessment aligned to the objective.
-
Blended learning Domain-specific is a kind of Learning
It remains its own entry because its identity is fixed by the learners and objectives, physical and online settings, required presence, allocation and integration of activities, learner control over time place path or pace, teacher role, interaction and feedback, assessment alignment, access constraints and outcome evidence.
-
Cold hardening Domain-specific is a kind of Learning
It remains its own entry because its identity is fixed by the organism species tissue and developmental stage, inducing temperature range and duration, environmental covariates such as photoperiod and water, sensing and signaling, membrane lipid and osmolyte changes, protective proteins and gene expression, extracellular ice dehydration and cellular protection, measured freezing or chill tolerance, acclimation and deacclimation timeline and distinction from dormancy adaptation and vernalization.
-
Cooperative education Domain-specific is a kind of Learning
What makes it its own entry: A structured work-integrated learning model that alternates or integrates academic study with supervised paid or credited employment tied to curricular outcomes.
-
Deep belief network Domain-specific is a kind of Learning
It remains its own entry because its identity is fixed by the visible data and multiple hidden-variable layers, within-layer connection restrictions, directed lower generative links and undirected top pair, restricted Boltzmann machine components, greedy unsupervised layerwise pretraining, approximate inference and reconstruction, generative distribution, supervised fine-tuning option and distinction from DBM autoencoder and generic deep neural network.
- Dogme language teaching Domain-specific is a kind of Learning
What makes it its own entry: the domain-specific identity determined by instruction is conversation-driven, materials-light, and responsive to language emerging from the participants.
- Domain-general learning Domain-specific is a kind of Learning
The theory concerns learning mechanisms generalized across content.
- Dreyfus model of skill acquisition Domain-specific is a kind of Learning
What makes it its own entry: A staged model in which learners move from context-free rule following through situational judgment toward increasingly intuitive expert performance.
- Graph neural network Domain-specific is a kind of Learning
The architecture learns representations from relational examples.
- Inferential Theory of Learning Domain-specific is a kind of Learning
**Learning** is the strict parent because the terminal result is a durable update of an agent's internal knowledge used in future inference.
- Land-based education Domain-specific is a kind of Learning
What makes it its own entry: An educational approach centered on sustained relationships with local land, waters, community and place-based knowledge, especially within Indigenous pedagogies.
- Latent Learning Domain-specific is a kind of Learning
Earlier exposure yields retained information expressed in later behavior, a specific form of learning.
- Linguistic Bootstrapping Domain-specific is a kind of Learning
**`learning` — strict subsumption.** Every linguistic-bootstrap event is a durable, experience-driven update in a learner that changes later interpretation or production.
- Memorization Domain-specific is a kind of Learning
It remains its own entry because its identity is fixed by the learner and target information, sensory or symbolic modality, encoding task, rehearsal elaboration or mnemonic strategy, organization and chunking, retention interval, consolidation and interference conditions, retrieval cues, recall recognition or reproduction criterion, accuracy and forgetting or transfer.
- Model-Free Reinforcement Learning Domain-specific is a kind of Learning
**Learning** is the broader abstraction this entry instantiates: model-free RL is a strict learning specialization whose feedback is sequential reward and whose update bypasses an explicit planning model.
- Multiple instance learning Domain-specific is a kind of Learning
It remains its own entry because its identity is fixed by the instance feature space, bag construction and membership, bag labels and supervision split, assumption connecting instance and bag labels, instance encoder or classifier, aggregation pooling or attention, bag-level objective, prediction target at bag or instance level and evaluation leakage and interpretability limits.
- Multisensory learning Domain-specific is a kind of Learning
**Representational Modality** (`prime:representational_modality`).
- Natural approach Domain-specific is a kind of Learning
What makes it its own entry: Any relaxed communicative class is not automatically the named method; its identity depends on the Krashen-Terrell acquisition assumptions and instructional priorities.
- Nurse education Domain-specific is a kind of Learning
Nurse Education instantiates Learning because organized experience and feedback produce durable changes in nursing knowledge, judgment, and professional capability.
- One person, one language Domain-specific is a kind of Learning
It remains its own entry because its identity is fixed by the child and developmental stage, caregivers and designated languages, interaction settings and consistency, exposure amount and quality, community and schooling language, child response and code-switching, family goals and longitudinal proficiency and wellbeing evidence.
