新颖且有用

#1. Executive Summary

Creativity is not a mystical gift but a cognitive, dispositional, and environmental process whose product is defined by two joint criteria: novelty and value (usefulness, appropriateness, fit). This chapter’s central thesis is that most popular beliefs about creativity are half-true in a way that misleads working creators. The evidence supports a decidedly unromantic picture: creativity is largely ordinary cognition applied under the right conditions, it depends heavily on domain expertise as raw material, it is powered by intrinsic motivation, and — for a working career — it obeys something close to a probabilistic quantity-quality law in which sustained output, not rare inspiration, is the strongest lever an individual controls.

Three conclusions matter most for a working writer. First, divergent thinking is a component of creativity, not its definition — divergent-thinking tests predict real-world creative achievement only weakly, and the field has a genuine measurement problem it has never fully solved. Second, the celebrated phenomena — insight as lightning, incubation as magic, constraints as enemies, brainstorming as the engine of group creativity — each partly survive contact with the research but only in nuanced, qualified forms; the strong popular versions are false. Third, the most reliable multi-year strategy is a divergence-convergence discipline: generate abundantly, disengage deliberately, evaluate ruthlessly with the humility that your self-evaluation of novel work is unreliable, and run a portfolio of cheap bets because you cannot predict in advance which work will land.

On AI: the current evidence (Doshi & Hauser 2024; Girotra/Meincke et al. 2023) shows generative AI reliably raises the floor of individual output and idea fluency while lowering collective diversity — homogenizing what groups of writers produce. This shifts the bottleneck of creative work from production toward evaluation, taste, and distinctiveness, which is precisely where a serious writer should now concentrate.

#2. Why This Topic Matters

Creativity is the cognitive capacity that lets a mind produce something both new and useful — the engine behind every scientific theory, every technology, every novel and every solved problem that could not be reached by rote. For the individual thinker it is the highest-order integration of everything else in this curriculum: it consumes the fuel of curiosity, runs on the substrate of domain knowledge built by deliberate practice, is corrupted by cognitive biases in self-evaluation, requires signal-versus-noise filtering to separate good ideas from the many bad ones, and rewards an antifragile portfolio posture toward uncertain bets.

Getting the science right has real stakes. If you believe creativity is a fixed gift, you will not practice it. If you believe divergent thinking is creativity, you will train the wrong skill. If you believe you must wait for the muse, you will produce little. If you believe constraints are your enemy, you will squander their generative power. And in an era where machines generate fluent prose on demand, misunderstanding where human creative value actually lives — in judgment, distinctiveness, and revision rather than raw generation — is a career-level error. This chapter aims to replace a set of comfortable myths with a working model accurate enough to change how you sit down to write.

#3. Foundations

#Core definition

The modern consensus, traceable through Amabile, Runco, and Sternberg, defines creativity as the production of work that is both novel (original, surprising) and valuable (useful, appropriate, or fit for its context). Both criteria are necessary: novelty without value is mere randomness; value without novelty is competent reproduction. Crucially, value is judged by observers within a domain — a point that becomes the foundation of both Amabile’s measurement method and Csikszentmihalyi’s systems view.

