The Excellent Rear Left Wheel

There’s a metaphor in Think! Before It’s Too Late that does more work than Edward de Bono probably intended. The rear left wheel of a motor car, he writes, can be absolutely perfect — flawlessly engineered, optimally inflated, beyond criticism on any technical ground — and still be insufficient. You need the other three wheels. Insisting on the excellence of the one does not address the absence of the others.

He means this as a defense of his lateral thinking toolkit against educational traditionalists who ask why, given such impressive achievements in science and technology, we should change anything about how we teach human beings to think. But the metaphor cuts both ways. Apply it to de Bono’s own program and you start to see the precise shape of what Think! gets right, what it leaves undemonstrated, and what it reveals — almost despite itself — about the most consequential unanswered question in educational design today.

What the Book Actually Argues

The argument has three layers, and they require separation.

The first layer is diagnosis. Western education, de Bono contends, has been shaped by a specific inheritance: the Greek Gang of Three — Socrates, Plato, Aristotle — transmitted through the Renaissance Church, which needed logic, truth, and argument to defeat heresy and had no use for design, creativity, or perceptual flexibility. The result is an educational tradition that excels at pattern recognition and standard-solution application while systematically neglecting three cognitive modes: design thinking (constructing value-delivering arrangements for the future), perceptual thinking (examining how situations are framed before reasoning about them), and possibility thinking (treating hypotheticals as exploratory tools rather than logical failures).

This diagnosis is largely defensible. You can trace it in curriculum structures — the dominance of content delivery over metacognition, the elevation of critical analysis over generative synthesis, the near-total absence of explicit instruction in how to perceive a problem before solving it. David Perkins’s research, cited in the book, reportedly shows that 90% of errors in ordinary reasoning are perceptual rather than logical. If that finding holds under scrutiny, it constitutes a damning indictment of educational systems that spend nearly all their cognitive-skill investment on logic.

The second layer is mechanistic. De Bono claims the brain is a self-organizing asymmetric patterning system. Information settles into existing channels — like rain finding the riverbeds cut by earlier rain. New information is processed through dominant patterns. The implication is that what is obvious in hindsight is structurally inaccessible in foresight: the ant on a specific leaf reaches the trunk with certainty; the ant on the trunk reaches that specific leaf with odds of approximately 1:8,000. Creative insight isn’t mysterious. It’s geometrically necessary in asymmetric systems. You need an entry point from the periphery — a random word, a provocation, a deliberate lateral movement — because the dominant path prevents you from finding the side path by reasoning forward from the center.

This is the book’s most important and most underappreciated claim. It provides the first genuinely mechanistic explanation for why creative ideas are logical in hindsight without being reachable by forward logic. Murray Gell-Mann, the physicist who discovered the quark and founded the Santa Fe Institute for complex systems research, commissioned simulations to test the model and confirmed its behavioral predictions. That’s not proof of the literal neural claim, but it is significant evidence that the formal properties of the model are coherent.

The third layer is prescriptive: teach de Bono’s specific tools — PMI, OPV, C&S, lateral thinking, Six Hats — and watch performance improve across every domain. This is where the argument requires the most scrutiny and receives the least.

The Atkey Problem

The empirical foundation for de Bono’s educational prescription rests heavily on a single source. The Atkey organisation, he reports, has shown that teaching his thinking as a separate subject improves performance in every other subject by between 30 and 100 percent. He cites this finding five times across multiple chapters, in different registers — as established fact, as counterargument to critics, as justification for curriculum reform, as evidence for mandatory thinking courses.

Thirty to one hundred percent improvement across every subject. This would be among the largest educational intervention effects documented in research. The systematic review literature on tutoring, feedback, and metacognitive instruction — which includes some of the most robustly replicated findings in educational psychology — generally reports effect sizes in the range of 0.3 to 0.7 standard deviations. A 30-100% improvement in academic performance would require effect sizes substantially larger than anything in that literature. If true and generalizable, it would not be a footnote in a popular book. It would reorganize every curriculum on earth.

The Atkey organisation appears to be a proprietary training company, now operating as Blue Sky Skills. The research is never cited to a peer-reviewed publication. No independent replication is mentioned.

This is not a reason to dismiss the tools. It is a reason to be precise about what is established and what is claimed. The PMI framework — directing attention explicitly to Plus aspects, then Minus aspects, then Interesting anomalies — produces demonstrable shifts in group judgment in de Bono’s own facilitated experiments. Twenty-nine of thirty twelve-year-old boys changed their minds about student payment after applying PMI, without teacher intervention. Eighty-six percent of senior Canadian executives changed their position on equal pay after applying Consequences and Sequels. These before-and-after demonstrations describe something real about the gap between claimed and actual consideration. They just don’t have control groups, and they don’t demonstrate transfer across subjects at 30-100%.

The distinction matters for education. There is a version of de Bono’s program that is well-supported: explicit attention-direction tools reliably shift the scope of consideration within a given session, and this effect is large enough to be pedagogically significant. There is a version that is not yet supported: these tools produce large cross-domain academic performance gains when taught as a standalone subject. Curriculum designers need to know which version they are acting on.

The Embedded vs. Explicit Fork

Here is where Think! Before It’s Too Late most productively collides with contemporary educational technology.

Recent work on intelligent tutoring systems — AutoTutor being the paradigm case — takes a different structural bet. AutoTutor embeds cognitive skill development inside content tutoring through extended Socratic dialog. It scaffolds student reasoning on specific problems, uses latent semantic analysis to assess conceptual proximity, and pushes toward what the research calls “deep reasoning” through questions that require elaboration, self-explanation, and inference rather than factual recall. The approach has published effect sizes. It reports null results alongside positive ones. It distinguishes what the data shows from what is concluded.

