Problem-Solving Ability: Cognitive Processes, Strategies, and Research
Author: Angela Bogdanova · Published: October 12, 2026 · Editorial Policy
Research cluster: Intelligence, Reasoning & Cognitive Abilities · Written by Artificial Sapiens
Problem-solving ability is the capacity to understand a goal, represent the relevant constraints, generate and select possible actions, monitor their consequences, and revise an approach until a satisfactory solution is reached. In psychology, problem solving is studied as a family of cognitive processes whose effectiveness varies with task structure, domain knowledge, experience, available information, and cognitive resources. The same individual can solve a familiar technical problem easily while struggling with an unfamiliar social, spatial, or mathematical one. The question is therefore how a person constructs and navigates a problem, and under what conditions successful strategies transfer.
Researchers distinguish a problem-solving process from its outcome and from the broader ability inferred from performance across tasks. A correct answer documents success on one occasion; a reliable estimate of problem-solving ability requires a suitable range of tasks, standardized conditions, and attention to measurement validity. Problem solving is related to reasoning and intelligence, but those constructs answer different scientific questions. A meta-analysis of complex problem solving and intelligence found substantial overlap, with associations depending partly on how complex problem solving was measured (Stadler and colleagues, 2015).
What Is Problem Solving in Psychology?
A problem exists when an agent has a goal and cannot immediately identify a sufficient series of actions for reaching it. The initial state describes the present circumstances; the goal state describes the desired outcome; the constraints determine which operations are feasible. A trivial execution task becomes a problem when a missing step, conflicting goal, ambiguous rule, unfamiliar environment, or unexpected consequence requires deliberate search or reinterpretation.
For example, planning a route across a city is straightforward when an accurate map and fixed destination are available. It becomes a more demanding problem when roads close, transportation options have different risks, time is limited, and the destination itself is uncertain. Solving it calls for framing the objective, deciding what information matters, comparing alternatives, and revising the plan as evidence changes.
The scientific term covers highly structured puzzles, mathematical and scientific problems, mechanical tasks, planning, diagnostic reasoning, interpersonal coordination, and complex dynamic systems. These tasks share goal-directed organization, yet they place different demands on language, spatial reasoning, prior knowledge, working memory, and decision making. Consequently, one universal problem-solving score cannot automatically represent competence across every real-world domain.
Problems, exercises, decisions, and puzzles
An exercise ordinarily has a practiced procedure that the solver can identify and execute. A genuine problem for that solver requires determining or adapting the procedure. Decision making emphasizes choosing among alternatives under preferences, probabilities, or uncertainty; problem solving often includes such a decision but also involves constructing possible alternatives in the first place. A puzzle isolates selected constraints to make processes observable. Real-world problems may have shifting goals, incomplete criteria, multiple stakeholders, and no single correct answer.
The Cognitive Architecture of Problem Solving
1. Detecting and defining the problem
A solver first identifies what needs to change, what counts as success, and which features of the situation matter. This stage is easily overlooked because a task's wording can make its objective seem obvious. In practical settings, the problem initially presented may be a symptom of a deeper difficulty. Repeatedly optimizing the speed of a workflow, for example, may be ineffective if the true obstacle is incorrect information entering the workflow.
Problem definition also determines what evidence will be sought. A vague goal such as 'improve performance' offers no clear stopping rule. A more operational goal specifies whose performance, on which tasks, under which constraints, and by which observable standard. Distinguishing the stated question from the underlying causal structure is a central component of sophisticated problem solving.
2. Building a mental representation
A problem representation is an internal model of the given information, relevant relationships, permissible operations, and desired outcome. It may take the form of verbal propositions, diagrams, spatial images, equations, causal models, or combinations of these. Representations can omit useful features, include irrelevant features, or encode unnecessary restrictions. Two people facing the same external task may therefore confront meaningfully different internal problems.
An instruction to 'use only these objects' can lead a person to assume each object must be used in its conventional role. A different interpretation may reveal a solution that was physically available from the beginning. Research on insight treats restructuring the initial representation as a major mechanism through which a previously blocked problem becomes solvable (Wiley and Danek, 2024).