#The historical arc

• Galton (1869) launched the hereditary study of genius, framing eminence as an individual-differences problem.
• Graham Wallas (1926), in The Art of Thought, gave the first influential stage model: preparation → incubation → illumination → verification (a later reading by Sadler-Smith, 2015, restores a fifth “intimation” stage Wallas actually described). This model still anchors popular thinking a century later — and, as critics note, it inadvertently glorifies the moment of conception while dismissing the “drudgery” of realization.
• J. P. Guilford’s 1950 APA presidential address is the field’s founding event as an experimental science. Guilford lamented psychology’s neglect of creativity and proposed divergent production (generating many varied responses) as its measurable core, distinct from the convergent thinking measured by IQ tests.
• E. Paul Torrance operationalized this into the Torrance Tests of Creative Thinking (TTCT, 1966), scoring fluency, flexibility, originality, and elaboration — still the most-used divergent-thinking instrument.
• Teresa Amabile (1982–1996) shifted the field toward the social psychology of creativity, introducing the Consensual Assessment Technique (CAT) and the intrinsic-motivation principle.
• Dean Keith Simonton built the historiometric tradition, quantifying creative careers across history and formulating the equal-odds rule and chance-configuration (blind-variation-and-selective-retention, BVSR) theory.
• Mihaly Csikszentmihalyi (1988–1999) proposed the systems model (individual–domain–field), relocating creativity from inside the head to the interaction of a person, a body of knowledge, and a community of gatekeepers.
• Finke, Ward & Smith (1992) launched the creative cognition approach with the Geneplore model.
• The current era adds cognitive neuroscience (Beaty and colleagues on brain-network dynamics) and, since 2022, the study of generative AI as both a tool and a theoretical provocation about what “novel and valuable” means when machines generate fluently.

#Key frameworks to hold in mind

• The Four P’s (Rhodes, 1961): creativity can be studied as Person (traits), Process (cognition), Product (the output judged novel+valuable), and Press (the environment).
• The Four-C model (Kaufman & Beghetto, 2009, Review of General Psychology): creativity exists at four levels — mini-c (personally meaningful insight in learning), little-c (everyday creativity), Pro-c (professional-expert creativity), and Big-C (eminent, domain-changing creativity). A working novelist lives in the Pro-c band and is reaching toward Big-C.

#4. Current Scientific Understanding

#The measurement problem (well-established, and foundational)

The single most important thing an educated person should know about creativity research is that its core instruments imperfectly capture what they claim to measure. Divergent-thinking tests (TTCT and “alternate uses” tasks) have good reliability but weak long-term predictive validity for real-world creative achievement. As John Baer has argued for decades, and as Sawyer (2012) concluded, the domain-general “creativity” these tests assume may be much smaller than the domain-specific component — meaning no general test could ever succeed well. Keep this caveat live throughout: much creativity research may be partly a science of what is easy to measure.

The main alternative, Amabile’s Consensual Assessment Technique (CAT) — have appropriate domain experts independently rate products for creativity — is considered the “gold standard” and achieves high inter-rater reliability (typically coefficient alpha in the 0.80–0.90 range). But it measures products against a domain’s current consensus, is labor-intensive, and a 2021 study (Barth et al., Journal of Creative Behavior) found that while mean product ratings are highly stable, individual raters drift over weeks — a subtle reminder that “value” is a moving social judgment.

#Divergent vs. convergent thinking (well-established that both are needed)

Real creativity requires cycling between divergent generation and convergent evaluation/selection, not divergence alone. Guilford’s original distinction has held, but the popular equation “divergent thinking = creativity” has not.

#Insight and the “Aha!” (contested; popular version falsified)

The Gestalt idea of insight as sudden restructuring is real as a phenomenology, but the most famous demonstration — the nine-dot problem as proof of “thinking outside the box” — does not support the folklore. Burnham & Davis and Alba & Weisberg (1981) showed that explicitly telling people the solution requires going outside the square barely improved solution rates (by about five percentage points, within sampling error). Kershaw & Ohlsson (2004) demonstrated that solving it depends on trainable perceptual/expertise factors, and that even with training only around half of subjects solved it, and not in a smooth flash. Insight is more the product of expertise than of a mystical leap.

#Incubation (real but modest; mechanism unresolved)

Sio & Ormerod’s (2009) meta-analytic review in Psychological Bulletin (135(1):94–120) synthesized 117 studies and confirmed a genuine but low-to-medium incubation effect (mean d = 0.29), with divergent-thinking tasks benefiting more, and — importantly — the benefit being greater when the break is filled with an undemanding interpolated task than with rest or a demanding task. But the mechanism remains debated among (a) unconscious spreading activation, (b) forgetting of misleading fixation, and (c) opportunistic assimilation of environmental cues. Sio & Ormerod concluded both conscious and unconscious processes are likely involved. So “incubation as magic” is overstated; “incubation as a real, deliberate, low-load disengagement” is defensible.

