Problem Solving Skills: The Science of Better Solutions

Problem solving is often treated as a personality trait — something you either have or you don't. The research says otherwise. Problem solving skills can be learned, practiced, and measurably improved. The difficulty is that most people never receive systematic instruction in them; they pick up habits and heuristics through experience, which means they also pick up the blind spots.

This post covers what problem solving skills actually are, what the research says about the different types, and what kinds of practice produce durable improvement.

What Are Problem Solving Skills?

A problem exists when there's a gap between your current state and a desired state, and the path between them isn't obvious. Problem solving is the cognitive process of closing that gap.

Research since the 1970s has distinguished between well-defined problems — where the starting state, goal state, and allowable operations are all specified — and ill-defined problems, where one or more of those elements is unclear. Most real-world problems are ill-defined. You rarely know exactly what "solved" looks like until you're looking at the solution.

This distinction matters for skill development. Well-defined problems reward algorithmic thinking: follow the procedure, get the answer. Ill-defined problems require something different — the ability to impose structure on ambiguous situations, generate and evaluate multiple solution paths, and tolerate the discomfort of not knowing where you're going.

The Four Types of Problem Solving

Not all problem solving looks alike. Cognitive psychologists have identified four distinct modes, each suited to different kinds of problems.

Algorithmic Problem Solving

Algorithms are step-by-step procedures that, when followed correctly, guarantee a correct solution. Most math problems, most debugging tasks, and most rule-based business processes fall into this category. Algorithmic problem solving is reliable but brittle — it only works when the problem is well-defined enough to have an algorithm.

Heuristic Problem Solving

A heuristic is a rule of thumb that guides search through a problem space. Unlike algorithms, heuristics don't guarantee correct solutions — but they get you to good-enough solutions faster than exhaustive search. "Work backwards from the goal," "break the problem into smaller parts," and "find an analogous problem you've already solved" are all heuristics.

George Pólya's How to Solve It (1945) is still one of the best systematic treatments of mathematical heuristics. The heuristics he identified — draw a diagram, consider a simpler related problem, examine special cases — transfer well beyond mathematics.

Insight Problem Solving

Insight problems are those that produce the sudden shift in understanding where a solution becomes obvious all at once. The nine-dot problem, Duncker's candle problem, and most classic lateral thinking puzzles are insight problems. They resist systematic search and require restructuring the problem representation.

Research by Stellan Ohlsson established that insight occurs when you abandon an incorrect assumption about the problem — what he called "re-representation." The stuck solver is usually stuck because they've mentally framed the problem in a way that rules out the solution. The breakthrough comes when that frame collapses.

Analogical Problem Solving

Analogical reasoning involves recognizing structural similarity between a new problem and a problem you've already solved, then mapping the known solution across to the new domain. It's one of the most powerful problem solving tools available, and also one of the most trainable.

Mary Gick and Keith Holyoak's classic 1980 experiments showed that people could solve Duncker's radiation problem — how to use X-rays to destroy a tumor without damaging surrounding tissue — much more readily after reading a story about a military general who split his forces to converge on a fortress from multiple directions. The structural analogy is exact; the surface features are completely different. Most participants missed the connection entirely. When told the stories might be relevant, transfer rates jumped dramatically.

Core Problem Solving Competencies

Beyond the four types, research identifies several competencies that distinguish skilled problem solvers from novices.

Problem Representation

How you frame a problem determines what solutions become visible. Novice problem solvers accept the framing given to them. Expert problem solvers spend significant time reformulating: flipping goal and constraint, abstracting to the core structure, asking what the problem looks like from a different stakeholder's perspective.

This is what problem reframing is about in practice. The technique isn't a creativity exercise — it's a direct intervention in problem representation. The same underlying situation, described differently, generates different solutions.

Working Memory Management

Complex problems exceed working memory capacity. Skilled problem solvers externalize: they draw diagrams, build models, write out partial solutions, and use external structures to offload cognitive load. This is why mind mapping, sketching, and writing are problem solving tools, not just communication tools. The act of getting a partial idea onto paper frees up working memory for the next step.

