Learning Is Biological

Learning is biological in the plainest possible sense: experience leaves a living system changed. The change can be fleeting or durable, available only in one setting or transferable across many, consciously described or expressed through faster and more accurate action. It involves brains, but not brains isolated from bodies, relationships, tools, histories, and environments. The biology matters because learning is not merely information delivered. It is adaptation.

That statement is both more modest and more radical than the popular language of “brain-based learning.” It is modest because neuroscience rarely tells a school leader exactly which schedule, platform, or lesson to choose. Findings about synapses, memory systems, reward prediction, or brain networks do not travel directly into a procurement decision. It is radical because a biological process cannot be commanded into existence by covering content. Exposure is not change. Completion is not retention. Performance under familiar prompts is not necessarily transfer.

The National Academies’ synthesis treats learning as an interaction among the learner, the material, the activity, and the social and cultural context. ¹ This systems view guards against two opposite mistakes. One is to treat the mind as software that receives neutral content. The other is to use biology as destiny: to assume that a genetic association, a scan, or an early performance fixes what a person can become. Biology is not the opposite of environment. Development is the history of their interaction.

Neural plasticity gives the chapter its starting point, not a marketing promise. Experience-dependent change and memory consolidation are part of the National Academies’ account of learning, including the roles of prior knowledge and sleep. ¹ A finding that practice changes a measured neural response does not prove that any particular practice is educationally effective. The educational question is whether the change supports the capability that matters.

Dopamine is a useful example of how a legitimate mechanism becomes an educational myth. Dopamine research includes accounts of reward-prediction error and motor learning. ^2,3 It is not a volume knob that a gamified app simply turns up. Reward cues can increase engagement while directing attention toward points rather than understanding. A product may be stimulating without producing durable learning. Leaders should be suspicious when a complex neuromodulatory system is offered as a one-word explanation for a commercial feature.

The same caution applies to regional brain stories. Neural accounts of learning involve interacting systems rather than one educational switch. ¹ But a classroom activity does not become validated because it “activates” a named region. The relevant chain is longer: does the activity change behavior or understanding, does the change persist, does it transfer, and can the result be explained without a simpler account such as additional time, feedback, or practice?

For leaders, the biological view establishes a design test. A learning environment should create repeated opportunities for meaningful action, timely information about the result, recovery and consolidation, and variation sufficient to reveal whether knowledge can travel. It should also protect the physical and emotional conditions that make sustained attention possible. These principles are less glamorous than a brain graphic. They are also more demanding, because they require institutions to organize time around learning rather than organize learning around a timetable.

From mechanism to educational design

The distance between a biological finding and an educational decision deserves explicit attention. A mechanism may be established under controlled conditions, an instructional technique may have an average effect across studies, and a school may still fail to implement it effectively. These are not contradictions. They are different levels of claim.

At the mechanism level, researchers ask what processes participate in attention, memory, prediction, movement, emotion, or adaptation. At the learning-design level, researchers compare tasks and conditions: retrieval versus restudy, spaced versus massed practice, worked examples versus unsupported problem solving. At the implementation level, leaders confront schedules, training, relationships, language, materials, incentives, and competing demands. Evidence can weaken at each transition because the question changes.

This distinction helps leaders resist two common errors. The first is neuro-essentialism: assuming that naming a brain process makes an intervention more scientific. The second is implementation fatalism: concluding that because context matters, research cannot guide action. A better approach carries evidence forward with its boundary conditions. Spacing and retrieval, for example, have substantial support as learning techniques, but their useful form depends on the material, the interval, the learner’s prior knowledge, the quality of feedback, and the desired retention period. ^4,5,6

The educational unit is not a neuron. It is a person acting in an environment over time. That person may be tired, threatened, curious, grieving, fluent in several languages, unfamiliar with the institution’s language, confident in one domain, or carrying a history of exclusion. None of those realities makes biology irrelevant. They are how biology is lived.