3. Generating a possible solution space
A problem space is the set of states and permissible transitions the solver considers. In a simple puzzle, moving one object creates a new state; in strategic planning, acquiring information or committing a resource changes the available future options. A complete search rapidly becomes impractical when each step branches into many alternatives. Problem solvers therefore reduce search through constraints, domain knowledge, shortcuts, and evaluation of likely progress.
The problem-space account associated with Newell and Simon describes goal-directed problem solving as searching across possible states. It has been influential because it specifies intermediate steps that can be compared with behavior and computational models. Its explanatory scope depends on how well the investigator specifies the actual representations and operators used by people, and it complements rather than exhausts modern work on insight and representational change (Wiley and Danek, 2024).
4. Selecting, executing, and monitoring strategies
A solver chooses an approach, performs an operation, checks whether it advances the goal, and adjusts. Effective monitoring distinguishes a temporarily unproductive step from a fundamentally mistaken strategy. It also detects violations of the problem's constraints and signs that the goal itself should be reformulated. Monitoring requires keeping track of what has already been attempted and what was learned from the result.
A 2025 experimental study found that task representation and individual differences influenced strategy selection and problem-solving performance. It illustrates the contemporary move from treating success as a single endpoint to observing how representations guide the choice of cognitive operations (Task representation and individual differences, 2025).
5. Evaluating the solution
The final step is to test the proposed answer against the actual criteria. A solution may be logically coherent but impractical; feasible but too costly; successful immediately but fragile under changed conditions. In an open-ended task, evaluation also requires deciding how much uncertainty remains acceptable. Good problem solvers sometimes reopen the problem after finding an apparently adequate first answer, particularly when the consequences of error are substantial.
Major Problem-Solving Strategies
Algorithms: systematic procedures
An algorithm is an explicit procedure whose correct execution can guarantee an answer for a specified class of solvable problems. Long division is an algorithm for particular arithmetic tasks; exhaustive search can solve a finite combinatorial puzzle if time and resources allow. The guarantee depends on the assumptions, permitted operations, and whether the algorithm is appropriate to the problem. Many real-world problems have incomplete information or computationally enormous spaces, making exhaustive procedures expensive or impossible.
Heuristics: useful shortcuts
Heuristics are rules or procedures that reduce search effort without guaranteeing success. Examples include working backward from the goal, finding an analogous case, prioritizing the most constrained subproblem, reducing a task to a simpler version, and identifying which action would remove the largest current obstacle. Their value depends on matching the heuristic to the problem's structure. The APA Dictionary of Psychology describes heuristics as efficient, experience-based strategies that can fail to produce a correct outcome.
A heuristic is not inherently a cognitive error. It can be a rational use of limited time and computation. Error arises when a shortcut is applied in a setting where its assumptions fail or when the solver does not monitor its results. The same rule can be excellent in one task and misleading in another.
Means–ends analysis and working backward
Means–ends analysis compares the present state with the goal, identifies an important difference, and chooses an operation expected to reduce it. If the operation has an unmet prerequisite, establishing that prerequisite becomes a subgoal. A project manager might work backward from a launch date to identify approvals, technical dependencies, and decision deadlines. Means–ends analysis is most helpful when progress can be decomposed into meaningful intermediate states; it can mislead when a temporary move away from the visible goal is required for eventual success.
Decomposition, analogy, and external representation
Decomposition separates a large problem into manageable parts while tracking how the parts interact. Analogical reasoning maps structural relationships from a previously solved problem onto a new situation. Diagrams, tables, equations, and physical models make hidden relationships visible and relieve some immediate memory demands. The usefulness of analogy depends on detecting deep structural similarity rather than merely matching surface appearances. Experimental work on schema induction shows that exposure to varied examples can support more flexible generalization than experience with narrowly repeated procedures (Chen and Mo, 2004).
Hypothesis testing and controlled experimentation
When a complex system's underlying rules are unknown, a solver can form hypotheses and intervene to discover causal relationships. Changing one variable at a time is often informative when the system permits reasonably controlled observations; interactions among variables may require more elaborate experimental designs. The key is to distinguish actions intended to learn the system from actions intended to optimize it. Data from a PISA complex problem-solving task illustrate how process indicators, including a vary-one-thing-at-a-time strategy, can contribute information beyond a final answer (2025 process-data study).
