On this page

Learning Theories — the family of frameworks that explain how learning happens, and the umbrella concept for theory-related ideas. In AI in education, learning theories shape both how AI systems are designed (the pedagogy they embody) and how the field interprets whether AI "works": the same tool can be a scaffold under constructivism assumptions, a reinforcement engine under behaviorism, or a cognitive-load hazard under Cognitive Load Theory.

Questions to Consider

  • Think about an AI tutor or adaptive system you have used or seen. What assumptions did it make about how people learn — did it reward right answers (behaviorism), build understanding (constructivism), or manage mental effort (cognitive load)? Did its makers ever state those assumptions?
  • A recurring gap runs through the field: discourse espouses constructivism while AI implementations default to drill-and-feedback mechanics. Where have you seen a tool claim to support deep learning but actually just reinforce surface responses?
  • The same AI tool can look like success under one theory and failure under another — inflated homework scores read as learning under behaviorism but as a failure to build durable understanding under constructivism. Which lens is fairer for judging whether students actually learned?
  • Because every AI tutor embeds a theory whether its designers say so or not, 'does it work?' may be the wrong question. What is the better question to ask about an educational AI, given the theories it could embody?
  • Generative AI is prompting educators to consider new theories — like learning as iterative co-construction between humans and AI, or AI as a cognitive partner across the lifespan. Does the rise of AI genuinely require new learning theories, or do existing ones still suffice?
  • Learning-gains measurement is itself 'theory-laden': an instrument built on one theory may not capture the gains another predicts. How might two researchers with different theoretical commitments look at the same data and reach opposite conclusions about whether an AI tool worked?

Introduction

Learning theories are the umbrella concept for the pedagogical assumptions behind AI in education. They sit at the heart of the field because every AI tutor, adaptive system, and feedback tool embeds assumptions about how people learn, whether the designers state them or not. Each theory has its own literature, and together they form the conceptual lens through which AI's design and effects are evaluated. The research field that tests those frameworks is distinct: the learning sciences study learning empirically through experiments, classroom trials and design-based research, and treat a theory as something to be confirmed or falsified, whereas the theory literature sets out the frameworks themselves. The singular phrase "learning science" names the theory strand.

The learning-theory landscape

Several families of learning theory recur in AI education research, each with its own concepts:

A 2026 systematic review of 163 GenAI studies maps the literature onto five theory traditions and finds constructivist and connectivist designs dominant, with behaviorist reinforcement designs confined largely to primary-level adaptive platforms (Saranya & Saravanan, 2026). Its reviewers inferred each study's theory from described practice because primary studies rarely named one, which is itself a finding about how the field reports theory.

Learning theories and learning gains

Learning theories are ultimately evaluated by their outcomes, and learning gains is where theory meets evidence. Each theory makes a different prediction about what counts as learning and how to measure it: behaviorism predicts observable performance gains on drill-and-feedback tasks; constructivism predicts deeper understanding that transfers to novel problems; sociocultural theory predicts gains in participation and mediated problem-solving; and cognitive-load theory predicts gains only when instruction respects working-memory limits. Measurement choices, then, decide which theory's prediction an evaluation can confirm. In the AI era, the sharp divergence between AI-assisted performance and unassisted learning gains (see Rismanchian et al. (2026), Strömberg et al. (2026)) can be read through this lens: a behaviorist reading sees inflated homework scores as success, while a constructivist reading, focused on durable understanding, sees the same evidence as a failure to learn.

The pooled evidence is positive, and it measures a narrower construct than the theories do. A PRISMA meta-analysis of 53 studies puts GenAI assistance at g = 0.40 on academic achievement, 0.72 on higher-order thinking, 0.76 on writing and 0.81 on motivation, with GenAI feedback the largest mechanism at g = 1.27 (Dong (2026)). Those are gains on the instruments the included studies chose, so they settle a behaviorist or cognitivist reading of "it works" and leave the constructivist prediction about durable, transferable understanding untested. The country subgroups also diverged, from China's g = 0.71 to Turkey's nonsignificant g = 0.01, so the pooled estimate is an average to test locally rather than a property of the tool.