A ladder of responsible inference

Before attaching a neuroscience claim to a program, leaders can ask four questions. Was the finding observed in humans performing a task relevant to learning? Does it predict a meaningful behavioral or educational outcome rather than only a measured signal? Has the proposed intervention been compared with credible alternatives? Has it worked under conditions sufficiently similar to the intended setting?

The further an answer falls down that ladder, the more provisional the decision should be. A product should not receive a stronger evidentiary status because its brochure contains a brain image. A familiar classroom practice should not escape scrutiny merely because it predates neuro-marketing. The same claim discipline applies to both novelty and tradition.

Recovery, stress, and the design of time

Learning requires time not only for exposure and practice but also for consolidation, feedback, and return. Schedules built entirely around coverage can create the appearance of productivity while reducing opportunities to retrieve and use knowledge later. Leaders control some of these conditions through calendar design, course pacing, homework policy, transition load, and the protection of sleep and recovery.

The implication is not a universal biological timetable. Learners and domains vary, and the evidence rarely justifies a single optimal schedule. The implication is that time is part of the intervention. If a school changes curriculum but leaves no space for deliberate practice, feedback, revisiting, and transfer, it has changed content without redesigning learning.

The biological perspective is ultimately a check on institutional impatience. Durable adaptation takes repeated, meaningful encounters. It cannot be inferred from coverage, excitement, or short-term fluency alone. Systems should be designed to observe what remains after time has passed and what becomes available when the prompt changes.

Prediction, reward, and the danger of the short circuit

Biological research can deepen an explanation without completing an educational argument. Dopamine research is a clear example. Reward-prediction error models describe how outcomes that are better or worse than expected can update future behavior. ² That is a serious account of a learning mechanism. It does not follow that points, streaks, animations, or praise improve conceptual understanding, nor that a designer can infer or optimize a learner’s dopamine level from screen behavior.

Motor-learning research makes the boundary even clearer. Nikooyan and Ahmed studied reward and visual feedback during an abrupt visuomotor-rotation task; adding reward to visual feedback accelerated adaptation under the tested conditions. Zhao and colleagues’ review distinguishes learning stages and reports variation across experimental conditions. ^7,8 A comparative review examines proposed dopamine roles across species, while Phillips and colleagues recorded dopamine dynamics as mice learned a skilled reaching task. ^3,9 These findings matter because they complicate the cartoon in which dopamine simply means pleasure. They remain several inferential steps away from mathematics, reading, history, or a school’s incentive system.

The defensible educational conclusion is restrained. Feedback carries information; anticipation and consequence affect behavior; and the same reward can organize attention differently at different points in learning. A reward may help a learner begin, persist, or notice progress. It may also narrow the goal to obtaining the reward. Whether learning improved must be tested through the relevant capability after the reward, prompt, or support is removed.

The cerebellum is not a purchasing argument

Hull’s review describes cerebellar prediction signals beyond a simple account of supervised motor correction. ¹⁰ Huvermann and colleagues examined human feedback learning through complementary cerebellar-stroke and single-pulse stimulation experiments. The measured prediction-error signal changed, while learning of action–outcome associations remained largely intact. ¹¹ This is mechanistic evidence, not a demonstration that stimulating a region improves school learning. It makes the cerebellum scientifically interesting without making “cerebellar activation” an educational outcome.

Almost any coordinated activity recruits many neural systems. Naming one of them does not show that the activity produces durable academic learning, transfers beyond the practiced task, or works better than a credible alternative. Claims that a motion game, typing routine, instrument, or exercise improves “focus” because it stimulates a brain region collapse a long chain of inference into a slogan. The appropriate outcomes are behavioral: what can the learner now do, under what conditions, for how long, and with what transfer?

Learning theories operate at different levels

Behaviorist, cognitive, constructivist, and network accounts ask different questions. Reinforcement accounts help explain how consequences shape behavior. Cognitive accounts model attention, memory, and knowledge organization. Constructivist traditions emphasize the learner’s active interpretation of experience. Social and network accounts examine how knowledge is distributed among people, practices, and tools. None is made true by adding technology, and none is refuted because a lesson uses direct instruction.