#The expertise-creativity relationship (well-established with a strong caveat)

Big-C creativity almost always rests on deep domain knowledge — the “10-year rule” (Hayes; Ericsson). Kaufman & Kaufman (2007), analyzing 215 modern fiction writers, found they took an average of 10.6 years between first and best publication. But the rule is a near-necessary condition, not a sufficient one, and its deliberate-practice version is contested for creative domains: unlike chess or music, creative fields give sparse, ambiguous, delayed feedback, so the tidy deliberate-practice story (studied in the Expertise chapter) transfers poorly. Expertise is the substrate creativity recombines; it does not guarantee the recombination.

#Intrinsic motivation (well-supported; qualified by later work)

Amabile’s intrinsic-motivation principle — people are most creative when driven by interest, enjoyment, and challenge rather than by controlling extrinsic pressures — is supported across many studies. But it has been revised: Amabile’s own later “motivational synergy” view holds that informational or enabling extrinsic motivators (recognition that confirms competence, resources that unlock work) can add to creativity, while controlling ones (surveillance, contingent reward, evaluation pressure) undermine it. The reward-and-creativity debate (Eisenberger vs. Amabile/Hennessey) remains genuinely unsettled at the margins.

#Group creativity (well-established: classic brainstorming underperforms)

Osborn’s face-to-face brainstorming reliably underperforms the same number of people working alone. Mullen, Johnson & Salas’s (1991, Basic and Applied Social Psychology 12(1):3–23) meta-analysis of 20 studies found large effects favoring nominal groups (individuals ideating separately, pooled) over interacting groups on both the number of non-redundant ideas (r = .57) and idea quality (r = .56); they concluded the “productivity loss of brainstorming groups is highly significant and of strong magnitude,” worsening as group size grows. The main culprit is production blocking (waiting your turn), with evaluation apprehension and free-riding as secondary factors. Brainwriting and nominal-group techniques largely fix this.

#The person and the trait evidence (well-established)

Among the Big Five, Openness to Experience is the most consistent and robust personality predictor of creativity across measures, tasks, and ages. Karwowski & Lebuda’s (2016, Psychology of Aesthetics, Creativity, and the Arts 10(2):214–232) meta-analysis found Openness the strongest predictor of creative self-beliefs and behaviors (r = .467; 95% CI .402–.531), followed by Extraversion (r = .264; 95% CI .225–.304). Openness is linked to the dopaminergic exploration system.

#5. Interdisciplinary Perspectives

The four disciplines this chapter integrates each answer a different question, and their friction is instructive.

The deepest tension is between the cognitive lab’s process view (creativity as ordinary cognition inside one head) and Csikszentmihalyi’s systems view (creativity is not “in” the person at all — it is the event where an individual’s variation is selected by a field for inclusion in a domain). The AI evidence sharpens this: when GPT-4 can generate ideas rated higher on average than students’, the differential tradition’s “who is creative” and the cognitive tradition’s “how is novelty generated” both lose force, and the systems question — who or what gets selected by the field as valuable — becomes the one that matters. The bottleneck moves from generation to judgment, which is a social-systems variable, not a cognitive one.

