Chapter 2 — What Is Agency? Why Do We Want It?

Agency = capabilities + decision-making + origination + answerability.

The thief is skilled, decisive, and inventive, and answerable to no one. The influencer has plenty of voice, and stands behind it in public; the products are someone else's. The middle manager is competent, empowered, and accountable, executing a vision that was never theirs. And the last: that is a teacher under a pacing guide, inventive, expert, answerable to everyone, while the big decisions are made somewhere else.

Agency, at least as I am using the concept in the context of the classroom, is origination of and answerability for a move on an object of knowing, within a context and its constraints. Origination, decision-making, and capability describe something about the actor's capacity to act; answerability describes what happens when that action enters a world containing other people, knowledge, consequences, and responses.

In the classroom, origination defines whether an answer came from a student's own thinking, or was copied from lecture notes, from friends, or from AI. Can they originate actions in the shared conditions of the classroom? Can they originate their path of learning itself?

Decision-making involves voice, but it also requires the personal capacity to make a choice, and having the amount of informed knowledge required to make that choice.

Capabilities, as agency, means becoming more skilled, more knowledgeable, and increasingly sought after for expertise.

Answerability is the teacher answering the student, the student answering the teacher and the class, and each taking responsibility for their position. By "responsible" I mean that the person's identity and sense of self are present in their words, rather than the words simply telling the teacher what she wants to hear.

So my definition of agency as I am using it applies only in the context of learning: agency is the student's origination of and answerability for a move on the object of knowing, within a specified context and its constraints, including the classroom's people, relations, and the collaborative formation of knowledge. In a Bakhtinian sense this also includes texts and historical others, since we argue in our heads with them as well, even if we don't say it out loud.

The table below organizes the ways I have watched learning show up across classrooms and homeschooling.

As the table suggests, agency is not one thing. There is no hierarchy here; it is more about how the person shows up, or can show up, in a given context. In this chapter I argue for swapping the term motivation for a clearer language of agency: recognize students as knowers, and learn the patterns of agency that make these forms legible to the teacher.

My daughter, Heidi, once told me she disliked her high school art class. This surprised me because, as a homeschooler, she could lose herself in the color and rhythm of drawing for hours. What she didn't love was the way art class was being taught. "She's too directive. I wish she'd show us a skill, maybe a few examples, and then just let us play: give us time to get good at it, time to make more decisions about what we want to make, and time to craft at our own pace."

Heidi articulated what many learners feel but rarely name: the need for guidance with space, structure with freedom. She didn't want to be left entirely on her own, nor told exactly what to do. She wanted to be taught with, not at. To learn from someone more experienced while still steering her own creative direction. She wasn't rejecting instruction; she was rejecting instruction that left no room for her emerging agency in art. Her teacher, probably under pressure to meet curricular pacing requirements, emphasized coverage over ownership.

Over the years I've met students with very different relationships to learning: students like Angie, who needed to think in her own way; like Heidi, who needed room to originate; students like Marcus, whose learning was obvious everywhere except inside school; and students like Nancy, who had a startling re-visioning of what the lesson even was. They weren't more or less capable. They were differently oriented. Motivation language is insufficient because it can't name what students are doing as knowers.

How focusing on motivation unseats knowers

When teachers say we want students to be "motivated" or "self-directed," what do we actually mean? Motivated to comply, or to create? Self-directed, or directed within whose boundaries? Vague language obscures students' lived experience and weakens our ability to design environments that support their development as knowers. And if you support teachers, through coaching, walkthroughs, or policy, be aware that vague language doesn't just obscure, it limits perception.

Often in educational research, a term means one thing in psychology and something different in education. Edward Deci (1971) defined intrinsic motivation as "the doing of an activity for its inherent satisfactions rather than for some separable consequence," distinguishing it from extrinsic motivation. Later, Deci and Ryan (1994) tried to humanize classroom motivation by showing how students can come to endorse what school asks of them, especially when teachers support autonomy and students engage with a sense of choice and personal commitment. This is what they called identified regulation. That is, the student identifies what the teacher is doing as their own choice.

That distinction matters. In the day-to-day logic of schooling, intrinsic motivation often gets collapsed into identified regulation: endorsing what school assigns. In school-improvement talk the two get treated as interchangeable; if a student "buys in," we call it a win. But even when internalization occurs, students' motivation can remain bound by someone else's timing, goals, and values. Some critics, like Matusov, describe this as "benevolent control." Students solve problems; they don't pose them. Schooling trains students to adapt rather than to encounter, to fit meaning rather than make it. Identified regulation can explain endorsement, but it does not account for how students might question, revise, or reinvent curricular purposes. Leadership can confirm a room looks orderly and "on task," and it can still be a room where knowers are being unseated.

Schooling's central problem is its curriculum return-on-investment logic: coverage, pacing, deliverables. In that frame, what can be counted is what gets named, and what can't be counted gets treated as noise. This is not a teacher character issue. It's a systems issue, because pacing guides, benchmark calendars, and evaluation categories define what teachers can afford to notice. Teachers may be given some language for formative thinking, for partial models and "not-yet" explanations, but rarely the time or permission to cultivate those beginnings as learning. So motivation becomes the lever. When we don't have time for thinking, we reach for compliance. When we can't cultivate knowing, we manage behavior and call it engagement.