Leaders should choose methods by the learning demand. A novice may need explicit explanation and worked examples before open inquiry is productive. ¹² A learner developing fluency may need repeated, corrective practice. A learner preparing for transfer must encounter changed examples, contexts, and representations. A community investigating a contested problem must learn how evidence, expertise, and power circulate. The scientific task is not to select one grand theory. It is to connect a bounded theory to an observable learning claim.

Genetic association is not an educational prescription

Neurotransmission reflects many interacting biological processes, and genetic variation contributes to differences across populations. Educational outcomes, however, are highly polygenic and inseparable from family, social, institutional, and environmental conditions. A large association study can identify small statistical relationships without yielding a reliable diagnosis, ceiling, or classroom intervention for an individual learner. ¹³

The environment is not merely an external variable added after biology. Research on genetic nurture shows that parental genotypes can be associated with children’s outcomes through environments parents help create, even when the relevant alleles are not inherited. ¹⁴ That finding undermines simple nature-versus-nurture stories. Education leaders should reject single-variant explanations of motivation, empathy, attention, or potential and should never use genotype to track, admit, exclude, or lower expectations.

Biology therefore supports neither fatalism nor optimization theater. It establishes that learners are living, developing systems whose histories matter. It also establishes why a measurement taken at one moment cannot become a verdict on a person’s future.

Plasticity is specific, costly, and historical

Plasticity is sometimes described as though the brain were indefinitely malleable clay and education simply chose the shape. The metaphor hides three facts. Change is constrained by what already exists, change in one system does not imply general improvement, and adaptation can be beneficial in one environment while costly in another. A learner becomes fluent through a history of particular actions, feedback, language, tools, relationships, and opportunities.

Transfer depends on more than improvement in the original practice task. ¹⁵ Repeated performance tunes perception and action to recurring structures. If the assessment reproduces those structures, the change is visible. When the surface, timing, context, or required decision changes, the learner must recognize what still applies. The biological capacity to adapt does not guarantee that the adaptation will travel.

This is also why “brain training” requires an educational outcome beyond progress in the training task. Improvement inside an application may show that the learner learned the application. That can be worthwhile, but it is not evidence of stronger reasoning, attention, memory, or academic capability elsewhere unless those outcomes were measured credibly. The closer the outcome resembles the practice, the less the study tells us about far transfer.

Prior knowledge changes the cognitive demands of a lesson. ¹² An expert sees organized relations where a novice sees separate details. A fluent reader can devote more capacity to argument and implication because decoding demands less deliberate control. A novice may need explicit guidance that an expert experiences as redundant. This difference is not a stable ranking of people. It is a description of the knowledge available for this task at this time.

Error must remain informative

Learning systems adapt through differences between prediction and outcome, but educational error is not one thing. An incorrect answer may reveal a misconception, missing prerequisite, misread instruction, inaccessible format, language gap, impulsive guess, transcription failure, or sensible interpretation of an ambiguous question. A score that records only wrongness discards the evidence needed for teaching.

Feedback should help the learner locate the difference. Sometimes that means a direct correction; sometimes a worked example, contrast, hint, question, or invitation to explain. The right form depends on what caused the error and what the learner must eventually do independently. Feedback timing has to be considered with the task, learner, and intended outcome; a review of formative feedback does not support one universally superior timing rule. ¹⁶

Prediction-error research in dopamine and cerebellar systems gives a mechanism-level reason to take discrepancy seriously. ^2,10,11 It does not tell a teacher exactly which response to give. Educational design still requires evidence about the task, learner, goal, and consequences. A mechanistic explanation can make a hypothesis plausible; comparative learning evidence must determine whether the design works.

My concern is that attaching permanent consequences to every attempt can make uncertainty costly to reveal. That is a design risk to examine, not a claim that all learners react identically. I recommend protected practice in which an error can change the next action without immediately changing a person’s status.