De Bono’s CoRT programme does something almost structurally opposite: it teaches thinking as a decoupled explicit subject, with tools labeled by acronym, practiced on neutral scenarios, intended to transfer automatically into every domain.

These are competing hypotheses about the same problem. Does cognitive skill development transfer better when embedded in content (AutoTutor’s bet) or when isolated as explicit transferable technique (de Bono’s bet)? The educational research literature offers partial answers — explicit instruction in metacognitive strategies shows transfer effects, but those effects are sensitive to how explicitly the transfer is taught. Neither program directly addresses the other’s architecture.

What neither program addresses adequately is the perception problem. AutoTutor’s entire design assumes that students have correctly perceived the problem and need help reasoning about it. Its scoring mechanisms — comparing student responses to expert answers using semantic similarity — presuppose that the student is working within the right conceptual frame. But if Perkins is right that 90% of reasoning errors are perceptual rather than logical, then the student who most needs help isn’t the student who reasons incorrectly from a correct perception. It’s the student who has framed the problem wrong from the start, and who will reason impeccably toward the wrong answer.

De Bono’s OPV tool — Other People’s Views — is a blunt instrument for addressing this. You stop. You deliberately adopt a different frame. You ask what the situation looks like from the position of each stakeholder. The Karee mine finding, if taken at face value, describes a violence reduction so large (210 fights per month to 4) from so simple an intervention that it demands explanation. The explanation the data suggests is that the miners were operating inside a single perceptual frame — threat, insult, tribal loyalty — and the OPV requirement forced a different entry point into the same situation. That’s the asymmetric patterning argument applied directly.

Neither intelligent tutoring systems nor critical thinking curricula have this tool. They are optimizing the logical layer. The perceptual layer is largely unaddressed in both traditions.

What the Book Gets Wrong About Itself

Think! is, structurally, an argumentative book. It makes claims, marshals evidence, attacks opposing positions. This is, by de Bono’s own framework, the wrong mode for the kind of thinking it advocates.

The unfalsifiability problem is its most serious methodological flaw. De Bono argues that institutions defend the status quo by using critical thinking to attack novelty. He then categorizes any challenge to his work as an instance of this pathology. This closes the argument. A theory that immunizes itself from refutation by categorizing evidence against it as a symptom of the disease is not a scientific claim — it is an ideology. The book is written by someone who has been right about several important things for forty years and who has, in the process, stopped building in the structures that would tell him where he is wrong.

The twenty-three reasons section is not twenty-three reasons. It is approximately six to eight recurring themes — GG3 heritage, Church influence, perception neglect, complacency, institutional inertia, language limitations — restated with different labels. This is organizational inflation passing as analytical precision. The Palace of Thinking and Society of Thinkers proposals appear in the same register as demonstrated tools despite being aspirations without governance structure, budget, theory of change, or failure mode analysis.

The recognition section — cataloguing awards, the minor planet named after de Bono, Accenture rankings — functions as credentialing in a book that has spent considerable energy criticizing academic credentialism. The irony is unacknowledged.

These are not small problems. They are the precise problems the book teaches readers to identify.

What the Book Gets Right That Matters

The asymmetric patterning argument is structurally correct and educationally underutilized. If you accept that the brain forms dominant pathways and that creative solutions require entry from a different starting point, then the entire educational practice of “think harder about the same problem” is misguided. Effort applied to the dominant path does not increase the probability of reaching the side path. A different entry point does. This is teachable. The random word tool, the provocation technique, and the concept fan are all implementations of the same structural insight.

The judgment/design distinction deserves more traction than it has received. Education teaches students to identify what is — to recognize, categorize, analyze, and evaluate. It teaches almost nothing about constructing what could be. Design is not a vocational specialization. It is a cognitive mode that every person navigates every day: designing a conversation, a career, a family policy, a response to a conflict. The absence of explicit design thinking from general education is not a minor gap. It is a structural absence with significant consequences for how people approach problems that don’t have standard solutions — which is to say, most of the problems that matter.

The OPV, PMI, and C&S tools are demonstrably attention-shifting in facilitated settings. They are also genuinely simple to teach and use. The 2009 publication date of this book means de Bono was proposing explicit metacognitive tools decades before the educational psychology literature began producing large-scale evidence for their value. He was, on the question of whether attention-direction tools change judgment, ahead of the evidence — and the evidence, where it exists, tends to support the direction if not the magnitude.

The Question Education Won’t Ask

Here is the question the book raises without quite posing it directly:

If we accept that perception precedes logic, and that perceptual errors outnumber logical errors by approximately nine to one, then what is the ratio of educational investment in perceptual training to logical training?

In nearly every curriculum, the answer is close to zero to everything. Logic, evidence evaluation, argumentative structure — these receive explicit instruction across twelve years of compulsory schooling. Perception — how to examine the frame of a problem, how to generate alternative interpretations, how to deliberately adopt a different starting point — receives almost none.

De Bono’s tools are one answer to this problem. They may not be the only answer, or the optimal implementation, or effective at the magnitudes claimed. But the problem they address is real and unaddressed. That is the rear left wheel argument applied correctly. The existing wheel — logical-critical training — may be excellent. The missing wheels are not less important for being harder to engineer.

The students designing educational AI systems in 2025 are inheriting this gap. AutoTutor’s architecture optimizes deep reasoning. It doesn’t address perceptual framing. Every next-generation tutoring system built on the same assumption — that the student is perceiving the problem correctly and needs help reasoning about it — will produce the same blind spot at scale.

That’s the honest conclusion the book earns, even when the evidence for its specific prescriptions remains thin. The diagnosis is defensible. The mechanism is coherent. The prescription needs controlled testing it has not yet received.

The rear left wheel is excellent. Education still needs the other three.