#6. Mental Models

1. Novelty × Value as defining axes. Plot any idea on two axes. High-novelty/low-value = noise; low-novelty/high-value = cliché; high-novelty/high-value = creative. Use it as a diagnostic on your own drafts. Fails when “value” is genuinely unknowable in advance (frontier work).
2. The Four P’s. When creativity stalls, ask which P is the bottleneck: Person (skill/trait), Process (method), Product (the work itself), or Press (environment). Most writers over-attribute blocks to Person when the real problem is Press or Process.
3. Divergence-convergence cycling. Separate idea generation from idea evaluation in time — never do both at once. This is the operational heart of the whole chapter.
4. The Geneplore generate-and-explore loop (Finke, Ward & Smith, 1992). Creativity alternates a generative phase (building rough “preinventive structures”) and an exploratory phase (interpreting, testing, refining them). It maps cleanly onto drafting (generate) and revising (explore).
5. Quantity begets quality / the equal-odds rule (Simonton). Because you cannot predict which work will be a hit, and hit rate is roughly constant, more attempts = more hits. Produce more. Fails as a naive law — Openness moderates it (Forthmann et al.), and it describes aggregates, not guarantees.
6. Incubation as deliberate disengagement. Schedule breaks with low-cognitive-load activity (walking) after real preparation. Not magic; a modest, real edge (d = 0.29).
7. Constraints as creativity. A U-shaped relationship: moderate constraints focus and boost creativity; too many or too few hurt. Self-imposed constraints (“eight-sentence story,” “no adverbs”) are a tool, not a cage.
8. The portfolio/barbell model. Run many small, cheap creative bets with uncapped upside (from the Antifragility chapter); protect against the downside of any single project failing. Creative careers are power-law-shaped: a few works carry most of the value.
9. Sturgeon’s Law on your own drafts. “Ninety percent of everything is crap” — including your first drafts. The job of convergence is to find and rescue the 10%. This is signal-versus-noise applied inward.

#7. Common Misconceptions

• “Creativity is a rare gift.” No — it is a broadly distributed, trainable capacity (the Four-C model shows a developmental continuum). Intelligent people fall for this because eminent Big-C examples dominate attention (an availability bias).
• “Divergent thinking = creativity.” Divergent thinking is one weakly predictive component; convergent selection matters just as much.
• “Constraints kill creativity.” The evidence shows moderate constraints usually help (Rosso; Stokes; the constraint-creativity U-curve).
• “Brainstorming is how groups create.” Classic verbal brainstorming underperforms individuals working alone (Mullen et al., 1991).
• “Inspiration strikes before the work.” Usually it strikes during and after the work. Action precedes motivation.
• “You must wait for the muse.” Prolific professionals write on schedule; waiting for inspiration is negatively related to output.
• “Creativity and discipline are opposites.” They are complements — the equal-odds rule and the 10-year rule both make discipline the precondition of creative luck.
• “AI will replace creative work.” The evidence shows AI raises the floor and homogenizes; it displaces routine generation while raising the premium on human judgment and distinctiveness.
• “Creative people are disorganized and chaotic.” Openness predicts creativity, but Conscientiousness predicts scientific creative achievement, and prolific careers require systematic habits.

#8. Real-World Applications

The research points in consistent directions for a working writer. Separate generation from evaluation in time — because doing both at once triggers the fixation and self-censorship that kill early ideas, and because your judgment of your own novel work is unreliable (a confirmation-bias problem from the Cognitive Biases chapter). Build domain knowledge relentlessly, because creativity recombines what you already deeply know — wide reading in and around your genre is not procrastination, it is loading the combinatorial store. Use constraints deliberately to escape the blank page. Protect intrinsic motivation by insulating the drafting phase from market/evaluation pressure (“write with the door closed”). Favor quantity across a portfolio, because you cannot pick your winners in advance. Use incubation strategically — walk away from a stuck problem after genuine preparation, filling the break with low-load activity. And position AI as a floor-raiser and sparring partner, not an author, precisely because its collective homogenization effect means that the writer who leans on it hardest becomes the least distinctive.

