Psychological Knowledge of Results

Abstract

Knowledge of results is augmented feedback about the outcome of a performance — how far a movement missed its target, whether an answer was correct — supplied after the fact to guide learning, which MeSH classifies under reinforcement. For most of a century it was treated as the near-automatic strengthener of correct responses, following Thorndike's law of effect. Modern motor-learning research overturned that view: the feedback that most improves performance during practice can impair long-term retention, because the learner comes to lean on it. The field now separates guiding performance from promoting learning, and studies how the frequency, timing, and control of feedback shape a durable skill. This article surveys knowledge of results, the guidance hypothesis, feedback intervention theory, and the motivational accounts, with three interactive demonstrations.

Keywords: knowledge of results, augmented feedback, guidance hypothesis, motor learning, feedback intervention theory

When a learner throws at a hidden target and is told only “eight centimeters short,” that single piece of information — knowledge of results — is often enough to drive rapid improvement. The finding that performance tracks the feedback so closely made knowledge of results one of the most studied variables in the psychology of skill, and it seemed to confirm a simple idea: tell people how they did, and they get better (Salmoni, Schmidt, & Walter, 1984). The century of research since has complicated that idea in an instructive way. Feedback that produces the fastest gains while it is present frequently produces the poorest performance once it is withdrawn, which is the situation that actually matters — the game, the recital, the surgery (Winstein & Schmidt, 1990). Knowledge of results guides, but guiding is not the same as teaching.

What Knowledge of Results Is

Knowledge of results is post-response feedback, from a source outside the performer, about the outcome of a movement or decision relative to its goal. It is one kind of augmented feedback — information added to what the performer can already sense directly — and is distinguished from knowledge of performance, which describes the movement pattern itself rather than its result. A coach who says “the putt stopped a foot short” supplies knowledge of results; “the wrist broke too early” supplies knowledge of performance (Salmoni, Schmidt, & Walter, 1984). Both are contrasted with the intrinsic feedback — vision, proprioception, the sound of the strike — that arises from the action itself.

MeSH files the descriptor under reinforcement, a placement that reflects the field's original assumption: that telling a learner the result of a correct response reinforces it, in the same sense a reward does. That assumption, inherited from Thorndike, turned out to be only part of the story, and the gap between knowledge of results as a reinforcer and knowledge of results as information is the thread that organizes the modern literature (Kluger & DeNisi, 1996).

Feedback and the Law of Effect

The experimental study of knowledge of results begins with a direct test of Thorndike's law of effect, which held that responses followed by a satisfying state of affairs are stamped in. If feedback works by reinforcement, then a response labeled “right” should be strengthened automatically, without the learner needing to reason about it. Trowbridge and Cason gave participants a line-drawing task and varied what they were told after each attempt, finding that specific quantitative feedback produced steady improvement while a mere “wrong” or nonsense syllable produced almost none (Trowbridge & Cason, 1932). The result looked like reinforcement but already hinted at something more: what mattered was the information content of the feedback, not its evaluative sign.

By mid-century the informational reading had largely displaced the pure reinforcement account. Bilodeau and Bilodeau showed that learning depended on the number of trials that carried knowledge of results, not on the number of trials as such, and that withdrawing feedback halted improvement almost immediately — feedback was functioning as a guide the learner used trial to trial, not as a reward accumulating in the background (Bilodeau & Bilodeau, 1958). Adams built this into a formal account, his closed-loop theory, in which knowledge of results is the error signal a learner compares against a stored perceptual trace of the correct movement, gradually reducing the discrepancy (Adams, 1971). Schmidt's schema theory then generalized the idea, casting feedback as the material from which learners abstract a rule relating movement parameters to outcomes rather than a single trace (Schmidt, 1975).

The Guidance Hypothesis

The turning point came from a distinction that early work had blurred: the difference between performance during practice and learning, which can only be measured by a later retention or transfer test with feedback removed. When researchers began routinely running such tests, a paradox appeared. Conditions that produced the best performance while knowledge of results was available — feedback after every trial, immediately, in full detail — often produced the worst performance on a delayed retention test, while conditions that looked inferior during practice produced more durable learning (Salmoni, Schmidt, & Walter, 1984).

