Unlock the Intrinsic Motivation of Knowledge Workers

Unlock the Intrinsic Motivation of Knowledge Workers: extrinsic rewards degrade heuristic work, per candle-problem research and Hackman's two authorities.

Most leaders treat Unlock the Intrinsic Motivation of Knowledge Workers as license to remove management; hand out autonomy, flatten the chart, and expect self-direction to fill the gap. SAFe Principle #8 makes a narrower, more specific claim: extrinsic rewards measurably degrade performance on the heuristic, open-ended work that fills most knowledge workers’ days, and the fix runs through what leaders actually delegate, not through how far they step back.

---

Where this article sits

Journey stage 1 of 7: Readiness

readiness → use-cases → roi → pilots → kpis → operationalize → scale

this articlelinkedjourney stagepillarrelated (no direct link)

Your trail so far

The articles you visit light up on this map.

What Unlock the Intrinsic Motivation of Knowledge Workers Means as SAFe Principle #8

SAFe Principle #8 states that knowledge workers perform best when leaders stop managing their tasks directly and instead build the conditions for self-direction, because, as SAFe’s own Principle #8 page puts it, introducing an extrinsic reward to heuristic, exploratory work ultimately degrades the performance it was meant to improve.

The demand hides inside that permission. Building conditions is a harder discipline than issuing instructions, and most leaders underestimate the shift because “stop managing tasks” sounds like less work, when the practice actually asks for more precision about what to delegate and why.

Principle #8 as SAFe 6.0 States It: Daniel Pink and Peter Drucker

SAFe opens its own framework page for Principle #8 with Daniel Pink’s line, “It appears that the performance of the task provides its own intrinsic reward…this drive…may be as basic as the others”, and pairs it immediately with a definitional claim from management theorist Peter Drucker.

Drucker’s line, quoted directly on SAFe’s own Principle #8 page, is the operative sentence: knowledge workers “know more about their job than anybody else in the organization,” and that fact is part of the definition of the role, not an incidental observation about skilled staff. The consequence SAFe draws from it is structural: a leader attempting to specify how a knowledge worker should do their job is, by definition, working from less information than the person doing the work. Traditional task management assumes the opposite: that the person above the task understands it better than the person performing it, an assumption that held for supervised manual labor and collapses for specialized technical and creative work. SAFe frames the correction as an environmental one; leaders foster an empowering, supportive work environment rather than directing the mechanics of the task itself, because the domain knowledge required to direct those mechanics no longer sits with the leader. That reframing is why Principle #8 sits inside the broader set of SAFe Lean-Agile Principles as a leadership behavior change rather than a team-level practice: it changes what a Release Train Engineer or product manager is qualified to instruct, not just what a team is permitted to decide.

Extrinsic Versus Heuristic Tasks: Where the Traditional Incentive Model Stops Working

The traditional reward-and-punish incentive model works reliably on well-defined, linear tasks and stops working, sometimes sharply in reverse, on heuristic tasks that have no single obvious path to a solution.

SAFe’s own framework page draws this line with a concrete example: assembling a vehicle transmission is a well-defined task with a clear, linear sequence of steps toward a known result, and extrinsic incentives, pay per unit, bonus per quota, align cleanly with that structure because doing the task faster or more accurately is exactly what the reward is measuring. Heuristic tasks invert the shape of the problem. They require creative thinking, exploration, and experimentation to reach a goal that isn’t fully specified in advance; diagnosing an unfamiliar production incident, deciding how to sequence a backlog under conflicting stakeholder pressure, designing an integration between systems that don’t share documentation. Most of the work performed by knowledge workers falls into this second category, and SAFe’s finding on it is direct: introducing an extrinsic reward to heuristic work ultimately degrades the performance it targets, because the reward narrows attention toward the measured outcome and away from the exploratory search the task actually requires.

Task Type	Example	Effect of an Extrinsic Reward
Well-defined, linear (algorithmic)	Assembling a vehicle transmission	Reliably improves speed and throughput
Heuristic, exploratory	Diagnosing a production incident, sequencing a contested backlog, designing an undocumented integration	Narrows focus and degrades problem-solving performance

The practical filter for a leader deciding whether to attach a bonus, a spot award, or a quota to a piece of work is this table’s left column, not the seniority of the person doing it: a senior engineer doing algorithmic work responds to extrinsic incentives the same way a junior one does, and a junior engineer doing heuristic work is exposed to the same degradation a senior one is.