Nancy's case makes the disconnect visible. Nancy was engaged, careful, deeply absorbed. But because she had reframed the task inside her own interpretive frame, the teacher could not read what she was doing as math. He read it as a distraction. And once a teacher reads a student that way, "motivation" becomes the story: she got sidetracked; she didn't stay with the task. When teachers treat such moments as errors or distractions, they unintentionally teach students that their own ways of seeing are not important. In truth, these moments can be the beginning of rich discussion.

This is why I reach for the concept of agency, with its aspects of origination, decision-making, capabilities, and answerability. Agency, as I use the term in this book, is a developing capacity to act, with meaning and effect, towards an object of learning or its context, through which learners can shape purposes, participation, and possibilities. Agency arises between learner, task, and classroom community.

Students who have not experienced themselves as successful in school sometimes build confidence by practicing agency elsewhere: in work, sports, caregiving, or creative pursuits. That practice can support academic engagement; that's how powerful agency is for the individual. Often the life of agency is happening somewhere off-campus, a job, babysitting siblings, their own interest pursuits, and schools can either recruit it or punish it.

Genres of agency

In this section I map seven recurring genres of agency that I've noted across school contexts: four generative forms (engaged, epistemic, democratic, and authorial) and three shadows that differ in kind. Dysfunctional agency is the non-example; pseudo-agency is a decay; effortful agency is constrained, and sometimes useful. Each genre has a signature relationship to its school model's chronotope of culture, power, and historical sediment[2]: its rhythms, its authority, its definitions of legitimacy. Students may resist, survive, or thrive depending on how we set those conditions.

As I defined it earlier, agency is not the same thing as motivation or autonomy; endorsement does not equal authorship. Each genre is a pattern between learner, task, and classroom, and each exhibits one or more aspects of origination, decision-making, capabilities, and answerability.

The taxonomy is core to the book's purpose, and what is critically different about it is the focus: how student agency shows up in the classroom, not deciding whether individuals have or lack these forms. Being able to perceive the forms as they appear in student behaviors, discussions, or writing is what allows a teacher to take appropriate teaching action. Discriminating among them depends on where a move originates and what its object is.

There is a second way to say this, in the vocabulary of a theory I lean on later in the book. Variation Theory calls a lesson's target the object of learning (Marton; Moore-Anderson, Teaching Meaning), and my forms sort by their relation to it. Angie was messing around with how a concept worked: engaged, inside the object of learning. Nancy was in a different form of math altogether: epistemic, a different object for the same task. Marcus was driven by his own objects of learning: authorial. And democratic agency returns agency to students outside the object of learning, in the conditions that surround it. The map exists because we need to make sense of students' meaning-making, and meaning-making always has an object.

The Map

The first three agencies are shadow forms: what agency becomes when students as knowers are chronically unseated. (I focus on what we do have control over, our own classrooms, without suggesting classrooms are the whole or even primary cause of this displacement.) The last four are the forms student agency takes when the object is sense-making or meaning-making.[3] You have already met two of them in Angie and Nancy in Chapter 1.

Dysfunctional agency is the non-example that marks the definition's edge. The student is showing agency: origination, decisions, often considerable capability. What's misaimed is the object: the move points away from the object of learning, as refusal, disruption, sabotage, or disengagement. The capacity is fully intact, and it works against the student's own education, at least the education offered in the classroom.

Pseudo-agency is the appearance of agency without its substance: learning performed without sense-making, with an emphasis on maximal minimal effort. It is the decayed version of effortful agency.

Effortful agency is the high achiever doing real work toward goals they never set for themselves and don't care about. It gets good test scores and engages in sense-making, at least in the short term. Classrooms read it as virtue, but it leaves students divorced from their own subjectivity (Matusov, von Duyke, & Kayumova, 2016). And it is sometimes useful: persistence is a real capacity. The trouble comes when it is normalized as learning. Often the object of traditional school models.

Engaged agency is the student's effort to understand from their own position, still inside the teacher's lesson frame. It is easily misread, because it often presents in partial, frequently wrong, but authentic ways. Angie in Chapter 1. This form has three outward behavioral aspects, lurking, playing, and geeking out, which we will get to. It is the object of progressive school models.

Epistemic agency involves the student reframing the problem, proposing a different question, or making a counterclaim about what counts. It is meaning-making, not just sense-making. Nancy in Chapter 1. This rarely comes up in discussion in educational circles.

Democratic agency is student contribution toward the shared conditions and purposes of the classroom or the object of learning. It is voice acting on the world with others, not just voice as choice among limited options. Abundant in democratic-school settings.

Authorial agency is the student as author of their own intellectual life: authoring not just what they want to learn but how they want to learn it and how much. Problem posing, not just problem answering. Agents of their own subjectivities. The purpose of democratic school models. Marcus.

Please note that these forms exceed any single classroom's jurisdiction. The point is not to expect every teacher to host all of them, but to consider where an awareness of each might serve your students and your content. Engaged and epistemic agencies take the content as their object, and responding to them is what moves a class beyond delivering a curriculum toward developing the conversations that support students to metabolize meaning. Democratic agency addresses the shared conditions of learning, and classrooms differ in the affordances they offer for it. Authorial agency centers on the self across the school setting; your classroom may or may not meet a student on that journey, but a whole school might. These forms are bigger than one course, one teacher, one year. Awareness of them, for teachers and for students, opens possibilities.

