Prompts

Element · All elements
Evidence · 34 claims (12 for, 20 mixed, 2 against) · 38 studies (17 causal, 8 review, 6 quant-synthesis, 3 theoretical, 2 associational, 2 qualitative), q1–q4 · 8 of 38 report an effect size · 28 claims rest on one study

Description

A prompt is a cue, question, or instruction embedded in a learning task that directs learner attention, elicits a response, or scaffolds a cognitive process at the moment it is needed. Prompts range from simple procedural reminders ("check your units") to elaborative questions ("why does this solution work?") to metacognitive self-questioning stems ("what strategy am I using, and is it working?").

Design Implications

Prompts work by triggering self-explanations and strategic processing that learners would not otherwise engage in spontaneously Self-explanation prompts improve learning by forcing learners to connect steps to underlying principles. [+S]. Their effectiveness depends on timing and specificity: prompts delivered at the point of decision, and targeted at the process rather than the task level, produce the largest gains Feedback is most effective at the task and process levels rather than the self level. [+S]. Prompts should be faded as competence grows, since persistent prompting can become a crutch that learners cannot function without Fading support promotes transfer of responsibility from scaffold to learner. [+M].

Context

Requirements

• A clear target process or response the prompt is designed to elicit
• Placement at the decision point where the prompted action is relevant, not batched at the end
• A mechanism for the learner to act on the prompt (write, answer, revise), not merely read it (Practice)
• Fading plan: prompts should diminish as the learner internalizes the strategy (Fading)

Constraints

• Over-prompting produces prompt dependence — learners follow cues mechanically without building self-regulated strategies Fading support promotes transfer of responsibility from scaffold to learner. [-M]
• Prompts add extraneous load when learners lack the knowledge to respond to them; novices facing complex material may be better served by worked examples than by questions Example-problem sequences reduce cognitive load for novices compared with unsupported problem solving. [~M]
• Generic prompts ("think carefully") have little effect; vague cues waste attention without directing processing
• Effectiveness reverses with expertise: highly knowledgeable learners find explicit prompts redundant and distracting Guidance such as prompts becomes less effective — and can reverse — as learner expertise increases. [~S]

Target Learners

• Novices who do not yet spontaneously use strategies like self-explanation or monitoring Self-explanation prompts improve learning by forcing learners to connect steps to underlying principles. [+M]
• Learners with weak metacognitive habits who benefit from externalized self-questioning
• Less beneficial for experts, for whom prompts impose redundancy Guidance such as prompts becomes less effective — and can reverse — as learner expertise increases. [~M]

Target Learning Goals

• Metacognitive strategy acquisition: monitoring, planning, and evaluating one's own work
• Conceptual integration: connecting procedures to principles through elaborative questioning
• Transfer: internalizing prompted questions so learners generate them unprompted in new contexts

Affordances

• Scaffolding — a prompt is scaffolding in its most granular form: temporary, targeted support at a single decision point, designed to be faded
• Cognitive Load Management — well-placed prompts offload strategic decisions ("what should I check next?") so working memory is spent on content rather than on deciding how to proceed
• Active Learning — prompts convert passive reading or watching into response by demanding an answer, prediction, or revision at the moment of encounter
• Clear Structure — procedural prompts make the expected sequence of work explicit, reducing aimless search