Autonomy, Mastery, and Purpose: SAFe’s Own Three-Factor Framework

SAFe names its own three-factor model for fostering intrinsic motivation directly as Autonomy Mastery and Purpose, credited to the May 2023 update of SAFe 6.0 attributed to Dean Leffingwell.

In the blog post announcing that update, Leffingwell frames Principle #8 and its sibling, Principle #9 (Decentralize Decision-Making), as the “home” for SAFe’s thinking about the human element of agility; happy, motivated, empowered knowledge workers solving problems and increasing the value they deliver. The same post ties this explicitly to SAFe 6.0’s broader emphasis: the mirrored version of the update notes that flow, measurable, outcome-based value delivery, became the central organizing theme of the 6.0 release rather than an abstract question of whether a team “is Agile,” and Autonomy, Mastery, and Purpose function as the human-side counterpart to that flow focus. Autonomy is the authority to decide how work gets done; mastery is the room to get progressively better at solving hard problems; purpose is a clear line from the day’s work to something that matters beyond the task itself. None of the three functions as a one-time grant: each requires a leader to keep re-checking whether a decision, a process change, or a compensation policy is quietly withdrawing autonomy, mastery, or purpose it previously protected.

---

What Principle #8 Is Actually Rejecting: The Taylorist Inheritance

Principle #8 is not a rejection of management in the abstract: it is a rejection of a specific, dated, and named model called scientific management, built by Frederick Taylor in the early twentieth century and institutionalized as mainstream practice by Ford, General Motors, and Bell during the 1930s.

Naming the target changes the argument. “Traditional management” is vague enough to dismiss; a model with a name, an inventor, and a documented set of adopters is a historical claim a leader can check against their own org chart, and most will recognize pieces of it still running underneath their current process.

Scientific Management: Taylor’s ‘One Best Way’ and the Staff-Line Split

Taylor never called his own method Taylorism; he called it scientific management, and Michel Balle’s Gemba Coach column for the Lean Enterprise Institute traces exactly what that label meant in practice.

Taylor’s method was empirical in a specific sense: he studied the most effective workers performing a task and synthesized their observed behavior into a single prescribed sequence, the “one best way”, that could then be taught to everyone else. The split this created is the structural piece that matters for Principle #8. Staff specialists did the studying and produced the answer; line workers were taught, or in practice frequently forced, to apply that answer without needing to understand why it worked. Balle’s perspective situates this inside a larger claim about what “science” meant to Taylor’s generation: a Newtonian universe that could be mapped piece by piece until the full mechanism was known and could be handed down as instruction. This assumption is the one Drucker’s knowledge-worker definition, opened earlier in this discussion, was written to overturn; and the practical effect on this staff-line split specifically is that it loses the justification it needed to keep issuing detailed task instructions.

How Ford, General Motors, and Bell Made Taylorism Mainstream

Scientific management did not stay a niche efficiency technique; Ford, General Motors, and Bell built their production and management systems on Taylorist principles during the 1930s, which is how a staff-line split designed for a single factory standard became the default assumption for corporate management for the following century.

That institutionalization matters because it explains why the split persists long past the manual-labor context it was built for. Once staff specialists devising the “one best way” and line employees executing it became the standard organizational shape across the largest employers in the country, later generations of managers inherited the shape without inheriting the reasoning behind it; they managed knowledge workers the way their own managers had managed assembly-line workers, because that was the only management model widely taught and modeled. Modern alternatives to a centrally imposed playbook exist and are documented; the Agile Alliance’s Agile Playbook explicitly frames guidelines, rules, and guardrails as something a team assembles for itself rather than something a staff function hands down, which is close to the opposite instinct from Taylor’s one-best-way model even where it uses the vocabulary of standards and methods.

The Inversion: Why Drucker’s Knowledge Worker Is Taylor’s Premise Reversed

Drucker’s knowledge-worker definition inverts Taylor’s premise, and the inversion’s real casualty is structural, not just a matter of who knows more: it strips the staff-line split of the justification it needed to keep issuing detailed task instructions. Scientific management could defend telling a line worker exactly what to do only because the staff side supplied information the worker lacked; once that premise no longer holds, the instruction survives as habit, not as anything a leader can still defend on its original grounds.