#9. Case Studies

(a) The nine-dot problem and the limits of “insight.” Popularized by Guilford in creativity studies, the puzzle became the metaphor for “thinking outside the box.” But Alba & Weisberg (1981) and Burnham & Davis found that telling people the solution lies outside the square raised success only marginally (~5 points, within sampling error). Kershaw & Ohlsson (2004) showed solving it depends on trainable perceptual expertise. Lesson: insight folklore oversells a mystical leap; expertise and training do the real work. Archival caveat: this is a narrow lab paradigm, and generalizing from it to real creative breakthroughs is exactly the ecological-validity worry the field acknowledges.

(b) Amabile’s rewards experiment. In “Social Influences on Creativity: The Effects of Contracted-For Reward,” Amabile found that when creative work (e.g., collage-making, storytelling) was done for a contracted reward, participants produced work rated less creative by expert judges — reward made the process “high-pressured and businesslike” rather than playful and risk-taking. Lesson: controlling extrinsic incentives can degrade creative quality even when they raise effort. Caveat: later meta-analytic work (Eisenberger) contests the generality, and Amabile’s revised “synergy” view concedes enabling rewards can help.

(c) The documented failure of brainstorming — and its fix. Mullen, Johnson & Salas (1991) meta-analyzed 20 studies and found nominal groups beat interacting brainstorming groups on both quantity (r = .57) and quality (r = .56). Girotra, Terwiesch & Ulrich (2010) then showed a hybrid process (individuals generate alone, then combine) beats pure team brainstorming, and — notably — that “building on each other’s ideas” was counter-productive. Lesson: for group ideation, use brainwriting/nominal technique, not a free-for-all.

(d) Simonton’s equal-odds rule in creative careers. Across historiometric studies, Simonton found the number of hit works is roughly a linear function of total output, that hit-rate is approximately constant within and across careers, and that a creator’s single best work tends to fall in their period of peak productivity — but which work will be the masterpiece is unpredictable even to the creator (“creators must lack foresight regarding the sociocultural merits of their ideas”). Follow-up work (Forthmann et al.) qualifies this: Openness moderates the quantity-quality relation, so it is “not quite equal odds.” Lesson: produce a lot; you are playing a numbers game you cannot pre-solve.

(e) The hot-streak structure of careers (Liu, Wang, Sinatra et al.). Liu, Wang, Sinatra, Giles, Song & Wang (2018, Nature 559(7714):396–399), drawing on scientists’ most-cited papers, auction prices for artists, and IMDb ratings for film directors, found that high-impact works cluster in “hot streaks”: “hit works within a career show a high degree of temporal regularity, with each career being characterized by bursts of high-impact works occurring in sequence.” The hot streak is temporally localized, emerges at a random point in a career, and — strikingly — comes with no increase in productivity. The 2021 follow-up (Liu et al., Nature Communications 12:5392) found the onset is preceded by a specific behavioral sequence: exploration (diverse styles/topics) followed by exploitation (focus) — the only combination that lifts the probability of a streak (roughly a 20.5%, 13.8%, and 19.2% lift over baseline for artists, directors, and scientists respectively), with the effect explicitly correlational and modest. Lesson for a writer: explore widely, then commit hard; and because the streak is unpredictable and finite, keep producing rather than betting everything on being “due.”

(f) A named writer’s process: Stephen King, On Writing (2000). King’s documented method is a near-textbook divergence-convergence loop. He drafts at a fixed high volume — “I like to get ten pages a day, which amounts to 2,000 words” — to keep the story fresh; he writes the first draft “with the door closed” (private, generative) and revises “with the door open” (public, evaluative), a principle he attributes to his early editor John Gould; and he applies the revision formula “2nd Draft = 1st Draft – 10%.” He treats stories as “found things,” relics excavated like fossils rather than pre-plotted (“plotting and the spontaneity of real creation aren’t compatible”) — pure discovery-mode generation, disciplined afterward by convergent cutting. Lesson: high-volume closed-door generation + ruthless open-door convergence is exactly what the science recommends, embodied in one prolific career.