Shadow forms of agency

The zombification of student agency

When we attempt to solve disengagement through motivation, we fail to ask the prior question: what is happening in classrooms that disengages?

Teachers, parents, and students have long sensed that something about schooling can feel off, and a vast body of research confirms their intuition. The institution designed to open the world to students can also stifle their eagerness to explore it. Alienating students from their own knowing can paralyze their capacity for purposeful action, producing what Llewellyn called zombies and I call the zombification of student agency: students whose lives have been sucked out of them while being asked to jump through academic hoops. Or as Miracle Max said, "Mostly dead, but not all dead."[4]

These zombies may be all dead (disengagement), mostly dead (agency sickness), or functionally dead (pseudo-agency). They may still be compliant. They may still be "successful." But their relationship to learning has thinned into performance: doing school without being in school as a knower.

Agency breakdown isn't just a personality problem or a motivation problem; it's often a legibility problem. A student's agency does not disappear. It becomes institutionally illegible, unacceptable, or underdeveloped in terms of academic learning, and is misread as defiance, laziness, distraction, immaturity, or "not motivated." It is agency operating under conditions that make voice, meaning-making, or authorship hard to sustain.

I begin with the three shadow forms because they are so common, so easy to misread, and so often produced by the ordinary pressures of schooling. They are still agency, but they are unproductive for classrooms.

Dysfunctional agency: protective and hostile forms

What often remains unexamined is whether schooling as constructed, even in the best of possible school worlds, is part of the problem itself.

Freire's banking model of education suggests information is "put" into students' heads the way you put money in a bank; in that analogy, resistance can be read as a response to being treated as a container. Angie's agency resisted alienating conditions. In my own 2016 paper, I bracketed resistance and non-normative agency as non-agency. I've come to see that dysfunctional agency extends past that earlier framework: agency is taken when students are displaced as knowers, and it goes somewhere. Disengaged agency looks like the student consistently putting their head down on entering the classroom; being on their phone, gaming, shopping, watching movies, texting, sometimes FaceTiming. More hostile actions include talking loudly in class and arguing about the right to do so, starting fights, and sabotaging classroom resources while remaining committed to not learning.

This is what I mean by institutional legibility. When a student's agency doesn't fit the time-world of school, it becomes institutionally illegible or pathologized: invisible to teachers, uncountable by the system. Disengaged agency resists the classroom most directly, but other dysfunctional forms seem functional while being adaptations to dysfunctional conditions. It's performance learning, which, you have to admit, compared to a classroom dominated by dysfunctional behaviors, is pretty darned exciting.

Pseudo-agency: the illusion of engagement

Pseudo-agency is the appearance of engagement without its substance. It isn't effortful; it's calibrated to be as low-effort as the student can get away with. Marcus sometimes slipped into pseudo-agency, turning in the bare minimum to stay under the radar, but his real energy was elsewhere. Nancy, by contrast, resisted pseudo-agency by reframing the task; her teacher just couldn't see it.

Priya kept talking during my lesson. Not disruptively, just steadily. When I finally said, "If you're talking, you're not thinking about the lesson, and that's what concerns me," she said, "Well, I hear you, so I don't know why this is a problem." She was taking notes, looking at my slides; she always turned in all of her work. Other students looked up in surprise: why was I questioning her learning?

I checked in with her after the lesson to see whether she was really that remarkable at multitasking, and no. She had no idea what the lesson was about. I had a moment of sudden daylight: Priya actually seemed to believe that hearing something was the same thing as learning it. When we moved on to the follow-up work, she gave it equal weight with her ongoing social conversation, copying enough into her worksheet to produce visible proof she was on task while staying tuned to the gossip her friend was sharing. Hearing the lesson, turning in the paper: both, by her lights, counted as doing school. It did not matter where the words came from: my slides, a neighbor, AI. Mentally engaging with the object of learning was not the point. Visible proof, as a proxy for engagement, was.

Priya was not trying to be difficult. She believed she was learning, having spent years in a system that rewards looking like a learner over being one. And to be honest, how many of us have thought that hearing a term was the same thing as learning it? She was not performing engagement cynically; she had internalized her classroom performance so completely that she experienced it as the way learning occurs. Conveniently, she was so unengaged with the learning that she could also maintain a long, complicated, newsy dialogue with her friend.

That is how we get what I call pseudo-agency: performative studenting without learning. Students perform the rituals of learning but never make real contact with the ideas, and over time they learn to comply with surface demands while disengaged internally. Not through one bad teacher or one bad year, but through the accumulated experience of schooling that consistently rewards the performance of learning over the thing itself. This is especially prevalent in high-poverty schools, as we will see. And when pseudo-agency becomes the norm, when whole classrooms operate in it, teachers reward it, and walk-through administrators are happy with it, it creates a culture that confirms everyone's belief.

Pseudo-agency students remind me of the patients in Oliver Sacks's Awakenings. His patients had lost the neurological capacity to initiate action; they could respond to external cues, a ball thrown, a song played, but could not begin movement on their own. The analogy captures the reactive quality of pseudo-agency: reactive paper production, not generative learning.