The body is not an accessory to the learner

Learning occurs in a body, and the National Academies’ synthesis identifies sleep as relevant to memory and cognitive functioning. ¹ I recommend examining recovery, sensory access, physical safety, and the demands of the schedule before treating a difficult performance as a stable learner limit. An engaging interface is not evidence that those conditions no longer matter.

Embodied action can be central when the capability itself is physical: playing an instrument, forming letters, pronouncing sounds, operating equipment, performing a clinical procedure, or coordinating a team. In those cases, the body is part of what must learn. For abstract concepts, leaders should ask whether a proposed gesture, manipulation, drawing, or movement clarifies the intended relation and whether that benefit was actually evaluated. “Uses the body” is not an automatic mark of rigor, just as “uses the brain” says almost nothing.

Accessibility makes this principle more exact. The educational goal may require a particular sensory or motor performance, or those demands may be incidental barriers. A learner who cannot use one channel is not outside biology; the learning environment must find another path through a different configuration of perception, action, tool, and support. Multiple routes do not deny biological reality. They take variation in living systems seriously.

Translate biology into institutional questions

A leader does not need to become a neuroscientist to evaluate biological claims. Ask what was measured: neural activity, physiology, behavior, immediate task performance, delayed retention, or transfer. Ask who participated and whether the proposed learners resemble them. Ask whether the intervention itself was tested or only a mechanism associated with it. Ask what credible alternative received less attention because the biological story sounded novel.

Then examine the institution’s own design. Does the calendar allow return after forgetting? Does feedback arrive while it remains actionable? Can learners practice without permanent penalty? Are sleep and recovery undermined by workload? Are supports matched to barriers rather than predictions of potential? Does the assessment change context enough to reveal flexible use?

These questions make biology operational without turning it into destiny. They direct attention from colorful explanations toward the conditions under which capability actually develops. The point of a biological account is not to make education sound scientific. It is to remind institutions that learning is change in a person and that such change has conditions, limits, histories, and consequences.

The boundary between explanation and prescription

Biology explains why learning is possible and why conditions matter. It does not settle what is worth learning or how opportunity should be distributed. Those are ethical and political decisions. Nor does it make a measure objective merely because the measure comes from a body. Eye movement, heart rate, neural activity, and response time all require interpretation. The more intimate the signal, the stronger the duty to justify collecting it.

The leadership implication is simple: demand behavioral and educational evidence for educational claims. A neuroscience story may enrich the explanation, but it should not substitute for demonstrated learning, comparative evaluation, or attention to implementation. When a vendor begins with a brain mechanism and ends with an institutional purchase, ask to see every step in between.

A worked decision: from a brain story to a learning claim

Consider a hypothetical proposal for a school science course. The proposed system asks students to answer short questions, rewards correct responses with points, schedules another attempt after an error, and reports a readiness score. Its presentation invokes dopamine, prediction error, and neural plasticity. The school wants students to understand energy transfer well enough to explain an unfamiliar physical situation.

Nothing in this example has been evaluated. It is a decision exercise showing how I would separate the questions before a school adopts the proposal. The point is not to reject the system because it uses rewards or to approve it because its scientific vocabulary is legitimate. It is to determine what, exactly, would justify the educational claim.

Start with the capability, not the mechanism

The first task is to describe the desired performance without mentioning the product. A student should be able to identify relevant quantities, explain the direction of energy transfer, distinguish a closed from an open system where that distinction matters, and justify an answer in a changed example. Subject experts would need to refine these criteria and ensure that they reflect the curriculum’s actual goals.

That specification exposes what a correct multiple-choice response cannot settle. The student might have recognized a familiar phrase, eliminated implausible alternatives, or followed a cue. Those are possibilities to examine, not accusations. A useful assessment would include an explanation and a new example because those are parts of the capability the school says it wants. If rapid response is not essential, the system should not quietly make speed part of readiness.