The consequence runs further than swapping who holds the answer: it changes what kind of answer is even available. Lean’s own tradition makes the same argument from a different angle: lean deliberately avoids handing down fixed solutions and instead frames improvement work as “countermeasures” to specific problems, because performance is a reflection of the competence of the people doing the work, not a property of a process design that can be perfected once and distributed. Developing that competence, teaching people to solve their own version of a problem rather than applying someone else’s prescribed answer, is the harder, slower alternative to a one-best-way manual, and it is the alternative Principle #8 commits a leader to once the staff-line split is off the table.

---

The Candle Problem: Why Extrinsic Rewards Degrade Heuristic Work

Two experiments, seventeen years apart, turned this from a manager’s hunch into a measured result: Karl Duncker ran the original creative-problem-solving test in 1945, and Sam Glucksberg’s 1962 replication added the piece Duncker’s version didn’t test: a direct comparison between incentivized and unincentivized subjects solving the identical puzzle.

The mechanism is cognitive, not motivational: incentive pressure narrows attention toward the fastest visible path to the reward and away from the wider search a heuristic problem requires: a measurable effect in a controlled setting, not just something observed anecdotally in a sprint retro.

Duncker’s 1945 Candle Problem and Glucksberg’s 1962 Replication

Karl Duncker’s 1945 Candle Problem, as InfoQ’s Celina Refauter describes it in a piece reviewed by Shane Hastie, asks a subject to fix a candle to a cork wall using only a box of thumbtacks and a book of matches, arranged so that no wax drips onto the table below when the candle burns.

The obvious approaches, tacking the candle directly to the wall, melting wax to glue it, fail; the solution requires noticing that the thumbtack box itself can be emptied and repurposed as a candle-holder tacked to the wall, a reframing step rather than a mechanical one. Sam Glucksberg’s 1962 experiment built directly on Duncker’s test, comparing how quickly different groups of subjects solved it under different incentive conditions. The reframing step is the part a reward degrades: subjects under time or cash pressure fixate faster on the obvious, failing approaches and take measurably longer to notice the box’s second function, because incentive pressure narrows the search rather than accelerating it. That result generalizes past a psychology lab: any task that requires noticing an unconventional use for an existing resource, a workaround for a broken pipeline, an unusual read of a stakeholder’s actual priority, is structurally closer to the candle problem than to a transmission assembly line, and the same narrowing effect applies.

Relatedness, Competence, and Autonomy: The Three Needs InfoQ Names

InfoQ’s coverage of the candle problem credits three basic needs with driving human intrinsic motivation once extrinsic reward is off the table: relatedness, competence, and autonomy, and satisfying them has a documented performance-enhancing effect in complex or agile working environments specifically.

Relatedness is the presence of genuine social connection to the people a person works with, not just proximity to them. Competence is the sense that a task is challenging enough to test real skill without exceeding it: the same zone the candle problem’s reframing step sits inside. Autonomy is the room to choose how a problem gets approached, which is the direct organizational analog of removing the incentive pressure that narrowed Glucksberg’s subjects’ search. Agile Alliance’s writing on the psychology underneath agile practice makes a related point from a different angle: agile methods that get treated as purely mechanical process, standups, story points, velocity charts, without attention to the psychological needs underneath them tend to produce compliance rather than the problem-solving behavior the mechanics were meant to enable. Naming the three needs gives a leader something more specific to check than “is the team motivated”; which of relatedness, competence, or autonomy is actually missing on this particular team, this quarter.

Why the Brain Resists Boring Work, Per Stefan Falk

Boring, unchallenging work resists sustained effort for a specific neurological reason: Stefan Falk, writing in Harvard Business Review, traces it to a brain that rewards spending mental energy on expanding one’s own capability, but rewards conserving that energy even more, which is why tasks that don’t spark curiosity get avoided or rushed.

Falk’s framing reverses the usual advice to “push through” tedious work. If the resistance is a design feature of how the brain allocates mental energy rather than a discipline failure, pushing harder against it fights the wrong mechanism. The practical move Falk recommends instead is reframing: turning a tedious, unchallenging task into an interesting or challenging problem to solve engages the same energy-expanding reward pathway that makes hard problems feel worth the effort, rather than waiting for the task itself to become inherently interesting on its own. That reframing move connects directly back to the candle problem’s mechanism; both depend on approaching a task with attention wide enough to notice an unconventional angle, and both show that attention narrowing under pressure or boredom in the same direction, toward the first available answer rather than the best one.