Pseudo-agency is the predictable outcome of systems that prioritize coverage over meaning, pacing over mastery, memorization over reasoning, and standardization over curiosity. One antidote is teaching with complexity: richer questions, authentic inquiry, real cognitive struggle. But complexity collapses under the logic of pacing. When teachers are pressured to cover curriculum and students are rewarded for compliance, both parties collude, consciously or not, in a charade of learning.

Pseudo-agency often looks like success: on-task behavior, turned-in assignments, decent grades. What it represents is the substitution of completion for comprehension. Students come to see themselves as doers, not knowers, and when intellectual curiosity shuts down, the very cognitive growth all that effort is meant to stimulate lies dormant.

Christian Moore-Anderson, watching the same thing in his own science classroom, calls it skilful coping: students can skilfully cope with a task and still not learn what we hoped, because they are good at learning the contextual cues that produce the right words at the right time, and task completion can resemble understanding, especially when every student is expected to give the same answer (Teaching Meaning). Priya was coping skilfully.

Effortful agency: autonomy constrained, agency under surveillance

Now the most institutionally rewarded form: effortful agency, the kind that keeps classrooms orderly and students busy. It is often mistaken for the whole of motivation. It is often encouraged through extrinsic rewards.

My son Ben and his friends in middle school grew tired of the endless stream of external rewards: tickets, candy, prizes dangled in front of students to motivate "good behavior." Tired of the manipulation and the behaviors it sought to produce, the boys began forging bootleg tickets and using their counterfeits to fund small student-run enterprises: hand-knitted scarves, homemade jewelry, snacks. In other words, they rewarded authentically chosen activities over being manipulated into activities the teacher wanted. Agency can emerge even under restrictive systems; it just doesn't always emerge in sanctioned ways.

Students who are consistently unseated from their knowing develop workarounds. They learn to read what the teacher wants and produce it, without necessarily understanding it, decoding the hidden question beneath the teacher's question: not how do I make sense of this problem? but what does the teacher want me to say? They participate in the forms of learning, discussion, note-taking, answering questions, without making personal meaning of any of it, performing the cognitive labor of translating their thoughts into teacher-expected outcomes.

Matusov, von Duyke, and Kayumova (2016) coined effortful agency: a mode of action grounded in compliance, persistence, and self-discipline within predefined boundaries. These students color-code their notes, never disrupt class, memorize and regurgitate, and push through confusion. Their effort is admirable, sometimes heroic.

Educators love effortful agency. A former school administration of mine wanted to increase learning efforts, so they showed a clip from a football movie: exhausted players sweating, running, then crawling, forcing the ball into the end zone for the winning score. I remember thinking that no student should, or could, live like that every day to learn. It was an example of what I call effort porn, revealing more about what educators fantasize teenagers should do than about what adolescents actually need.

Effortful agency shows initiative, but there is an ethical question at its core. As Howard Blumenthal asked in a recent conversation on education: if students spend their entire lives learning only what others assign, when do they get to define themselves through learning? (Hargadon, 2021.)

The psychological costs can be high. Research on valedictorians, the ultimate effortful achievers, suggests they are no more successful, and often less fulfilled, than their peers after graduation. I remember running into a former high school valedictorian working full-time at Staples. I expected her to be doing great things in the world by now; instead she told me she did not know what she wanted to do with her life, and until she did, she needed to pay the bills. Barker (2017) makes the point bluntly: the students who perform best on the system's terms are not, as a group, the ones who go on to change the world. They have become experts at the game without developing a reason of their own for playing it.

Effortful agency is a particularly U.S. virtue; the valorization of relentless striving has roots in Weber's Protestant ethic, shows up plainly in Millennial hustle culture, and is increasingly rejected by the more sensitive and artistic Gen Z'ers. In effortful agency the student becomes a self-managing ideal worker: constantly evaluating, adjusting, and optimizing themselves to complete teacher tasks, rarely pausing to ask why the tasks should matter to them. Anderson and Winthrop (2025) name this same pattern Achiever Mode: external goals internalized without being integrated into any coherent sense of self.

At its core, effortful agency is agency under surveillance: the appearance of self-direction sustained by external expectation. Schools that prioritize effort sometimes aestheticize struggle itself. Effort can keep classrooms functioning, and it builds real capacities for persistence. But when schools treat effort as the only available form of agency, they reward discipline while leaving meaning-making, and the student's developing self, thin.

It is worth pausing on the relationship between pseudo-agency and effortful agency, since the two often look alike. Effortful agency is the student who is heroically compliant, who works hard at whatever the teacher assigns. Pseudo-agency sits on the same continuum but further from learning: effortful agency still concerns itself with learning, while pseudo-agency only resembles it, sometimes even to the student. Priya gives us the diffuse, easygoing version; the valedictorian gives us the disciplined one. What they share is the loss of contact with their own thinking as the object of school. Why these forms are so prevalent, particularly in high-poverty schools, is diagnosed in the chapters ahead.

Engaged agency

Meaningfulness within structure

Effortful agency describes disciplined action under imposed goals. Engaged agency is participation that means something to the student, inside a shared structure. Students don't merely comply; they shape their experience of the work while staying in relationship to the task, the teacher, and the classroom norms. In its strictest sense it is Deci and Ryan's identified regulation.