Why learning theories matter for AI in education

Learning theories matter for three reasons:

  • They predict AI's effects. Whether an AI tool improves or harms learning depends on which mechanism it activates. A tutor that gives away answers harms under a constructivist lens (it bypasses construction), is neutral under behaviorism (it reinforces), and raises cognitive-load concerns (it offloads rather than builds). Cognitive offloading and over-reliance capture the risk side of this.
  • They expose the theory-practice gap. Empirical work repeatedly finds that AI implementations embody different theories than the discourse claims — most notably constructivist language paired with behaviorist drill-and-practice mechanics.(Deutscher et al., 2026) Evaluating AI therefore requires asking which theory a system actually embodies, not just whether it "works."
  • They are being actively rethought. Generative AI is prompting educators to revisit whether the classical theories suffice. Generativism proposes that learning in the AI age increasingly occurs through iterative co-construction between human learners and AI systems, extending rather than replacing the classical four (behaviorism, cognitivism, constructivism, connectivism).(Li et al., 2026)

New theoretical directions from recent AIED work

Recent theoretical work extends the classical strand in several directions, each re-centering the human–AI relationship rather than treating AI as a neutral tool:

  • Human–AI co-regulation. A developmental framework positions AI not as an external instrument but as a cognitive partner that co-regulates thinking, learning, and self-control across the lifespan.(Artificial intelligence as a cognitive partner: a developmental framework for human-AI co-regulation in learning) Drawing on executive function, metacognition, distributed cognition, and sociocultural development, it casts AI in four roles — scaffold, metacognitive support, external memory / cognitive offloading system, and decision partner — with the framework most relevant from middle childhood onward.

  • Ensemble Cognition. A philosophical framework reconceptualizes thinking as emerging from dynamic interactions between human and artificial agents rather than residing solely in individual minds.(Jho et al., 2026) It challenges the "consciousness paradigm" (the autonomy, consciousness, and stability assumptions) and articulates five features — distributed agency, dynamic centrality, cognitive orchestration, multi-representational integration, and context-sensitive switching — while distinguishing AI's functional agency from moral responsibility.

  • Self-Directed Growth / A2PL. An extension of self-directed learning integrates Generative AI with learning analytics to cultivate Self-Directed Growth, operationalized through the Aspire to Potentials for Learners (A2PL) model.(Mao, 2025) It reconfigures learner aspirations (humanistic), complex thinking (constructivist), and self-assessment (pragmatic) into a single competency, positioning GAI as a non-prescriptive collaborative scaffold rather than a content provider.

  • Deceptive overgeneralization. An, McLaren, and Stamper (2026) extend the ACT-R / Knowledge-Learning-Instruction tradition by theorizing when observed correctness masks incomplete conditional understanding: learners compile an overgeneralized production that omits an application constraint yet still performs correctly — a failure mode that adaptive mastery systems, and even traditional instruction, can miss unless they test when to withhold an action.

  • Executable KLI theory. Rachatasumrit, Koedinger & Carvalho (2025) ground the Knowledge-Learning-Instruction framework in an executable computational model (the Apprentice Learner framework with an ACT-R-style memory mechanism) that reproduces a cross-over interaction in human data: pure practice aids verbatim fact memory (by delaying forgetting) while example-integrated practice aids generalizable skill induction. Because KLI ties constant (fact) knowledge to memory processes and variable (skill) knowledge to induction, the result is a predicted content–treatment interaction rather than a contradiction between testing and worked-example recommendations — and the model's success only when a memory mechanism is present demonstrates that practice and examples play distinct, complementary roles.

  • Embodiment as an ontological challenge to AI's paradigm. Videla, Penny and Ross (2026) argue an ontological divide separates embodied, enactive cognition from the representational idiom AI embodies, so AI should be decentered as the epistemic center rather than treated as a neutral tool.