Only then should the team inspect the mechanism story. Dopamine research can make a hypothesis about feedback or reward plausible. It does not identify the best reward schedule for this course. The distinction between a mechanistic explanation and a tested educational intervention remains intact even if every neuroscience sentence in the presentation is accurate.

Separate the components that were bundled together

The proposed system changes several things at once: the number of practice opportunities, the timing of return, the feedback supplied, the reward display, and the teacher’s information. If the whole arrangement eventually outperformed an alternative, that would not by itself reveal which component produced the difference.

The school need not isolate every component before any low-risk use. It does need to avoid attributing an observed package effect to the most marketable feature. A comparison with existing practice would answer a question about the package. A comparison in which both groups receive the same questions, timing, and feedback but different reward displays would address a narrower question about that display. These are different evaluations with different costs and purposes.

The choice should follow the decision. If the decision is whether to replace the present course practice, the whole-package comparison may be most useful. If the school already uses equivalent retrieval and feedback but is being asked to purchase a costly reward layer, the incremental question matters. No neurological explanation should conceal which comparison is missing.

Define success after the support changes

The next question is when and under what conditions students will demonstrate the capability. A same-session quiz with hints available is evidence about performance in that supported setting. A later explanation without hints addresses a different claim. A new physical example asks whether the learner can recognize the relevant relation outside the practiced form.

For this hypothetical evaluation, I would specify those outcomes before examining results. I would also record which tools are permitted. An accessible representation or assistive input may remain appropriate even when an instructional hint is removed. Independence does not mean withdrawing every support; it means being precise about which assistance would perform the capability that the assessment is intended to reveal.

The school should not describe a later null result as failure of the students to retain the system’s benefit. It may show that the earlier benefit was narrower than expected. Equally, a useful access improvement should not be dismissed because it does not raise an unrelated score. A truthful evaluation can report several outcomes without forcing them into a single success label.

Examine who can use the proposal

Now consider the students whose interaction with the system differs from its assumptions. One uses a screen reader. Another needs additional time to produce text. Another understands the physical relation but is still learning the language of instruction. These are hypothetical differences to plan for, not invented accounts of how named students performed.

The team should ask whether the interface exposes equivalent information, whether timing is relevant to the goal, and whether the response mode preserves the intended reasoning demand. It should examine who was included in any supporting study and who was excluded. A claim about the average participant cannot establish that a route is accessible to everyone the school intends to serve.

The readiness score requires particular care. Before using it to restrict access to later work, the school would need evidence for that decision, not just evidence that the score correlates with performance somewhere. What are the consequences of a mistaken low score? Can a learner provide contrary evidence? Is the recommendation periodically reconsidered? A practice tool should not acquire authority over opportunity simply because it produces a number.

Decide what can responsibly happen next

A proportionate next step might be a limited instructional pilot, provided existing rights and support are preserved and no unsupported high-stakes classification follows. The school would identify a responsible adult, a way to report difficulty, the data necessary for evaluation, and conditions for changing or ending the pilot. It would not begin by collecting physiology merely because the mechanism story mentions the brain.

The report should state what was implemented, for whom, against which alternative, and with what outcomes. It should distinguish evidence from unresolved explanation. A useful result could justify continuing the practice while leaving its proposed dopamine mechanism untested. An unhelpful result could justify revising the practice without refuting the entire science of reward learning.

That is the discipline a biological view should bring to education. It connects mechanisms to hypotheses, hypotheses to tasks, and tasks to outcomes while keeping the learner’s access and agency visible. The scientific vocabulary becomes useful when it improves the questions. It becomes a gatekeeping device when it is used to end the questions before the educational evidence has arrived.

Implications for leaders

• Define the durable change expected from an experience before choosing the experience.
• Separate evidence of engagement from evidence of learning.
• Require products making neurological claims to show relevant educational outcomes and limitations.
• Protect sleep, movement, safety, recovery, and sustained attention as learning infrastructure.
• Refuse biological measures that are more invasive than the decision requires.