The guidance hypothesis explains this. Frequent, immediate knowledge of results guides the learner powerfully toward the target on each trial, but that very guidance discourages the learner from developing their own error-detection and problem-solving, and encourages them to depend on the feedback. When the feedback disappears, the guided learner has nothing to fall back on. Winstein and Schmidt demonstrated the effect directly: a group given feedback on only half their trials, faded across practice, performed as well as a 100% group during acquisition and better on a no-feedback retention test (Winstein & Schmidt, 1990). Reducing the relative frequency of knowledge of results, delaying it, or summarizing several trials at once all tend to help retention for the same reason: they force the learner to process their own intrinsic feedback.

When Feedback Backfires

If knowledge of results were simply information that a rational learner uses, more of it could never hurt outright — it might be redundant, but not harmful. Yet feedback sometimes makes performance worse in an absolute sense. Reviewing several hundred effect sizes, Kluger and DeNisi found that although feedback interventions raised performance on average, they lowered it in roughly a third of cases, and that the average effect concealed enormous variability (Kluger & DeNisi, 1996).

Their feedback intervention theory located the difference in where feedback directs the learner's attention. Feedback that focuses attention on the task and how to do it tends to help; feedback that draws attention up to the level of the self — praise, threats to self-esteem, comparison with others — tends to consume cognitive resources on self-evaluation and to hurt. This reframing carried directly into education, where Hattie and Timperley organized effective feedback around three questions — where am I going, how am I doing, where to next — operating at the levels of the task, the process, and self-regulation, and warned that feedback aimed at the self as a person is the least effective of all (Hattie & Timperley, 2007). Shute's synthesis for instructional design reached the same conclusion, emphasizing specific, timely, task-focused feedback over evaluative judgments of the learner (Shute, 2008).

Self-Controlled and Motivational Feedback

The most active contemporary line treats knowledge of results as acting through motivation and attention, not information alone. A robust finding is that learners who control their own feedback — requesting it when they judge they need it — learn better than matched learners given the identical feedback on a schedule they did not choose. Chiviacowsky and Wulf showed that self-controlled learners tend to ask for feedback after trials they believe went well, using it to confirm and consolidate rather than to correct, and that the autonomy itself contributes to the benefit (Chiviacowsky & Wulf, 2002).

Wulf and Lewthwaite drew these effects together in the OPTIMAL theory of motor learning, which holds that feedback enhances learning partly by raising expectancies for success, supporting autonomy, and directing attention externally — motivational and attentional channels that operate alongside its informational role (Wulf & Lewthwaite, 2016). On this view, feedback framed to signal that a trial was good (rather than merely reporting error) strengthens learning through the learner's expectancies, which is why comparative feedback suggesting improvement can help even when the objective information is held constant (Chiviacowsky, 2016). The informational account and the motivational account are not rivals so much as complementary descriptions of how the same feedback exerts its effect.

Worked Example

A simple error-correction model makes the guidance paradox concrete. Let a learner's absolute error on trial n be E_n, and suppose that on any trial carrying knowledge of results the learner removes a fixed fraction k of the current error: E_n+1 = E_n(1 − k). Take a starting error of E₀ = 8.0 cm and k = 0.25. With feedback on every trial the error falls geometrically: 8.0, then 6.0, 4.5, 3.375, 2.53, 1.90 — a smooth acquisition curve, E_n = 8.0 × 0.75ⁿ, that approaches zero the more feedback is given.

Read literally, the model says knowledge of results is pure guidance and more is always better: every feedback trial multiplies the error by 0.75, so 100% feedback minimizes error at every point. That is exactly the prediction the guidance hypothesis refutes. The model captures performance while feedback is present and is silent about what the learner retains once it is gone, because it contains no term for the learner's own error-detection — the capacity that frequent guidance suppresses and that the retention test exposes. Winstein and Schmidt's faded 50% group, worse than this curve predicts during acquisition yet better on retention, is the evidence that the quantity the model omits is the one that matters (Winstein & Schmidt, 1990). The arithmetic that fits the practice curve is the same arithmetic that misses the point of practice.