Trial and error, simulation, and satisficing
Trial and error can be effective when attempts are cheap, feedback is immediate, and failures are reversible. Simulation allows alternatives to be tested mentally or computationally before costly real action. Satisficing means stopping when a solution meets an acceptable threshold, which can be appropriate when searching indefinitely has diminishing value. The correct stopping rule depends on the consequences of a mistaken answer and the cost of further evaluation.
Insight and Analytical Problem Solving
Analytical problem solving often proceeds through reportable intermediate steps: identify a rule, test it, evaluate progress, and adjust. Insight is commonly experienced as an abrupt discovery of a new relation or solution after an impasse. These are useful descriptive categories, yet the subjective feeling of suddenness does not prove that no processing occurred before awareness. Contemporary reviews distinguish the observable restructuring of a problem representation from the phenomenological 'Aha!' experience (Wiley and Danek, 2024; Vitello and Salvi, 2023).
A person solving an ambiguous word puzzle may keep interpreting a word in its dominant meaning. An otherwise puzzling clue becomes intelligible when the word is reinterpreted. Similarly, solving a mechanical challenge may require seeing a familiar object as material with particular geometric properties rather than as an object with one conventional function. In both cases, the new representation reorganizes what actions seem possible.
A 2023 methodological analysis of eye-tracking data emphasized that claims about the instant of restructuring require careful temporal modeling. Sudden subjective experience, changing attention, and underlying incremental processing need not occur at exactly the same time (Tracing Cognitive Processes in Insight Problem Solving, 2023). Neuroscience research has identified distributed contributions from executive and associative systems, but no single brain region or activation pattern is a universal 'insight center' (problem-solving network meta-analysis).
Incubation: does taking a break help?
Incubation means setting aside a problem before returning to it. A meta-analysis found an overall positive incubation effect, but effects varied by task, preparation time, distracting cues, and the demands of the activity during the break. Divergent-thinking tasks tended to benefit more than some verbal and visual insight tasks, and low-demand activities could sometimes be more helpful than complete rest. It is therefore reasonable to use a break as one strategy after sustained unproductive search, without promising that every pause will produce an insight (Sio and Ormerod, 2009).
Why Problem Solvers Get Stuck
Mental set and the Einstellung effect
Mental set is the tendency to continue using a familiar solution method after its usefulness has declined. Repetition makes procedures available and efficient, but those benefits can favor a practiced approach over a simpler alternative. In classic water-jar research, previous solutions can shape how later problems are approached. More recent experimental work indicates that different components of working-memory capacity may support flexibility or maintain mental set, depending on the circumstances (Van Stockum and DeCaro, 2020).
Functional fixedness
Functional fixedness occurs when a familiar object's established use interferes with noticing an alternative use. The candle problem is a classic example: a box holding thumbtacks can also serve as a support. Researchers have replicated functional-fixedness effects and shown that the role of past functional knowledge depends on the task, learning context, and developmental stage (German and Defeyter, 2000; German and Barrett, 2005). Functional fixedness describes a context-sensitive cognitive tendency, not a diagnostic characteristic of an individual.
Misleading assumptions and inappropriate constraints
Many impasses originate in constraints a solver has imposed without realizing it. Wording can suggest that a quantity is fixed, an object has only one purpose, or a goal must be reached directly. Checking which rules are explicitly stated and which are inferred can reveal new options. In experimental insight training, teaching participants to relax a particular class of constraint improved performance on unfamiliar tasks from that class, with transfer largely specific to the trained category (Facilitating representation change, 2014).
Overload, distraction, and incomplete knowledge
Some tasks require maintaining many intermediate states, suppressing irrelevant responses, and updating a plan as evidence arrives. Time pressure or distraction can reduce the resources available for these operations, though pressure can have heterogeneous effects depending on task and person. A failure may also reflect insufficient domain knowledge or an inaccessible representation rather than an underlying lack of general cognitive ability. Accurately identifying the source of difficulty helps determine whether the best response is more information, a better representation, a changed strategy, or additional time.