- Online Machine Learning Domain-specific is a kind of Learning
Sequential revision of a persistent predictive or decision state from arriving observations or feedback is specialized experience-driven learning.
- Peer Learning Domain-specific is a kind of Learning
Peer Learning most directly **instantiates `prime:learning`**: experience supplied through peer interaction updates a learner's knowledge or capability and should carry forward into later performance.
- Perceptual learning Domain-specific is a kind of Learning
The candidate literally instantiates prime:learning; its cognitive_psychology restrictions provide the domain-specific residual.
- Population-based incremental learning Domain-specific is a kind of Learning
It remains its own entry because its identity is fixed by the search space and encoding, probability-vector initialization, population sample size, objective and ranking, positive and negative update rules, learning rates, mutation probability and shift, stopping condition and repeated-run evaluation.
- Proactive learning Domain-specific is a kind of Learning
**Learning** is the narrowest accepted prime because both prediction and acquisition policy are updated from labeled experience.
- Processability theory Domain-specific is a kind of Learning
What makes it its own entry: A second-language acquisition theory predicting developmental sequences from the hierarchy of grammatical information-processing procedures learners can execute in real time.
- Reinforcement learning Domain-specific is a kind of Learning
Reinforcement Learning instantiates Learning because experience durably updates an agent's internal policy, value, or model so that later behavior changes; its distinguishing signal is sequential reward.
- Retrieval Practice Domain-specific is a kind of Learning
Retrieval practice is learning specialized to a durable human-memory update produced by effortful cue-to-target recall rather than passive re-exposure.
- Second-language acquisition Domain-specific is a kind of Learning
Second-language acquisition is a domain-specific kind of Learning: Second-language acquisition is a recurring identity in social sciences, humanities, and arts defined by this frozen evidence: Process in which a second or additional language is acquired after the first language has been learned.
- Self-persuasion Domain-specific is a kind of Learning
The person learns or revises attitude through self-generated reasoning.
- Self-supervised learning Domain-specific is a kind of Learning
Data-derived targets drive durable model-parameter updates that alter later predictions.
- Silent Way Domain-specific is a kind of Learning
It remains its own entry because its identity is fixed by the learners and target language, teacher-silence convention, visual or manipulative materials, problem sequence, learner production, peer interaction, feedback and self-correction, assessment and adaptation.
- Simulations and games in economics education Domain-specific is a kind of Learning
It remains its own entry because its identity is fixed by the economic concept and model assumptions, learner population and roles, rules actions information and timing, resources payoffs chance and strategy, interaction rounds, recorded individual and aggregate outcomes, theoretical benchmark, facilitation and debrief, learning objectives and assessment and realism validity and accessibility limits.
- Situated Learning Domain-specific is a kind of Learning
Situated Learning is a strict kind of Learning in which the durable experience-driven update is formed through legitimate participation in an authentic social practice.
- Statistical learning in language acquisition Domain-specific is a kind of Learning
The phenomenon learns linguistic units from experience-dependent statistics.
- U-shaped development Domain-specific is a kind of Learning
It remains its own entry because its identity is fixed by the learner population and task, performance measure and baseline, observation times, initial strategy, decline and recovery criteria, item equivalence, longitudinal versus cross-sectional design, proposed representational reorganization, competing explanations, uncertainty and replication.
- Universal law of generalization Domain-specific is a kind of Learning
The law characterizes how learning transfers from experienced to novel cases.
- Woodshedding Domain-specific is a kind of Learning
What makes it its own entry: the domain-specific identity determined by the practitioner isolates a performance target, undertakes sustained corrective practice away from the performance setting, and demonstrates later transfer into integrated execution.
- Absorptive Capacity Prime is a kind of Learning
Absorptive capacity is a specific kind of learning where the agent acquires capability by recognizing and assimilating external knowledge.
- Collective Systemic Learning Prime is a kind of Learning
Collective systemic learning is a specialization of learning in which the agent doing the updating is a multi-component system rather than an individual.
- Conditioning (Behavioral) Prime is a kind of Learning
Behavioral conditioning is a specialization of learning; it is the family of contingency-detection mechanisms that durably update behavior through pairing.
- Constructivist Learning Prime is a kind of Learning
Constructivist learning is a specialization of learning that frames the durable update as active meaning-making rather than passive reception.