(g) Human-AI collaboration in creative writing (Doshi & Hauser, 2024). In a controlled experiment (N = 293 writers) published in Science Advances (10(28):eadn5290), writers given GPT-4 story ideas produced eight-sentence stories that were “evaluated as more creative, better written, and more enjoyable, especially among less creative writers.” But “generative AI–enabled stories are more similar to each other than stories by humans alone.” The authors frame this as a social dilemma: “an increase in individual creativity at the risk of losing collective novelty.” Girotra/Meincke et al. (2023, “Ideas are Dimes a Dozen,” Wharton Mack Institute working paper) similarly found GPT-4 generates ideas faster and rated higher on average; comparing 200 student ideas with 200 GPT-4 ideas, of the top 10% (40 ideas) only five were from students and 35 were from ChatGPT — but the AI ideas were less novel/unique. Honest assessment: AI raises the floor and the average; it does not (yet) raise the ceiling of distinctiveness, and it narrows the collective space. For a writer whose value is distinctiveness, this is both a tool and a warning.

#10. Practical Framework

#Principles (the non-negotiables)

1. Never generate and evaluate at the same time.
2. Feed the combinatorial store daily (read widely; capture everything).
3. Produce volume; select ruthlessly; expect Sturgeon’s Law to apply to you.
4. Protect the closed-door draft from all evaluation pressure.
5. Run a portfolio; you cannot pick winners in advance.

#The Creative Practice Protocol

Divergence-convergence session structure (a single writing block):

• Warm-up (5 min): re-read yesterday’s last paragraph only — not the whole manuscript.
• Divergent drafting (50–90 min, door closed): write forward at volume toward a fixed word target. No editing, no deleting, no judging. If stuck, use a constraint (“write this scene in dialogue only”).
• Hard stop, then disengage. Do not revise the same day.
• Convergent pass (separate session, door open): cut, restructure, apply the −10% rule, test against novelty × value.

Idea-generation system feeding the knowledge base:

• Keep a permanent capture inbox (idea log). Aim for volume — this is your equal-odds engine.
• Weekly, migrate the best fragments into your knowledge base, tagged by project. Most will die; that is correct.

Incubation-and-review loop:

• After genuine preparation on a stuck problem, deliberately disengage with a low-cognitive-load activity (walk, shower, chores) — the condition Sio & Ormerod found maximizes incubation.
• Let finished drafts rest (King uses roughly six weeks) before the convergent pass, to weaken fixation and restore evaluative distance.

Filtering rule (which ideas to develop):

• Score each candidate on novelty (to your genre) and value (to your intended reader), plus energy (does it still pull you after a week?). Develop only ideas that survive on all three after an incubation gap. Kill the rest without mourning.

Portfolio rule (how many bets, when to kill/promote):

• Run a barbell: one or two “core” projects in active drafting (exploitation) + several small cheap experiments (short stories, premises, openings — exploration).
• Promote an experiment to core status when it survives three independent review gaps and still generates energy and reader interest.
• Kill a core project when it has stalled through two full incubation cycles without renewed pull — sunk cost is not signal.
• Mirror the Liu et al. hot-streak sequence: explore broadly, then, when something catches, commit hard.

#A two-week exercise

• Days 1–7 (divergence/exploration): every day, generate ten distinct story premises (100 words each) — 70 total. No editing. This is deliberate quantity to exploit the equal-odds logic and defeat perfectionism.
• Day 8 (incubation): do not look at them. Take a long walk daily.
• Days 9–10 (convergence): score all 70 on novelty × value × energy. Select the top three.
• Days 11–13 (exploitation): draft an opening scene for each of the three, door closed.
• Day 14 (selection + reflection): apply the −10% revision to each; pick the one to develop; write one paragraph on why — and notice how unreliable your Day-1 favorites turned out to be. That gap is the chapter’s lesson made personal.