Remember Heidi, who wanted her art teacher to "show us a skill, maybe a few examples, and then just let us play"? That's engaged agency. She wasn't rejecting instruction; she wanted guidance with space, to learn from someone more experienced while still expressing her own creative ideas. She's not dissimilar to Angie.

Engaged agency is not freedom from structure. It is agency exercised inside structure, when the classroom ecology legitimates students acting on instruction with their own purposes as it unfolds. Students name what isn't landing, request a different representation, practice new information in their own way, propose a method, or redirect attention toward a question that matters to them, all without leaving the lesson behind. Reeve and Tseng's (2011) concept of agentic engagement names the observable behaviors students use to contribute to instruction as it unfolds: asking, proposing, requesting, resisting, redirecting. Resistance being an important signal to the teacher that their lesson is not flying.

And Patall (2024) offers a detailed account of engaged agency and the conditions under which students experience it: classroom activities that are relevant, self-endorsed, and worth caring about. She discusses the importance of each in autonomy-supportive teaching. Patall's caveat is: autonomy support does not land evenly. Students' prior experiences of success, belonging, and recognition shape their ability to use autonomy supports at all. Those who have felt competent and recognized elsewhere often thrive in autonomy-supportive classrooms; those who have not may need real scaffolding before they can act independently. That's a cautionary warning to think about.

Engaged agency takes three distinct forms as learning progresses.

The shape of early engagement

1 · The frame: learning we can't recognize

Oregon Trail. Two education professors once observed a nine-year-old boy at an after-school program interacting with The Oregon Trail. The student seemed to click around randomly, glance around, doodle, and comment to friends. One professor concluded the child was mindlessly pressing buttons and needed instruction. Two days later, on his own, the second professor came across the same child playing with focus and proficiency. Within weeks he was highly competent.

The paradox isn't that the child learned quickly. It's that a trained education professor couldn't recognize learning while it was happening, because it didn't look like the way engagement is expected to occur in a classroom.

When children learn a new computer program, they rarely begin by reading the manual. They watch others, get a feel for what the program does, and then poke around: testing buttons, discovering limits, figuring out which actions produce which effects. This is how learning often unfolds: exposure, partial understanding, exploration, coherence, transfer, and sometimes invention. Children tolerate the anxiety inherent in learning, and their own early clumsiness, when conditions allow low stakes, time to linger, social safety, and a sense of play. Over time competence builds, and eventually understanding transfers: they learn new programs faster, customize tools, and are sought out by others as "experts".

In school, we often punish the behaviors that support engagement but precede competence.

2 · The three models each misread early engagement

Every school model has a theory of engagement. Traditional schools believe effort precedes interest: work hard enough and understanding arrives, with interest following. Progressive schools believe interest precedes effort: give students real choice and interesting work and they will engage willingly. Democratic schools believe self-direction precedes both: let students own their learning entirely and effort and interest take care of themselves.

Each theory has something true in it, and when engagement breaks down, each reaches for its signature tool: more structure, more choice, or more freedom. What none of them fully reckons with is that engagement doesn't arrive whole. It develops through phases, and the phases don't look like learning to eyes trained to find outputs.

3 · Ito's phases: a language for the evolution of engagement

In Hanging Out, Messing Around, and Geeking Out (2010), Mizuko Ito and her colleagues give us a language for these phases. They tracked young people's participation in voluntary, social, youth-driven learning contexts and found a recurring pattern in how interest deepens in the wild. "In the wild" is important, because classroom students are captive and often have to structure for engagement differently.[5]

Ito's "agency genres," as she calls them, name the student's observable behaviors at each phase and how each functions in building understanding. The phases are not developmentally sequential; a learner can be lurking in one aspect of a domain while messing around in another. They name an iterative process from orientation toward expertise. I don't list them as genres of agency in my taxonomy because, for my purposes, they sit inside engaged agency and add granularity to it. And the point of the phases is worth saying plainly: engagement changes the form of agency that gets expressed. Lurk to geek is the same kid.

4 · Lurking

Behaviorally it looks like students are present but not yet visibly participating. They hover near the action and language of a domain, watching, listening, absorbing. In school this looks like disengagement, because it produces no visible effort or talk. But Ito shows that this phase often marks the beginning of real orientation: What is this activity? Who belongs? What kinds of participation are possible? What is allowed and what is not?

As a homeschooling parent, I watched my younger child learn to read simply by playing near the phonics lessons of their older siblings. They got all the information without any pressure of evaluation. And a Montessori colleague told me about a student who refused reading lessons for three years in the multi-year 1st–3rd grade classroom typical of the method. This student always sat nearby, turned away, playing with something, quiet. Then one day he simply picked up a book and read. His teacher had been concerned, but was sensitive enough, as a progressive teacher, to note that something valuable was happening in all that adjacent sitting.

Lurking violates the rules of legibility in traditional schools: we expect thinking to be shown on paper and engagement to be verbally demonstrated. Silent proximity is a precondition for further learning. Like the silent period in early language acquisition, quiet is not disinterest; it is intense interest. The issue isn't whether learners are peripheral. It's whether their peripherality is treated as legitimate (cf. Lave and Wenger) or assumed to be a form of disengagement in need of correcting.

When a student's lurking is not seen as legitimate, they can be redirected in ways that pull their attention away from the very learning that is underway. Or, as teachers are often instructed, to assign collaborative roles, which means lurking students are expected to talk before they feel ready.