How the theories connect

Rather than treating learning theories as abstract philosophy, each is grounded in the AI-in-education research that uses it. The constructivism and behaviorism pages document how AI designs embody (or betray) each theory; Cognitive Load Theory, self-regulated learning, metacognition, and transfer of learning connect theory to specific AI mechanisms and outcomes. This mirrors how other umbrella domains such as feedback and assessment are treated, as a coherent system of interacting concepts rather than isolated pages.

Learning theories and "education about AI"

Learning theories also appear as content in AI literacy curricula: learners study behaviorism, cognitivism, constructivism, and connectivism to understand the pedagogical assumptions behind the tools they use.(Li et al., 2026) Teaching this strand gives students (and educators) the vocabulary to critique why an AI product is built the way it is — and whether its mechanics serve the learning goal at hand.

  • The mediational agent. Warschauer, Tate, and Ritchie (2026) argue generative AI breaks the sociocultural distinction between mediational means and social interaction, proposing the mediational agent — a system that both mediates action and generates contingent, non-accountable contributions, occupying a hybrid space between a tool and a social partner. This yields five human-first habits of participation (primacy of human cognition, purposeful engagement, supervisory agency, epistemic vigilance, reflective self-regulation).(Warschauer et al., 2026)

Theories proposed for the AI era

Alongside the classical families, a body of theory addresses learning with AI systems specifically, and these frameworks carry the design implications that the older theories leave open. Agentivism (Yan and Gašević 2026) is a mid-range example. It defines learning as durable growth in human capability rather than successful task completion, and names four mechanisms: delegated agency, epistemic monitoring and verification, reconstructive internalization, and transfer under reduced support. It also states six testable propositions, among them that AI support preserving learner responsibility for problem framing, criteria setting and justification produces stronger learning than support that delivers answers.

Elsayed (2026)'s Pedagogical Symbiosis makes a stronger ontological claim: the learner is a Post-Human entity whose cognition is hybrid rather than tool-assisted, organized by four principles: cognitive offloading and augmentation, epistemic co-construction, metacognitive symbiosis, and dynamic identity formation; operationalized as a Symbiotic Portfolio rubric and a "Cognitive Choreographer" teacher role; it is explicitly untested.

Implications for Practice

For instructors

  • Choose theory by what you need it to do: explain how learning happens, describe the learner's state, or explain why a practice spreads. Applying the wrong layer measures the right variable against the wrong prediction.
  • State the outcome measure's theory commitment explicitly, since inflated homework scores read as success under behaviorism and as failure under constructivism from identical data.
  • When a framework cannot predict a difference between configurations, it cannot be tested in your setting and will not help you decide.

For developers

  • Publish the theory your system embodies and the prediction that would falsify it; "grounded in constructivism" without a falsifiable difference is a claim with no content.

For researchers

  • Design for adjudication between theories rather than accumulation of frameworks. Maturity shows up as competing predictions resolved, not as vocabulary growth.

Common Misconceptions

"One of the learning theories is the correct one, so theory choice is a matter of taste." The theories make different claims about what counts as evidence, so the same data can support opposite conclusions. The clearest instance: inflated homework scores read as learning under behaviorism and as a failure to build durable understanding under constructivism. Theory choice therefore determines whether an intervention succeeded, which makes it a measurement decision rather than a stylistic preference.

"A tool that supports deep learning will look like deep learning under any framework." It will look like success under one and failure under another, as above. That is why the field's outcome measures need to be justified against a stated theory rather than borrowed as if neutral.

"More theories mean a better-understood field." Accumulation without adjudication is the opposite of progress. Theories proposed for the AI era multiply faster than they are tested, and the sign of maturity is competing predictions being resolved — which requires designs that can distinguish them, not catalogues that accommodate them.

"Education about AI is the same as AI in education, theoretically speaking." They draw on different strands: one asks how learners come to understand AI as a subject, the other how AI changes how learning happens. Conflating them produces literature reviews that mix two distinct evidence bases and report a coherent picture of neither.

Connected Concepts

Connected Articles

Connected FAQs

Embed this page

Copy the code below to embed this page in a learning management system or other website. The embedded version shows only the page content — it hides the site header, navigation, and footer.