#11. Criticisms and Limitations

Applying this chapter’s own critical standards to its sources:

• The measurement problem is unsolved. Divergent-thinking tests have weak ecological/predictive validity; the CAT measures against shifting domain consensus and shows rater drift. The field may be partly studying what is easy to measure rather than creativity itself.
• The lab may not capture real creativity. Insight paradigms (nine-dot, remote associates) are narrow; incubation effects, though meta-analytically real, are modest (d = 0.29) and mechanistically unresolved.
• The domain-generality debate is open. If creativity is largely domain-specific (Baer, Sawyer), then general theories and general tests are of limited use — a challenge to the whole psychometric edifice.
• Defining “value” is genuinely hard. Value is judged by a field whose standards shift over time (Simonton’s point that acclaim and neglect can reverse). This makes creativity partly unmeasurable in the present tense.
• Historiometry is correlational and selective. Simonton’s and Liu et al.’s findings rest on eminent, long-career samples; the hot-streak work is explicitly correlational, modest in effect size, and visible only in hindsight.
• The deliberate-practice/10-year rule transfers poorly to feedback-sparse creative domains — a caveat the Expertise chapter’s enthusiasm should be tempered by here.
• Cultural variation. Most of this literature is Western; what counts as novel and valuable — and even whether individual creativity is prized over tradition — varies across cultures. (Notably, Agarwal, Naaman & Vashistha, 2025, CHI, found AI writing suggestions homogenize toward Western styles.)
• AI raises unresolved ethical and practical questions — authorship, training-data provenance, and the collective-diversity cost of individually rational AI use.

No perspective here is absolute. The honest summary is that creativity science is a young, methodologically contested field whose most robust findings are also its least glamorous.

#12. Future Directions

The frontier is being reshaped by generative AI. The most important open question is whether machine fluency raises or lowers the bar for human creativity. The current evidence points to a specific reallocation: as generation becomes cheap and abundant, the scarce, valuable human contribution shifts to evaluation, curation, taste, and distinctiveness — the convergent half of the loop. Girotra/Meincke et al. put it bluntly: cheap idea generation “may substantially reduce the importance of the idea-generation phase and shift managerial focus to the idea-evaluation phase.” Expect creativity education to be redesigned around judgment rather than ideation, and expect anti-homogenization techniques (diverse AI personas, prompt variation — Wan & Kalman, 2025; Meincke et al., who found chain-of-thought prompting most effective at boosting AI idea diversity) to become a research and workflow priority. Computational-creativity research will keep probing whether LLMs can be genuinely novel or are structurally bound to the center of their training distribution (the homogenization findings suggest the latter, for now). Neuroscience (Beaty et al., 2025, Communications Biology, N = 2433 across five countries) will continue mapping creativity to dynamic switching between the default-mode and executive-control networks — a biological echo of the divergence-convergence model. For the working writer, the durable bet is that the value of a distinctive, well-judged human voice rises precisely as fluent generation becomes free.

#13. Recommended Resources

Beginner

• Creativity — Mihaly Csikszentmihalyi (1996). The systems model and rich interviews with eminent creators; readable and paradigm-shifting.
• On Writing — Stephen King (2000). A working master’s process that, unintentionally, illustrates nearly every research finding in this chapter.
• Ethan Mollick, “Automating Creativity” (One Useful Thing). A clear, evidence-based synthesis of the AI-ideation studies.

Intermediate

• Explaining Creativity — R. Keith Sawyer (2012). The best single-volume, evidence-forward textbook; strong on the measurement problem and domain-specificity.
• Kaufman & Beghetto (2009), “Beyond Big and Little: The Four C Model of Creativity,” Review of General Psychology. The developmental framework in one paper.
• Amabile, “Componential Theory of Creativity” (HBS working paper, 2012). The intrinsic-motivation principle from the source.