5 · Messing around

Behaviorally, participation becomes playful and exploratory. Learners poke at the learning thing, trying approaches, testing limits, learning through low-stakes failure. Messing around depends on time, access, room for wrong moves, and help on demand: conditions schools frequently fail to provide. It is a form of play, and like all play it lowers the anxiety of learning (Peter Gray).

The knee-jerk reaction of many teachers is: stop fooling around, or you're off task, without realizing these stop signals are aimed at exactly the behaviors we want. When students explore and fiddle with the materials they are using, say in a lab, they are engaging. Recently I introduced my class to microscope work, with no lesson. Inevitably students fiddled with dials and lights, tried to see a piece of their own hair, then tried to look at anything they could find; we had to bring out a stereoscope and a scalpel to see specimens in different ways. This is perfect. A lot of learning happens without a lot of direction or speech. The same goes for digital and balance scales, things on my desk, teaching tools. The classroom is for them, not me.

What I find sad these days is how unfamiliar students are with basic materials. Many of my ninth and tenth graders initially don't know how to use a ruler, among other tools. When I asked about their middle school experience, they told me their teachers didn't let them use the rulers because they didn't want them fooling around. The very behavior that signals engagement became a reason to deprive them of the tool altogether. When we take materials out of students' hands to keep a room manageable, we are not only preventing mischief; we are foreclosing the play stage that later abstraction is built on. The result surfaces in my class as a tenth grader who can't use a ruler.

I get the anxiety. We've had our accidents. But students given a chance to fiddle around within reason will eventually settle, because their curiosity will be satisfied as the forbidden thing in their midst is accessible.

Messing around also occurs in dialogue. Students make what seem like off-topic comments, yet these can be a form of play. In museum education, researchers followed a father and daughter talking their way through a heart exhibit. Their conversation was full of off-topic subject matter, connections, and digressions. Yet in following visitors' retention of their learning, those who had conversations like this not only remembered the information; some expressed greater complexity ten years later than in their immediate post-exhibit understanding. People engage most deeply when they are allowed to bring knowledge into the context of their own lives in order to make sense of it: what I earlier called metabolizing meaning.

The part we get wrong is assuming this is something children grow out of. It is not age-bound, and it is not a type of learner. It is typical of any age when material is new.[6] I will never forget my Upper Elementary Montessori training at the Barrie School in Silver Spring, Maryland. We were five women at the lesson with our male teacher. We had all taken geometry in high school, yet here we were, giggling and gushing over the geometry materials used in problem solving. They let us physically reshape solids into the forms the equations describe, and feel where the volume formula comes from. We all said, "Oh, now I understand."

Never did one man make so many women so happy at one time.

I am the same way with new technology. I have to get my hands on it first. I learned how to build AI agents while writing this book: I need to see what a thing does first, and only then do I read the directions, and only the parts I need. It is not hands-on in the literal sense, but it is embodied; I have to act in the thing before the abstraction about it means anything. And typically I learn best by first applying it to my personal life, like creating an agent to grocery shop for me. Seen this way, a tenth grader handling a ruler for the first time and fooling around is not regressing to a childish stage. He is apprehending something new in a natural progression.

This is why I have come to distrust the language of learning styles, which has fallen out of favor for good reason.[7] When we call a student a "kinesthetic learner," we name as a fixed trait what Ito described as an orientation: where he currently stands in relation to this material. A style, if it were real, would be fixed; Ito's work shows how the mode changes as understanding grows.

Malik: the scene. Malik's story shows the phases coexisting.

Malik was a biology student who usually came in, lay across a lab counter, and went to sleep, or watched movies on his device. One thing I did notice: when he was awake, he fiddled with anything he could get his hands on. My pen got taken apart and reassembled. A rubber band got stretched and snapped. Lab material got examined from several angles before being used for usually unintended purposes.

So one day I had the Forensics class's materials out on a side table: a test for blood, and a simulation I'd built for blood spatter analysis. I typically cross-seed my students' interests this way. Malik came in, looked at the table as he walked by, then flopped on his favorite counter. I followed him and said, "Do you want to test for blood like a CSI?" He lifted his head. Most kids have seen CSI shows; in that sense they've already lurked.

He was up in thirty seconds, at the table, examining the materials with focused attention. I showed him briefly what each did and left him alone. No paperwork to fill out.

While he worked on the blood test simulation, I continued my ecology lesson with the rest of the class: energy pyramids, the transfer of matter and energy through an ecosystem, the vocabulary that names how a hawk and a mouse and a blade of grass are all part of the same energy economy at different levels. Malik, at the back table with his blood spatter kit, listened.

I didn't know that until later, when I brought a card activity over to him after my lecture: a set of physical materials I'd made that let students build an energy pyramid by placing organisms in the right positions, using the lesson's vocabulary to label each placement. Malik laid the cards out without asking for help, named several positions correctly, and caught many of his own mistakes. When I came over I pointed to where something didn't fit the definitions; he reworked the arrangement twice until I said it was complete. He had been lurking the entire time, tracking the lesson from the counter, absorbing the vocabulary. Letting him mess around with something else had also let him lurk in on my lesson, then be ready to try the materials until they worked. Like the Montessori non-reader, he was learning; it just didn't look like it.