Metacognitive Monitoring

Perhaps the most consistent finding in problem solving research: experts monitor their own progress more carefully than novices. They notice when they're stuck, recognize when a strategy isn't working, and switch approaches deliberately rather than persisting with a failing method.

This is a trainable habit. Metacognition as a practice — regularly asking "how is my current approach working?" and "what would I try if this approach were unavailable?" — measurably improves problem solving outcomes across domains.

Common Obstacles to Good Problem Solving

Several well-documented cognitive patterns reliably impair problem solving performance.

Functional Fixedness

Functional fixedness is the tendency to see objects and concepts only in their conventional roles. In Duncker's candle problem, participants struggle because they see the box of thumbtacks as a container rather than a platform. The box is available — they just don't perceive it as usable for that purpose.

The same pattern occurs in abstract problem solving. A standard organizational structure becomes an invisible constraint. A vendor relationship is treated as a fixed cost when it might be renegotiable. Skilled problem solvers deliberately ask: what is this thing at its most basic level, and what else could it do?

Mental Set

Mental set is the tendency to apply previously successful strategies to new problems even when they don't fit. Abraham Luchins demonstrated the effect clearly with his water jar problems: participants who learned a complex three-jar strategy to measure a specific volume of water kept using it even when a simpler two-jar approach was available right in front of them.

The cure is simple in principle and hard in practice: explicitly ask whether the approach you're about to use is right for this specific problem, or whether you're reaching for it because it worked last time.

Premature Closure

The tendency to stop exploring once you've found a workable solution. This produces adequate answers when better ones were available. Divergent thinking practice — generating many candidate solutions before evaluating any — is one of the most reliable antidotes. The first solution that comes to mind is usually the most obvious one, which usually isn't the best one.

Problem Solving Strategies That Transfer

Several strategies have enough research support that they're worth building into a default toolkit.

Decomposition. Break the problem into sub-problems small enough that each becomes tractable. Recursively apply decomposition until you reach problems you know how to solve directly. This is the core of systems thinking applied to problem solving — you're not solving the whole thing at once; you're solving a sequence of smaller things.

Working backwards. Start from the goal state and ask what would have to be true for the goal to be achieved. Then ask what would have to be true for that to be true. This is especially useful when the goal is clear but the path is not — it converts an open-ended search into a sequence of sub-goals.

Inversion. Instead of asking how to achieve the goal, ask what would guarantee failure. This flips the problem and often surfaces constraints and obstacles that weren't visible from the forward direction. Charlie Munger made this approach famous in investment analysis; it applies equally well to any domain where failure modes are easier to specify than success conditions.

Analogy search. When stuck, ask: where has a structurally similar problem been solved? The domains might look nothing alike — the solution might come from biology, military history, architecture, or cooking — but the underlying logic may transfer. This is the method behind analogical reasoning and forms the basis of TRIZ, the inventive problem solving system developed from systematic analysis of patents.

Imposing a constraint. Counterintuitively, limiting your options often improves solutions. Creative constraints research shows that restriction forces non-obvious recombinations and prevents the obvious-but-low-quality solution from crowding out better alternatives.

How to Build Problem Solving Skills

Most problem solving skill development happens incidentally — you encounter hard problems, you struggle, occasionally you have a breakthrough, and over time you accumulate useful heuristics. This is inefficient.

Deliberate practice for problem solving means:

• Solving problems with explicit attention to how you're solving them, not just whether you succeed
• Reflecting on what worked and what didn't, immediately after finishing while the process is fresh
• Seeking out problem types that don't fit your current toolkit, specifically to avoid over-relying on familiar strategies
• Practicing problem representation explicitly: before attempting to solve a problem, write out two or three different framings of it

Divergent thinking exercises train the core generative component of problem solving — the ability to produce many candidate solutions rather than stopping at the first one. This transfers across domains because the limiting factor in most problems isn't a shortage of obvious solutions; it's the absence of non-obvious ones.

Ready to train your creativity? Try science-backed exercises that measure and improve your creative thinking. Start a Free Exercise