Advanced

• Sio & Ormerod (2009), “Does Incubation Enhance Problem Solving? A Meta-Analytic Review,” Psychological Bulletin. The definitive quantitative treatment.
• Simonton, Creativity in Science / the equal-odds and BVSR papers. The historiometric tradition at full strength.
• Doshi & Hauser (2024), Science Advances; Girotra/Meincke et al. (2023). The primary AI-and-creativity experiments.
• Finke, Ward & Smith, Creative Cognition (1992). The Geneplore model at book length.
• Liu, Wang, Sinatra et al. (2018, Nature; 2021, Nature Communications). The quantitative science of creative careers.

#14. Self-Check

Attempt these from memory before re-reading:

1. Why are the two criteria novelty and value both necessary, and who decides “value” in Csikszentmihalyi’s model?
2. What is the measurement problem in creativity research, and how do divergent-thinking tests and the CAT each fall short?
3. What does the nine-dot problem actually demonstrate about insight, once you account for Alba & Weisberg’s and Kershaw & Ohlsson’s findings?
4. State the equal-odds rule and one important qualification to it.
5. Why does classic brainstorming underperform nominal groups, and what fixes it?
6. Distinguish controlling from enabling extrinsic motivation in Amabile’s revised principle.
7. What did Doshi & Hauser (2024) find about individual vs. collective creativity under AI assistance, and why is it a “social dilemma”?
8. How does the exploration-then-exploitation “hot streak” finding translate into a portfolio rule for your own projects?

Synthesis check: A strong answer set will connect these into one argument — that creativity is ordinary cognition (insight, incubation) cycling between divergence and convergence, resting on domain expertise and intrinsic motivation, judged by a shifting field, and best pursued through disciplined high-volume production across a portfolio because neither you nor the science can predict which work will land. If your recall of the AI findings did not include both the individual gain and the collective-diversity loss, revisit Sections 4 and 9; if your equal-odds answer omitted the Openness moderation or the unpredictability-of-the-hit point, revisit Section 6.

---

## Knowledge Card — Creativity: The Science of Novel and Valuable Production
- Core terms:
  - Creativity: production of work that is both novel and valuable (useful/appropriate/fit), judged within a domain.
  - Divergent vs. convergent thinking: generating many varied ideas vs. selecting/evaluating toward the best one.
  - Four-C model: mini-c (learning), little-c (everyday), Pro-c (professional), Big-C (eminent) levels of creativity.
  - Geneplore model: creativity as alternating generative (preinventive structures) and exploratory (interpret/refine) phases.
  - Equal-odds rule (Simonton): number of hits rises with total output; which work is the hit is unpredictable.
  - Consensual Assessment Technique (CAT): experts independently rate products; the field's "gold standard," with a shifting-consensus caveat.
  - Incubation: modest, real benefit of deliberate low-load disengagement after preparation (Sio & Ormerod, d = 0.29).
  - Homogenization effect: AI assistance raises individual output but narrows collective diversity (Doshi & Hauser, 2024).
- Core mental models:
  - Novelty × Value axes for diagnosing any idea or draft.
  - Divergence–convergence cycling: never generate and evaluate at once.
  - Quantity-begets-quality portfolio/barbell: many cheap bets, ruthless selection, explore-then-exploit.
- Connections to prior chapters: Curiosity (fuels exploratory generation); Flow (immersive generation, but in tension with the evaluation phase); Deliberate Practice/Expertise (domain knowledge as the substrate creativity recombines; 10-year rule debate); Cognitive Biases (unreliable self-evaluation; planning fallacy in creative projects); Signal vs Noise (idea filtering; Sturgeon's Law on your own drafts); Antifragility (creative portfolio as barbell of cheap bets with uncapped upside); Behavioral Economics (intrinsic vs. extrinsic motivation).
- Recommended next chapter: Judgment & Decision-Making Under Uncertainty — because as generation becomes cheap, evaluation and taste become the binding constraint on creative value.
- One habit to keep: Separate generation from evaluation in time — draft with the door closed at volume, revise with the door open ruthlessly.