---

Autonomy Without Anarchy: What Leaders Actually Delegate

Autonomy under Principle #8 is bounded, not unlimited; self-managing teams hold authority over exactly two things, their work and the process they use to do it, and do not hold authority over who is on the team or what the team’s purpose is.

That boundary is the part leaders skip when they hear “autonomy” and either overcorrect into a hands-off vacuum or refuse to delegate anything meaningful for fear of losing control. Neither response matches what the research on self-organizing teams actually describes.

Self-Organization Is Not Anarchy: Hackman’s Two Types of Authority

Self-organization does not mean a team operates without structure or guidance; Mountain Goat Software’s account of self-organizing teams is explicit that the team organizes itself around tasks, responsibilities, and collaboration within boundaries its leaders set, not in the absence of them.

Richard Hackman’s authority hierarchy, cited in that account, draws the boundary in two parts. Self-managing agile teams have authority over their work, the decisions about how to build what’s been asked for, and authority over their process, the decisions about how they collaborate, sequence, and communicate while building it. They do not have authority over who is on the team, and they do not have authority over the team’s purpose; both of those remain leadership decisions. The distinction matters because it converts “give the team autonomy” from an all-or-nothing posture into two specific, delegable authorities a leader can hand over deliberately while retaining two others deliberately. A Product Owner who lets a team decide its own sprint sequencing but still owns which team gets staffed on which initiative and why that initiative matters is operating exactly inside Hackman’s boundary, not violating Principle #8 by withholding something.

Philip Anderson’s ‘Guided Evolution’ and the Leader’s Remaining Job

Philip Anderson’s contribution from The Biology of Business, quoted in the same Mountain Goat Software piece, names the job that remains for a leader once the two bounded authorities are delegated: management commits to guiding the evolution of behaviors that emerge from the interaction of independent agents, instead of specifying in advance what effective behavior is.

Guided evolution is a harder discipline than either extreme it sits between. It is not the staff-specialist role of designing the “one best way” in advance, and it is not stepping back entirely and hoping good behavior emerges unassisted: it is watching what a team of independent agents actually produces once they’re working inside their two bounded authorities, and shaping the conditions around them rather than the specific decisions they make. The Agile Alliance’s writing on flat structures makes the same point from a structural-change angle: flattening a hierarchy without first understanding what actually drives the people inside it tends to produce confusion rather than autonomy, because removing the visible authority structure doesn’t automatically supply the guided-evolution work a leader still owes the team. Lean’s own coaching literature on motivating adoption reinforces the involvement half of guided evolution directly: people doing the work generate the best improvement ideas, and organizing a team to catch and act on those ideas, rather than importing ideas from outside and asking for compliance, is what ownership of improvement actually looks like in practice.

The User Story Test Case: Who Actually Writes Them

Who writes a team’s user stories is a concrete, weekly-recurring test of how seriously an organization applies Principle #8, and Scaled Agile’s own guidance answers it directly: agile teams should write most of the stories on their own backlog, not the Product Owner alone.

The reasoning given traces straight back to Drucker’s original definition; teammates become experts in what they are doing because most people want to succeed at the work in front of them, and the Product Owner’s role is to manage and prioritize the team backlog, not to author the detailed statement of every piece of work inside it. Treating “who writes the story” as a Principle #8 decision heuristic rather than a fixed org-chart rule changes how a Scrum Master or RTE resolves the recurring friction point where a Product Owner, under delivery pressure, starts writing stories themselves to move faster. The short-term speed gain trades against the exact authority Hackman’s hierarchy assigns to the team, their work, and a pattern of the Product Owner reclaiming that authority under pressure is a reliable early signal that a team’s Principle #8 practice is eroding, even when nobody has announced a policy change.

---

Compensation Under Principle 8: Paying People Without Crowding Out Intrinsic Motivation

Compensation under Principle #8 is not a solved problem the framework hands a leader: it is the question every executive raises immediately after hearing “stop using extrinsic rewards,” and the honest answer combines a narrow, deliberately compressed salary band with a wider diagnostic for the non-monetary motivators money alone will never satisfy.