Questions to carry forward

1. Which routines in your institution measure exposure or compliance while being described as learning?
2. Where does a biological explanation improve design, and where is it being used as decoration?
3. What evidence would show that a change persists and transfers beyond the original setting?

Notes

1. 1National Academies of Sciences, Engineering, and Medicine. 2018. “How People Learn II: Learners, Contexts, and Cultures”. The National Academies Press.↩
2. 2Glimcher, Paul W.. 2011. “Understanding Dopamine and Reinforcement Learning: The Dopamine Reward Prediction Error Hypothesis”. Proceedings of the National Academy of Sciences, vol. 108, 15647–15654.↩
3. 3Wood, A. N.. 2021. “New roles for dopamine in motor skill acquisition: lessons from primates, rodents, and songbirds”. Journal of Neurophysiology, vol. 125, no. 6, 2361-2374.↩
4. 4Dunlosky, John, Rawson, Katherine A., Marsh, Elizabeth J., et al.. 2013. “Improving Students’ Learning With Effective Learning Techniques: Promising Directions From Cognitive and Educational Psychology”. Psychological Science in the Public Interest, vol. 14, no. 1, 4–58.↩
5. 5Cepeda, Nicholas J., Pashler, Harold, Vul, Edward, et al.. 2006. “Distributed Practice in Verbal Recall Tasks: A Review and Quantitative Synthesis”. Psychological Bulletin, vol. 132, no. 3, 354–380.↩
6. 6Roediger, Henry L. III and Karpicke, Jeffrey D.. 2006. “Test-Enhanced Learning: Taking Memory Tests Improves Long-Term Retention”. Psychological Science, vol. 17, no. 3, 249–255.↩
7. 7Nikooyan, Ali A. and Ahmed, Alaa A.. 2015. “Reward Feedback Accelerates Motor Learning”. Journal of Neurophysiology, vol. 113, no. 2, 633–646.↩
8. 8Zhao, Jingwang, Zhang, Guanghu, Xu, Dongsheng. 2024. “The effect of reward on motor learning: different stage, different effect”. Frontiers in Human Neuroscience, vol. 18, 1381935.↩
9. 9Phillips, Chris D., Hodge, Alexander T., Myers, Courtney C., et al.. 2024. “Striatal Dopamine Contributions to Skilled Motor Learning”. The Journal of Neuroscience, vol. 44, no. 26, e0240242024.↩
10. 10Hull, Courtney. 2020. “Prediction Signals in the Cerebellum: Beyond Supervised Motor Learning”. eLife, vol. 9, e54073.↩
11. 11Huvermann, Dana M., Berlijn, Adam M., Thieme, Andreas, et al.. 2025. “The Cerebellum Contributes to Prediction Error Coding in Reinforcement Learning in Humans”. The Journal of Neuroscience, vol. 45, no. 19, e1972242025.↩
12. 12Sweller, John. 2024. “Cognitive Load Theory and Individual Differences”. Learning and Individual Differences, vol. 110, 102423.↩
13. 13Lee, James J, Wedow, Robbee, Okbay, Aysu, et al.. 2018. “Gene discovery and polygenic prediction from a genome-wide association study of educational attainment in 1.1 million individuals”. Nature genetics, vol. 50, no. 8, 1112-1121.↩
14. 14Kong, Augustine, Thorleifsson, Gudmar, Frigge, Michael L, et al.. 2018. “The nature of nurture: Effects of parental genotypes”. Science (New York, N.Y.), vol. 359, no. 6374, 424-428.↩
15. 15Barnett, Susan M. and Ceci, Stephen J.. 2002. “When and Where Do We Apply What We Learn? A Taxonomy for Far Transfer”. Psychological Bulletin, vol. 128, no. 4, 612–637.↩
16. 16Shute, Valerie J.. 2008. “Focus on Formative Feedback”. Review of Educational Research, vol. 78, no. 1, 153–189.↩