School models misread lurking and messing around. The traditional model misreads the student who needs to lurk before engaging, and offers more structure and accountability. The progressive model misreads the lurker, the student fiddling at the edge of a lab, the off-topic connector who needs to mess around before exploration means anything. The democratic model, to its credit, supports children in lurking and playing; it places a high value on both. Watching older children at play is complex and interesting, and so limited in most schools. Yet when democratic educators do provide a requested lesson, they are just as likely to teach it with the same misreadings as the other models. More about school models in the next chapter.

6 · Geeking out

Behaviorally this is where interest develops in dialogue and writing, and finally looks like expected outcomes. Learners seek competence because the domain has begun to matter to them. They pursue mentors, tolerate frustration, refine skills, adopt standards of practice, and determinedly use the vocabulary in their speech and writing. Identity starts to form: someone who knows something about this domain.

Students who have actually worked their learning out tend to share it without being asked; new knowledge wants to be expressed. Peer transmission is one of the most powerful learning mechanisms in a classroom, but only if it is not forced. It comes not as dialogue roles assigned before students are ready, but as the natural overflow of understanding when someone finally has something to say.

A great example is one of my sons-in-law, a carpenter who works much of his day alone. He loves history and politics, so he listens to podcasts and audiobooks about the presidents, capitalism, progressivism, all day. We have come to treat him as the one to check our historical facts on presidents, or to define political terms. He is becoming someone known for knowing something. When he comes home he can't wait to talk about what he has been listening to and thinking about.

Lurking asks: What's going on? Messing around asks: How does this work? Geeking out asks: Why does it work this way, and can I explain it to you?

That sequence matters, because high school routinely demands behavior more like geeking out from students who have never been given time to lurk or mess around. We want students to geek out; yet we cut off the phases that make geeking out possible. We are promoting anti-engagement in practice.

One bridge before moving on. Engaged agency fails in classrooms for students like Angie, who is engaged, just not the way the teacher wants her to be: be engaged in what I'm telling you, not in what comes from your own thinking. What engaged and epistemic agency ask of a teacher's perception, and of the follow-up move, differ; that difference is where we go next.

Epistemic agency

Reshaping knowledge

Partially correct answers, students' "not-yet" understandings, matter enormously for teaching and for engaged agency, but they are not what I mean by epistemic agency.

Epistemic agency shows up when a student reframes what counts as the task, what counts as a solution, or what kind of knowledge is being demanded. It is interpretive power: the learner doesn't only work within the knowledge assigned; the learner reshapes what counts as knowledge in this classroom, and sometimes the direction of the lesson itself.[8] Unlike engaged agency, which builds meaning through participation in the teacher's lesson, epistemic agency enacts meaning through reinterpretation of that lesson. You saw this in Nancy's coin flowers in Chapter 1.

Epistemic agency is also a skill, and one a person can learn to love. I did. Many people hit it without ever being supported to carry it through. It's funny: I never thought I was smart or a good student, because my brain always wandered to other possibilities in school. I remember being told in eighth grade that Roman numerals had to be changed because you couldn't do math with them, and I spent the rest of the class trying to work out the four operations in Roman numerals. Could you add? Where exactly did it break? Like Nancy's teacher, mine looked at my work, grunted, and moved on.

Democratic agency

Influencing the context

Where epistemic agency reframes meaning within a structure, democratic agency turns interpretive power outward, toward the shared conditions of learning in the classroom community itself. Democratic agency is shared authorship of the rules of the game: students help shape norms, procedures, and consequences, and learn to live with one another through deliberation and accountability.

One of my doctoral research sites was The Circle School (TCS), a democratic school on the East Coast now approaching its fourth decade. At TCS, students choose what to learn, whether to learn, and with whom. Their civil liberties are codified in a constitutional school document, and they vote on issues ranging from schedules to discipline to budgets and staff hiring. Children as young as four deliberate alongside teenagers in school meetings. Adults act as stabilizers of the school culture, anchoring norms and modeling judgment without dictating how learning unfolds. The school's approach is described more fully in the next chapter.

Authorial agency

Beyond the frame

In effortful, engaged, epistemic, and democratic agency, students are regulating themselves within structures largely framed by adults. Authorial agency begins when learners create curricular purposes inside school: when they generate new directions of study and author options rather than choose among them. It extends autonomy beyond selecting from a menu to authoring the menu, within the constraints and opportunities in and outside of a classroom ecology.

I define authorial agency in a bounded way: it begins when learners create curricular purposes for themselves, and I focus on authorial agency that occurs inside school though it isn't limited there. Culturally, we claim to value this deeply; innovation sits at the center of the American narrative. And as artificial intelligence becomes widespread, humans will need to get better at selecting goals worth pursuing, not simply completing tasks efficiently. If students spend their entire lives learning only what others assign, when do they get to define themselves through learning?

Authorial agency stands out in a school setting; outside one, it is ordinary. Almost everyone authors some portion of their life behind the scenes. It appears more readily outside school, in student-led businesses, skilled work, and social media, than within it. In schools it emerges unevenly: through independent study, long projects, mentorship, elective pathways. The teacher functions as curator, coach, or gate-opener rather than director.