Skipping this section is the most common way Principle #8 gets adopted only halfway: leaders remove spot bonuses and performance-tied incentives from day-to-day task management, then never resolve what that implies for the annual compensation conversation, and the resulting ambiguity tends to erode trust in the change faster than the old incentive model did.

Narrow-Band Compensation: Cottmeyer’s LeadingAgile Case

Mike Cottmeyer’s account of applying Dan Pink’s Drive to real compensation design, published on LeadingAgile, describes keeping all of the firm’s coaches inside a deliberately narrow salary range, with an active effort to narrow that range further over time.

The goal behind the narrow band is specific: reward individual performance without encouraging internal competition, local optimization, or a dynamic where one person feels rewarded while a peer on the same team feels punished. A wide salary spread inside one team creates exactly the local-optimization pressure Principle #8 is designed to remove. It reintroduces an extrinsic, comparative signal into work that’s supposed to run on autonomy, mastery, and purpose instead. Cottmeyer’s approach doesn’t eliminate performance differentiation entirely: it compresses the range enough that compensation stops functioning as the primary signal of standing on the team, which frees relatedness and mastery to carry that signal instead.

What Fair Base Salary Doesn’t Solve: Real Extrinsic Needs

A fair base salary with no performance component doesn’t work for every person on a team, and Cottmeyer’s own account concedes the point directly: the realities of a mortgage, consumer debt, household expenses, or saving for a child’s education create extrinsic needs that intrinsic-motivation theory doesn’t erase just by naming them.

Some people want to come to work, do solid work, and go home; others want to advance quickly and be compensated visibly for exceptional effort, and treating both groups identically under a flat, no-differentiation policy risks its own version of the problem Principle #8 is trying to solve: a compensation structure that ignores real variation in what motivates different people is no more attuned to the individual than Taylor’s one-best-way model was. The practical resolution isn’t a single formula; it’s holding the narrow band from Cottmeyer’s approach as the default while building a small number of legitimate, transparent paths, a senior technical track, a coaching or mentoring track, for the people whose extrinsic needs or ambitions require more differentiation than the band allows.

The Champfrogs Checklist: Ten Named Intrinsic Motivators

Jurgen Appelo’s Champfrogs Checklist, developed for Management 3.0, names ten specific intrinsic motivators and functions as a diagnostic tool: a team’s motivation problem is rarely “no autonomy” in the abstract, it is usually one specific, identifiable item missing from this list.

Motivator	What It Names
Curiosity	Having plenty to investigate and think about
Honor	Personal values reflected in the group, boosting loyalty
Acceptance	Approval from the people around you
Mastery	Work that challenges competence without exceeding it
Power	Enough room to influence what happens
Freedom	Independence in one’s own work and responsibilities
Relatedness	Good social contact with colleagues
Order	Stable rules and a predictable environment
Goal	A clear purpose behind the work
Status	A recognized position relative to others

Appelo describes the checklist as “a simple tool for change agents to match their ideas for change with people’s intrinsic motivation and desires,” aimed at team leaders, coaches, and HR staff rather than at executives designing top-down policy. That perspective matters for how it gets used: Champfrogs is a conversation tool for diagnosing a specific team’s specific gap, not a universal ranking of which of the ten motivators matters most in general: a team missing Order (stability during a reorg) needs a different intervention than one missing Power (no real influence over roadmap decisions), even though both would show up generically as “low motivation” without the checklist.

Curiosity, Honor, and Power

Curiosity is the presence of enough open problems to investigate, Appelo’s own phrasing is “I have plenty of things to investigate and to think about”, and it depletes fastest on teams where the backlog has become a queue of well-specified tickets with no ambiguity left to explore. Honor is the sense that a person’s personal values show up in what the group actually does, described as “my personal values are reflected in the group,” which is a loyalty driver distinct from simple job satisfaction. Power is the room to influence outcomes, captured directly as “there’s enough room for me to influence what happens around me.”

These three cluster around a team’s actual decision rights rather than its compensation or its social climate, which is why they connect most directly back to the Autonomy leg of SAFe’s three-factor model. A team can score well on relatedness and still be low on Power specifically if every architectural or prioritization decision routes through a single senior voice regardless of how collaborative the day-to-day work feels, the checklist separates that specific gap from the generic complaint of “we don’t feel empowered.”