- Exposure-Distribution Learning Prime is a kind of Learning
Exposure-distribution learning is Learning specialized to durable frequency-sensitive updates from an encountered observation distribution.
- Hebbian Learning Prime is a kind of Learning
Hebbian Learning is a specialization of Learning, retaining the parent's defining structure while adding the child's specific commitments.
- Inoculation Theory Prime is a kind of Learning
Inoculation is learning specialized to durable defensive updating from attenuated pre-exposure paired with successful response.
- Inquiry Prime is a kind of Learning
The accepted reference-grade review places Inquiry under Learning because the child instantiates or depends on the parent's broader structure while retaining its own constitutive identity.
- Learned Helplessness Prime is a kind of Learning
Learned helplessness is a specialization of learning in which the acquired internal capability is a belief that action and outcome are independent.
- Observational Learning (Social Learning) Prime is a kind of Learning
Observational learning is a specialization of learning in which acquisition occurs through watching others rather than through direct trial-and-error.
- Perceptual Expertise Prime is a kind of Learning
Perceptual Expertise is Learning specialized to a durable, exposure-driven change in representational resolution and later discrimination.
- Action Research Domain-specific includes Learning
Action research contains learning because each changed situation supplies experience that durably updates understanding and thereby alters later action and prediction.
- Service-Learning Domain-specific presupposes Learning
Service-Learning structurally presupposes Learning: Below is a list of examples (not all inclusive) of different service-learning programs offered throughout the United States.
- Adaptive Redirection Prime presupposes Learning
A redirection qualifies only when experience on the rejected course produces a durable update that is retained and used to select or execute the new course.
- Inquiry-Change Learning Loop Prime includes Learning
The loop contains learning because evidence must durably update the retained model or capability that selects later action.
- Learning Curve Effects Prime presupposes Learning
Learning curve effects presuppose learning because the predictable cost decline with cumulative experience is a downstream signature of durable self-update.
- Learning-Substrate Contamination Prime presupposes Learning
Learning-substrate contamination contains a normal learning process whose durable update converts adversarially selected evidence into later behavior.
- Pedagogy Prime presupposes Learning
Pedagogy presupposes learning because the deliberate teaching act is only intelligible relative to the learner-side acquisition it aims to cause.
- Shortcut Learning Prime presupposes Learning
Shortcut Learning presupposes Learning, whose structure must already obtain for the child mechanism to be meaningful or operational.
- Spaced Repetition Prime presupposes Learning
Spaced repetition presupposes learning because it is a procedure for strengthening durable memory updates against forgetting over time.
- Transfer of Learning Prime presupposes Learning
Transfer of learning presupposes learning because there must be acquired source-domain capability before it can be applied to a new context.
- Variation Strategies Prime presupposes Learning
Variation strategies presuppose learning because deliberately injected variation produces value only when the system can detect, retain, and act on the surfaced alternatives.
- Innovation Accounting Domain-specific is a decomposition of Learning
The ledger counts durable belief updates that carry forward to change later tests and strategy, rather than activity or transient metric movement.
- Pivot Domain-specific is a decomposition of Learning
A genuine Pivot preserves calibrated knowledge from the disconfirmed bet and makes that durable update load-bearing in the redirected strategy.
Hierarchy paths (2) — routes to 2 parentless roots
- Learning → is a kind of Adaptation
- Learning → includes, conditional Memory Consolidation
Not to Be Confused With¶
- Not Pedagogy (the pair): pedagogy is the teacher-side practice aimed at producing learning; learning is the learner's own acquisition. Pedagogy presupposes learning.
- Not Adaptation: adaptation is the broader pattern of a system changing structure/behavior under sustained pressure — including evolutionary and physiological change with no memory or agent. Learning is the experience/information-driven acquisition of capability by an agent that retains the change. (Proposed: learning -> adaptation, the memoried/informational kind.)
- Not
memory: memory is retention/storage; learning is the acquisition/update process that writes to and draws on memory. - Not Feedback: feedback is the loop that routes output back to input; learning may use feedback signals but is the resulting durable capability change, not the loop itself.
Notes¶
The learner-side sibling of Pedagogy. Learning parents the acquisition-side primes; Pedagogy parents the teaching methods.
One substrate-neutral pattern (experience-driven durable self-update), not a composite -> a prime, not a connector.