There's a greenhouse off my current classroom, long forgotten, until a student, Peter, expressed interest in using it. Over time he moved from plant experiments to entrepreneurial work: growing and selling plants, developing herbal products, and forming a club of other students whom he recruits and leads. He defines his agenda, sets standards for quality, recruits collaborators, and decides the requirements they should meet. Peter's work receives one credit, Teacher's Assistant, though it certainly deserves more; the school has no independent-study pathway. He should receive three credits for special project work, then maintain a portfolio as his ongoing record; but the school's credit structures weren't designed for learning that students author. They were designed for learning that teachers assign. His entrepreneurial learning will live on his transcript's margins.

Authorial agency reminds us that learning is not only about understanding the world but about creating it. It is learning that doesn't easily standardize. Our task as educators is not to direct that impulse but to notice it, make room for it to take form, and get our schools to legitimize and credit it.

Multiple agencies can coexist in a single student's experience. A learner may express effortful agency in one subject and engaged agency in another; may demonstrate democratic agency among peers but retreat from authorial expression in a classroom that treats learning as compliance. The next chapter traces how the different schooling traditions make some genres visible and others invisible.

Key takeaways: the forms, in the students who showed them

The non-example label on dysfunctional agency is doing real work: the student is showing agency, every component present, just not toward the object of learning. That is what makes it diagnostic rather than damning. It matters because the object of learning is never treated neutrally in a classroom: it is elevated to a moral height or vilified, so agency aimed away from it gets read as character rather than as aim. The three shadows differ in kind, not just in shade: a non-example, a decay, and a constrained form with real uses. The four generative rows are what the definition looks like when it lands on the object of learning.

Read the names as moments, not labels. The taxonomy classifies what showed up in the moment. It never classifies the student.

Endnotes — Chapter 2

1. Bakhtin said answerability means there is "no alibi in Being"; the art thief is the one person whose whole craft is the alibi. Bakhtin, M. M. (1993). Toward a Philosophy of the Act. University of Texas Press. [VERIFY edition/wording.]

2. Time-world / chronotope: I use "time-world" to name the patterned expectations a setting carries: what counts as learning, participation, and legitimate purpose. For readers who want a technical term for this kind of patterned space-time in human activity, Mikhail Bakhtin calls it a chronotope (Bakhtin, 1981).

3. Sense-making vs. meaning-making: Sense-making refers to organizing understanding within a given frame — figuring out how a model works, why a process happens, or how an explanation hangs together. Meaning-making includes the personal, relational, or ethical significance learning takes on in a life — why this matters to me, how it connects to who I am, what it obligates me to notice or do.

4. Agency sickness — first described in Llewellyn's The Teenage Liberation Handbook. "Zombification" — Grace Llewellyn, The Teenage Liberation Handbook (1991). Miracle Max, The Princess Bride.

5. Ito source: These three modes adapt Mizuko Ito and colleagues' ethnography of interest-driven participation — hanging out, messing around, geeking out (Ito et al., 2010). Their gradients described escalating depth outside school, where interest can be assumed. I am asking what the same gradients look like inside a required biology class, where interest cannot be assumed and has to be built, which is also why I call the first mode lurking rather than hanging out. Nor are the phases confined to voluntary settings: I have watched them run in semi-captivity, in Montessori classrooms, and in captivity, in my own required biology class; Malik is the demonstration. What changes across settings is not whether the phases occur but what the teacher must supply for them to be legitimate.

6. Bruner / Montessori / Piaget: The claim that unfamiliar material is met concretely first has a long lineage. Jerome Bruner described learning as moving through three modes of representation — enactive (knowledge held in action and manipulation), then iconic (images), then symbolic (language and abstraction); Montessori said it more plainly — the hands are the instruments of the intelligence — and Piaget's movement from concrete to formal operations tells a kindred story. Bruner was explicit that this is not age-bound: adults meet unfamiliar domains the same way, which is why a good curriculum spirals, returning to the same idea at successively higher modes — each truly new encounter starts low again. The phases (lurking, messing around, geeking out) co-occur and are not rungs; the modes of representation are a true progression, but a recursive, domain-by-domain one rather than an age-based one.

7. Learning styles: Careful reviews found no adequate evidence that matching instruction to a student's supposed "style" improves learning (Pashler et al., 2008), and more recent meta-analytic work reinforces that conclusion; studies of the belief itself find people tend to treat styles as something a person is born with (Nancekivell et al., 2020).

8. Episteme / epistemic: I use episteme to mean a framework of knowledge: the assumptions and norms that define what counts as a valid explanation, representation, or method within a discipline and a classroom. Following Michel Foucault (1970), an episteme is also historical — what counts as legitimate knowledge shifts across eras.

Extrinsic rewards: A substantial research tradition suggests that external rewards can undermine intrinsic motivation in some conditions. Extrinsic incentives can be effective when the endpoint is clear and procedural, but they can also suppress creativity, problem-solving, and the stamina complex learning requires.

Effortful agency: Matusov, E., von Duyke, K., & Kayumova, S. (2016). Mapping concepts of agency in educational contexts. Integrative Psychological and Behavioral Science, 50(3), 420–446.

Honesty footnote: I am making a developmental claim here — that unfamiliar material tends to be met concretely, in the hands, first — and not the stronger prescriptive claim sometimes attached to it: that deliberately fading instruction from concrete to abstract representations reliably improves outcomes. That prescriptive version (sometimes called concreteness fading) has more mixed empirical support, and I don't rest anything on it.