Freedom, Status, and Goal

Freedom is independence in how a person carries out their own work and responsibilities without needing to check in on every step, a narrower, more personal cousin of the team-level autonomy Hackman’s hierarchy describes. Status is a recognized standing relative to peers, distinct from Honor’s alignment-of-values framing; a senior engineer can feel their values are well represented on a team (high Honor) while still feeling their seniority carries no visible recognition (low Status). Goal is the presence of a clear purpose connecting the day’s work to something that matters, the team-level analog of the Purpose leg in SAFe’s Autonomy-Mastery-Purpose framework.

Teams under reorganization or acquisition pressure most often show a specific deficit here rather than a general one: Order and Status tend to drop first as reporting lines and role definitions become unstable, while Freedom and Goal often hold steady until the ambiguity drags on long enough to erode them too. Reading which of the ten motivators has actually moved, rather than assuming uniform demotivation, is what turns Champfrogs from a poster into a working diagnostic a Scrum Master or people leader can actually use in a one-on-one.

When Is the Right Fix a Compensation Change, and When Is It Something Else?

A raise request or a compensation complaint is not automatically a compensation problem, and treating it as one by default is its own version of the diagnosis Principle #8 asks a leader to skip past. Before adjusting the narrow band Cottmeyer describes, the more useful question is which of Champfrogs’ ten motivators the complaint actually maps to: a request framed as “I deserve more money” sometimes resolves once a leader recognizes it as a Status or Power gap instead, a signal that someone’s standing or influence isn’t matching their contribution rather than evidence their base pay is out of line. Genuine extrinsic needs, the mortgage or tuition bill Cottmeyer’s account concedes intrinsic-motivation theory can’t erase, still call for a real compensation answer. But running every dissatisfied conversation through the Champfrogs list first keeps a leader from treating money as a universal patch for problems that are actually about Order, Power, or Status, and from quietly rebuilding the flat, one-size-fits-all policy Principle #8 is designed to move a leader away from.

---

Summary

Principle #8 asks a leader to trade the certainty of a specified process for the harder, ongoing work of building conditions; because the heuristic work knowledge workers actually do degrades under exactly the incentive model that improves linear, well-defined work.

Diagnose Before Delegating

The mechanism running through every section above is the same one: naming what kind of task is in front of a person, well-defined and linear, or heuristic and exploratory, determines whether an extrinsic reward will help or hurt, and naming which of Hackman’s two authorities or Appelo’s ten motivators is actually missing determines what to delegate or address next. A leader who skips the diagnosis and defaults to a single company-wide policy, no bonuses anywhere, full autonomy everywhere, is running a new one-best-way model in place of Taylor’s old one, just with the values flipped. The Candle Problem and Stefan Falk’s brain-energy research already did the work of establishing why that narrowing happens; what’s left for a leader is the move it implies; remove the pressure or the boredom rather than trying to motivate a person past it, since neither responds to more effort applied in the same direction. Checking which condition is actually present, is this task heuristic, is this team missing Power or Order, is this compensation conversation about local optimization or about a genuine unmet extrinsic need, is the actual discipline Principle #8 asks for, in place of the discipline Taylor’s model asked for.

What Breaks When the Diagnosis Gets Skipped

Every failure mode described above traces back to collapsing a bounded, specific delegation into an unbounded one, or refusing to delegate at all. Crossing Hackman’s boundary in either direction breaks something specific: pushing past it produces the anarchy leaders fear, and stopping short of it, withholding even what the boundary permits, reproduces Taylor’s staff-line split under new terminology, with a leader who claims to have delegated but still owns every process decision no further from scientific management than the org chart makes it look. Compensation breaks the same way, and the fix is the same diagnostic move rather than a compensation-specific one: a spread wide enough to revive the pressure Cottmeyer’s narrow band exists to avoid calls for tightening the band, and a flat policy that’s actually starving someone’s mortgage or tuition bill calls for a real raise, not a Champfrogs conversation; mistaking one failure for the other is how a leader ends up overpaying to fix a Status problem or underpaying while calling it principled restraint. The throughline for a leader applying Principle #8 in practice is that every one of these tradeoffs is a specific, checkable diagnosis rather than a single switch to flip, and the ten-item Champfrogs list, the two-item Hackman hierarchy, and the extrinsic-versus-heuristic task distinction are the concrete tools for making that check instead of guessing at it.