Current Directions

Two questions organize current work. The first is whether the guidance effect and the motivational effects can be reconciled quantitatively, since reducing feedback frequency (guidance) and framing feedback positively (motivation) can pull in opposite directions on the same trial. Comparative and temporally framed feedback that signals progress improves learning even at frequencies the guidance hypothesis would call excessive, suggesting the two mechanisms operate on partly separate channels (Chiviacowsky, 2016; Wulf & Lewthwaite, 2016).

The second is how well the classic effects replicate and generalize. A meta-analytic revisiting of educational feedback confirmed that feedback's average benefit is real but highly variable and moderated by type, task, and level, echoing Kluger and DeNisi's central caution a quarter-century on (Wisniewski, Zierer, & Hattie, 2020). Systematic reviews of augmented feedback in motor rehabilitation likewise find consistent short-term performance gains but mixed evidence on lasting retention and transfer, leaving the guidance hypothesis's core claim — that the conditions optimizing practice are not those optimizing learning — as the field's organizing problem (Moinuddin, Goel, & Sethi, 2021).

Key Researchers

Edward L. Thorndike (1874-1949). Teachers College, Columbia University; his law of effect framed feedback about correct responses as a reinforcer that stamps in behavior, the assumption that shaped the first half-century of knowledge-of-results research. Wikipedia

Jack A. Adams (1922-2010). University of Illinois at Urbana-Champaign; his closed-loop theory of motor learning cast knowledge of results as the error signal a learner compares against a perceptual trace, giving feedback a precise role in skill acquisition.

Richard A. Schmidt (1941-2015). University of California, Los Angeles; his schema theory reframed feedback as the basis for generalized motor programs, and his guidance-hypothesis work showed that too-frequent knowledge of results can impair long-term retention.

Avraham N. Kluger. Hebrew University of Jerusalem; with Angelo DeNisi he built feedback intervention theory on a meta-analysis showing that a third of feedback interventions reduce performance, depending on the attentional locus they engage. ORCID

Gabriele Wulf. University of Nevada, Las Vegas; she established self-controlled and external-focus feedback effects and, with Rebecca Lewthwaite, the OPTIMAL theory linking feedback's motivational consequences to motor learning. ORCID

Suzete Chiviacowsky. Universidade Federal de Pelotas; she demonstrated that learner-controlled and comparative feedback schedules enhance motor learning, extending knowledge-of-results research into self-regulated and motivational practice. ORCID

John Hattie. University of Melbourne; his feedback model and meta-analytic syntheses translated knowledge of results for education, distinguishing feedback about the task, the process, and self-regulation from the far weaker feedback aimed at the self. Wikidata

Discussion

Knowledge of results has kept its central place in the study of skill because it is where a plausible equation — feedback strengthens correct responses, so more feedback means faster learning — breaks down under a single methodological correction. Once learning is measured by retention and transfer with feedback withdrawn, rather than by performance during practice, the conditions that maximize the two come apart, and the guidance hypothesis names why: frequent knowledge of results guides performance while suppressing the learner's own error-detection (Salmoni, Schmidt, & Walter, 1984; Winstein & Schmidt, 1990). The finding that feedback can lower performance outright, moderated by the attentional level it engages, closed off any purely informational account and forced the field to treat feedback as an intervention with motivational and self-related effects, not a neutral data stream (Kluger & DeNisi, 1996; Hattie & Timperley, 2007).

The payoff reaches wherever skills are taught. Coaching, physical rehabilitation, instructional design, and surgical training all confront the same trade-off between feedback that helps a learner get it right today and feedback that leaves them able to do it alone tomorrow, and the modern accounts — guidance, feedback intervention theory, and the motivational OPTIMAL framework — give practitioners a principled way to schedule, frame, and hand over control of knowledge of results (Wulf & Lewthwaite, 2016). The open problem the field still works is the one the retention test first revealed: how to give feedback that guides without creating the dependence that guidance breeds (Moinuddin, Goel, & Sethi, 2021).