Claims

• The general-AI group most often dictated ChatGPT to directly generate solutions while the regulated-AI group more frequently collaborated to improve self-generated ideas [+W] — attached 2026-10-12 from Wong et al. (2026), which proposed "Seven sample prompts scaffolding collaborative ChatGPT interaction in creative problem-solving".
• Names alone elicit weaker, less consistent identity-marked feedback language than explicit attribute prompts [+W]
• Missing prompt-element types significantly co-occur: failure to provide appropriate functionality explanation correlates strongly with missing expected output/return and with extraneous elements [~W]
• Coded prompt logs show different prompting strategies foreground different pedagogical themes, with technical prompts the largest strategy total (398 instances) [~W]
• Collaborative prompt use mediated the regulated-AI group's advantage over the general-AI group in independent originality [+W]
• Prompting is a major source of AI-assisted cognitive effort: copying prompts reduces effort but not time, and verbose model responses can make AI-assisted completion slower than predicted [~W]
• PCA metacognitive suggestions improve consensus reaching, task division and reciprocal interaction during collaboration [+M] — attached 2026-10-18 from Hayashi (2020), which proposed "PCA with five types of metacognitive suggestion delivered during dyadic explanation"; tests this page's relationship.
• Cognitive and metacognitive prompts did not increase self-reported online strategy use, the quality of strategies in learners' notes, or learning outcomes in hypermedia learning [-W] — attached 2026-10-19 from Nadja Gentner et al. (2020), which proposed "Hypermedia environment on empirical research methods with six page-linked cognitive and metacognitive prompts"; tests this page's relationship.
• Typical-gain teacher's students received the most inappropriate directions and prompts for setting a purpose and were seldom reminded to return to the purpose with evidence [~W]
• Cognitive load showed no significant mediation of the effects of navigation features or self-explanation prompts on video learning outcomes [~W]
• When facing students' major conceptual challenges, the teacher shifted from prompting cued questions to providing explanations himself [-W]
• Anticipatory self-explanation prompts induced lower extraneous cognitive load than retrospective prompts [~W]
• Problematizing scaffolds timed to metacognitive moments led students to learn the most domain knowledge [+W]
• Closed-book self-explanation prompts reduce the number of generated self-explanations when learners are not prompted to actively process the instructional explanations beforehand [~W]
• Combining cognitive and metacognitive prompts optimizes germane processing; metacognitive prompts alone help only when remedial planning is prompted and remediation is possible [~W]
• The effectiveness of CSCL scripts for domain learning does not significantly depend on the number of different collaborative activities prompted [~W]
• Cognitive and metacognitive prompts in learning journals helped early in a term but became detrimental by its end; adaptive fading-out mitigated this [~W]
• Framing reminders as losses reduces, but does not eliminate or reverse, the reminder bias in both between- and within-participants designs [~W]
• Prompting metacognitive strategies before cognitive strategies improves strategy quality and conceptual knowledge [~W]
• Correlations between the number of prompt mistakes and number of messages sent were non-significant or weak, giving little evidence that mistake count influenced prompting effort [~W]
• Metacognitive prompts during collaboration improve communication of knowledge and its underlying assumptions [+W]
• Navigation features and self-explanation prompts do not interact in their effect on video learning [~W]
• Task accuracy is far higher with reminders than internal memory and is unaffected by reminder framing [+W]
• Explicit training in constraint-seeking questions showed only moderate, non-generalizing benefits, whereas prompting explanation supports better question asking [~W]
• Participants show a significant reminder bias (overuse of reminders relative to the optimal strategy) when reminders are framed as gains [~W]
• Prompted cue type and timing of cue prompt do not interact in their effect on self-assessment accuracy (Experiment 1) [~W]
• Fading-out prompts yield more cognitive strategy use and higher learning outcomes than permanent prompts [~W]
• Prompt compliance varied widely across the four metacognitive prompts, with interactive prompts responded to more appropriately [+W] — attached 2026-10-23 from Rebecca Pape (2026), which proposed "Brainix metacognitive prompt set (set goal, prioritize, reflect prompts and calendar goals planner)"; tests this page's relationship.

The evidence by question

• Does prompting learners to explain what they study or solve improve their understanding, and on which outcomes? Prompting learners to self-explain improves understanding and problem solving on immediate tests by a moderate average amount, with little evidence yet on delayed or classroom gains. [+S]
• Do planning, monitoring and evaluation prompts help, and which ones? Metacognitive prompts improve learning [+S]
• Should support be withdrawn, and in what order? Fading support promotes the transfer of responsibility from instructor to learner. [+S]

Related Elements

• Think-Aloud — prompts can elicit think-aloud behavior; the two together make reasoning visible and rehearsed
• Fading — the mechanism for withdrawing prompts as strategies become internalized
• Practice — prompts embedded in practice direct effort during retrieval and application
• Feedback — a prompt is a question posed before responding; feedback is information provided after
• Advance Organizers — a prompt delivered before the task rather than during it

Patterns That Use This Element

• Cognitive Apprenticeship — coaching and articulation phases rely on prompts to elicit reasoning
• Direct Instruction — guided practice uses prompts to steer student responses before independent work
• 4C/ID — supportive information includes cognitive prompts attached to learning tasks

Examples

Reciprocal Teaching — Students are prompted with four question stems (summarize, question, clarify, predict) that they gradually internalize; the prompt cards are deliberately faded over successive sessions.

Khan Academy — Hint sequences in exercises function as graduated prompts: each hint reveals one step, prompting the learner to complete the remainder rather than showing the full solution.

Self-explanation prompt studies — Chi et al.'s prompted self-explanation paradigm: learners studying worked examples are asked "what is this step doing?" at each line, producing large gains over studying examples silently. - Revise educational SLM system prompts using the Persona and Context Manager patterns plus cognitive-apprenticeship scaffolding guidelines - Indirect training: prompting metacognitive regulation in regular subject classes to promote transfer across contexts - Design CSCL environments to remind learners to think as a group and scaffold cognitive group awareness

Key Sources

• Chi, M. T. H., Bassok, M., Lewis, M. W., Reimann, P., & Glaser, R. (1989). Self-explanations: How students study and use examples in learning to solve problems. Cognitive Science, 13(2), 145–182. doi:10.1207/s15516709cog1302_1
• Palincsar, A. S., & Brown, A. L. (1984). Reciprocal teaching of comprehension-fostering and comprehension-monitoring activities. Cognition and Instruction, 1(2), 117–175. doi:10.1207/s1532690xci0102_1
• Hattie, J., & Timperley, H. (2007). The power of feedback. Review of Educational Research, 77(1), 81–112. doi:10.3102/003465430298487
• Kalyuga, S., Ayres, P., Chandler, P., & Sweller, J. (2003). The expertise reversal effect. Educational Psychologist, 38(1), 23–31. doi:10.1207/S15326985EP3801_4

Page metadata

Field	Value
type	element
id	prompts
title	Prompts
description	A prompt is a cue, question, or instruction embedded in a learning task that directs learner attention, elicits a response, or scaffolds a cognitive process at the moment it is needed.
status	review
canonical	True
generated	by claude/unspecified on 2026-08-29
trust	unverified — no one has confirmed this page's content

Frontmatter is the page's machine-readable record — see the Schema & Guide.