In 2012, Professor Andrew Hooyman filmed three instructional videos at the University of Nevada. Each taught the same bowling motion used in cricket. Each showed identical demonstrations. Each covered the same technical details about foot placement, arm position, and release.

The only difference was three words.

The first video said: “Step forward with your front foot.”

The second said: “You need to step forward with your front foot.”

And the third said: “You could try stepping forward with your front foot.”

Need versus could. Should versus might. Must versus may.

When college students learned from these videos, those who heard “you could try” outperformed those who heard “you need to” on every measure. They were more confident, they were more in control, and they experienced higher positive emotion during practice.

Most importantly, when they returned the next day for a retention test, they performed significantly better. In one of the groups, the learning stuck, in the other two, it didn’t.

This reveals a simple but all-too-often overlooked truth: our words matter– a lot.

 

The Invisible Effect

 

Hooyman set out to study autonomy, a core driver of intrinsic motivation—and what he found revealed the subtle but powerful effect of language on performance.

Participants were divided into three groups during practice. One group received sovereign-supportive language, hearing phrases like “you could,” “you might want to,” and “feel free to.” Another group was given controlling language—“you need to,” “you must,” and “you should.” A third group received the same instructions without any framing and served as a neutral baseline.

At first, the language seemed to make no difference at all. During practice, all three groups improved at nearly identical rates. Their immediate performance was virtually indistinguishable– hidden from view.

But after practice, participants were asked to rate how much choice they felt on a scale from 1 to 10. The autonomy-supported group reported a strong sense of agency, averaging 8.4. The controlling group averaged just 6.1—a gap of 2.3 points.

That difference showed up the next day.

When participants returned for a retention test, the sovereign-supportive group demonstrated far superior accuracy. The learning had consolidated differently.

When the scaffolding was removed, performance diverged.  The gap appeared not in what people could do while being guided, but in what they retained—and how well they performed—once that guidance was gone.

 

Under the Hood

 

The human nervous system treats sovereignty as a biological requirement, not a preference.

Rats will choose a self-administered shock over an identical shock they cannot control. Monkeys will work for the opportunity to make choices, even when all options lead to the same outcome. Exercising control activates reward circuitry in the brain.

When you receive controlling language, your brain detects a constraint. The instruction “you must do this” signals a loss of sovereignty, even if you were planning to do exactly that thing anyway. This triggers a mild stress response. Research shows controlling language increases cortisol, which down-regulates the brain’s reward circuitry.

Sovereign-supportive language does the opposite. “You could try this” preserves the sense that you are choosing the action, even when following clear instructions. This keeps reward systems active. It maintains a positive effect. It reduces friction.

The brain learns better when reward systems are engaged and stress responses are quiet because our attention remains fully on the task. When we introduce stress, our brain is divided between the task and the consequences of our performance.

 

The Compound Effect

 

Since the difference in controlling versus sovereign-supportive framing is largely invisible in the moment of learning, it goes unnoticed.

This creates a dangerous trap. Teachers, coaches, and managers see a response to their controlling language in the moment and assume it works. Immediate compliance looks like success.

However, the cost shows up later. When learners practice alone. When they face variation. When they need to recall the skill under pressure.

When controlled, we do not retain the lesson. When experiencing sovereignty, we do.

Think of it as compound interest for skill acquisition. The controlling approach might produce the same initial return, but sovereign-supportive language generates significantly better long-term returns.

Each practice session becomes slightly more effective. Each repetition sticks a bit better.

And that advantage is unmistakable across time.

 

So Try Saying It Differently

 

The beauty of this research is its specificity. Your processes can stay the same, while the articulation of your message changes.

While controlling language removes perceived choice:

“You need to…”
“You have to…”
“You should…”

Sovereign-supportive language preserves it:

“You could try…”
“You might want to…”
“You may find…”

This isn’t about being permissive.The sovereign-supportive group in Hooyman’s study wasn’t given a choice about whether to perform the skill. They couldn’t opt out of the bowling motion. They received the same clear technical guidance as everyone else.

The sovereign-supportive language simply acknowledged that the learner remained the agent of their own action.

And it turns out, this matters a great deal.

 

Neutral Is Not Enough

 

There’s a second finding worth exploring: neutral language was barely better than controlling language.

The neutral group—those who received the raw instructions without controlling or supportive frameworks—rated their perceived control at 6.2 out of 10. Not quite as low as the controlling group, but nowhere near 8.4. Their retention was also nowhere close to the sovereign-supported group.

This means, sovereign support wins out over control and neutrality.

This makes sense from a systems perspective. The default instructional mode is already somewhat controlling. When an expert tells a novice what to do, the power gap creates an implied constraint. The learner assumes they should comply exactly.

Neutral language doesn’t disrupt this assumption. It simply avoids making it worse.

Sovereign-supportive language, on the other hand, actively counters it. By explicitly suggesting rather than prescribing, it reframes the relationship. The instructor becomes a guide rather than an enforcer. The instruction becomes information rather than command.

 

The Broader Pattern

 

This study fits within a larger body of research on self-determination theory, which identifies sovereignty as one of three basic psychological needs (along with competence and relatedness). Across hundreds of studies, sovereign support consistently predicts better outcomes in education, health behavior, work performance, and skill acquisition.

The Hooyman study is notable because it isolates the mechanism so cleanly. By holding everything else constant and varying only the language, it demonstrates that perceived sovereignty—not actual choice—drives the effect.

We don’t let students design their own curriculum or let athletes choose their training program. We simply frame the existing program in a way that preserves their sense of agency.

Other research supports this. Studies show that giving learners control over incidental factors—like the color of the equipment or the order of practice—enhances learning, even though these choices provide no strategic advantage. The benefit comes from the act of choosing, not from making better choices.

Similarly, research on feedback timing shows that letting learners request feedback when they want it produces better learning than providing the same feedback at the same frequency but without learner control. The information is identical. The timing is identical. The only difference is who initiates it.

The pattern is consistent: sovereign support enhances outcome.

 

What This Means for Design

 

If you’re designing a learning system—whether it’s a training program, an educational platform, a coaching method, or a personal practice routine—language choice matters.

The instruction “complete three sets of ten repetitions” becomes “you could aim for three sets of ten repetitions.”

The feedback “you need to shift your weight earlier” becomes “you might try shifting your weight earlier.”

The requirement “review this material daily” becomes “you may find daily review helpful.”

This isn’t semantic game-playing. It’s a mechanical adjustment.

When you preserve perceived sovereignty, you reduce the regulatory burden of managing controlling conditions. You keep reward systems engaged. You maintain a positive affect. You enhance memory consolidation. You build stronger, more durable learning.

The cost is zero. The implementation is immediate. The effect compounds over time.

Most importantly, this approach doesn’t require motivation. It doesn’t ask learners to “stay motivated” or “push through.” It simply reduces friction in the learning process by removing an unnecessary constraint.

The system works better because it’s designed better.

The action remains the same. The commitment remains the same. The standard remains the same.

Only the framing changes.

And that changes everything else.