Glossary

Augmented feedback.

Performance information added by an external source to what the performer can already sense; knowledge of results is one kind.

Bandwidth feedback.

Knowledge of results given only when error exceeds a set tolerance, so feedback frequency falls automatically as skill improves.

Closed-loop theory.

Adams's account in which knowledge of results is an error signal compared against a stored perceptual trace of the correct movement.

Faded feedback.

A schedule in which the relative frequency of knowledge of results is progressively reduced across practice to promote retention.

Feedback intervention theory.

Kluger and DeNisi's account holding that feedback's effect depends on whether it directs attention to the task or to the self.

Guidance hypothesis.

The proposal that frequent knowledge of results guides practice performance but harms retention by fostering dependence on the feedback.

Intrinsic feedback.

Sensory information arising from the action itself — vision, proprioception, sound — as opposed to augmented feedback.

Knowledge of performance.

Augmented feedback about the movement pattern that produced an outcome, contrasted with knowledge of results about the outcome itself.

Knowledge of results.

Post-response feedback from an external source about the outcome of a performance relative to its goal; the subject of this article.

Law of effect.

Thorndike's principle that responses followed by satisfying consequences are strengthened; the original rationale for treating feedback as a reinforcer.

Motor learning.

The relatively permanent improvement in skilled movement with practice, measured by retention and transfer rather than practice performance.

OPTIMAL theory.

Wulf and Lewthwaite's framework in which feedback aids learning through enhanced expectancies, autonomy support, and an external attentional focus.

Reinforcement.

The strengthening of behavior by its consequences; the MeSH category under which knowledge of results is classified.

Relative frequency of knowledge of results.

The proportion of practice trials on which feedback is given; lowering it often aids retention despite slowing acquisition.

Retention test.

A delayed test of performance with feedback withdrawn, used to measure learning as distinct from practice performance.

Schema theory.

Schmidt's account in which feedback lets a learner abstract a rule relating movement parameters to outcomes rather than a single trace.

Self-controlled feedback.

A schedule in which the learner decides when to receive knowledge of results, reliably improving learning over matched imposed schedules.

Transfer.

The application of a learned skill to a novel task or context, a stringent test of whether feedback produced durable learning.

Frequently Asked Questions

What is knowledge of results in psychology?
Knowledge of results is augmented feedback, provided by an external source after a response, about the outcome of a performance relative to its goal — such as how far a throw missed the target (Salmoni, Schmidt, & Walter, 1984).

How does knowledge of results differ from knowledge of performance?
Knowledge of results describes the outcome (the putt stopped short); knowledge of performance describes the movement that produced it (the wrist broke early). Both are augmented feedback, distinct from the intrinsic feedback the performer senses directly (Salmoni, Schmidt, & Walter, 1984).

Does more feedback always improve learning?
No. The guidance hypothesis shows that frequent, immediate knowledge of results speeds performance during practice but can impair long-term retention by fostering dependence on the feedback (Winstein & Schmidt, 1990).

Can feedback make performance worse?
Yes. A meta-analysis found feedback interventions reduced performance in about a third of cases, particularly when they directed attention to the self rather than the task (Kluger & DeNisi, 1996).

Why does reducing feedback frequency help retention?
Withholding feedback on some trials forces learners to process their own intrinsic feedback and develop error-detection, so they perform better when feedback is later unavailable (Winstein & Schmidt, 1990).

What makes feedback effective in education?
Feedback works best when it is specific and task-focused and answers where the learner is going, how they are doing, and what to do next; feedback aimed at the self as a person is the least effective (Hattie & Timperley, 2007).

What is self-controlled feedback?
It is a schedule in which learners request knowledge of results when they choose; learners who control their own feedback reliably learn better than those given the identical feedback on an imposed schedule (Chiviacowsky & Wulf, 2002).

How does OPTIMAL theory explain feedback?
It holds that feedback aids learning not only by conveying information but by raising expectancies for success, supporting the learner's autonomy, and promoting an external focus of attention (Wulf & Lewthwaite, 2016).

References