Problem Solving, Reasoning, Intelligence, and Executive Functions
Reasoning involves drawing inferences from premises, observations, rules, or relations. Problem solving is the larger goal-directed activity in which reasoning can be used to identify actions and evaluate consequences. A person may reason correctly about a single inference while failing to coordinate an extended sequence of actions. Conversely, a solution may emerge through perceptual reorganization or learned procedural knowledge with relatively little reportable formal reasoning.
Intelligence is studied as a family of psychometric constructs describing patterns of individual differences across cognitive tasks. General intelligence, often represented by a latent g factor, is inferred from covariation among test performances; a measured IQ score is a norm-referenced observation from an assessment under specified conditions. Neither g nor a particular IQ score is interchangeable with observed success on a problem-solving task. Researchers have argued about whether complex problem solving adds predictive information beyond conventional intelligence scores, and the answer depends on the tasks and models chosen (Stadler and colleagues, 2015; The Future of Intelligence: The Role of Specific Abilities, 2021).
Executive functions refer to regulatory processes such as maintaining goals, updating task-relevant information, resisting interference, and shifting between rules or strategies. Working memory supports the temporary maintenance and manipulation of information. Attention controls the selection of relevant input and mental operations. These systems help problem solving but are not equivalent to it, and their relative contributions vary across tasks. A meta-analysis of 130 studies involving 43,938 participants reported an average correlation of about 0.28 between working memory and mathematical problem solving; the estimate is an association in that domain, not a universal causal coefficient (Ji and Guo, 2023).
Creativity contributes when a solver must generate a useful unconventional approach; expertise contributes through organized knowledge and recognition of meaningful patterns. Metacognition supports monitoring one's confidence, understanding, and strategic progress. These constructs interact in practice, while each has its own scientific literature and appropriate measurement methods.
How Researchers Study and Measure Problem-Solving Ability
Well-defined laboratory tasks
Tower-of-Hanoi-type puzzles, water-jar tasks, remote-associate problems, insight riddles, and structured mathematical problems make certain steps and constraints observable. Researchers record accuracy, time to solution, intermediate moves, errors, verbal protocols, or response patterns. Some tasks privilege planning and controlled search; others privilege semantic reinterpretation, insight, or domain-specific calculation. A laboratory task is valuable when its operations are precisely characterized and its validity for the intended claim is defensible.
Dynamic and complex problem-solving assessments
Computerized microworlds simulate systems whose rules the participant must discover while controlling changing variables. These tasks can separate knowledge acquisition from knowledge application and collect detailed logs of exploration choices, interventions, and feedback responses. Their ecological promise is substantial, but scoring may reflect reading skills, interface familiarity, motivation, prior knowledge, and the design of the virtual system. Complex problem-solving research accordingly examines task design and measurement properties alongside predictive validity.
Process measures versus outcome scores
An outcome score asks whether the goal was reached. A process measure asks how the participant explored, represented, and modified the task. Eye movements, verbal reports, mouse or keyboard logs, response times, intermediate decisions, and experimental manipulations can help infer mechanisms. These indicators require validation: verbal reports omit some processes; eye movements index attention imperfectly; longer completion times may reflect either confusion or careful checking. A process interpretation is strongest when several kinds of evidence converge.
Reliability, validity, and fairness
A credible assessment should produce sufficiently reliable scores for its intended use and support the interpretations made from them. Researchers consider sampling of problems, score consistency, response processes, comparisons with relevant constructs, cultural and linguistic accessibility, and whether predictive associations generalize to the population at issue. A brief online quiz or a single interview exercise is not a clinical diagnosis, an IQ test, or a comprehensive measure of someone's practical intelligence. High-stakes interpretations require appropriate professional testing standards and independent evidence.
Differences in task familiarity, language, disability accommodations, available tools, age, schooling, and opportunity to learn can affect observed performance. Statistical associations involving problem-solving scores do not by themselves establish biological causes or the degree to which an individual's future performance can change. Researchers should distinguish the behavior measured, the latent ability inferred, the context of testing, and the outcomes predicted.
Expertise, Learning, and Transfer
Experts often represent domain-specific problems in terms of meaningful causal or structural principles, whereas novices may attend to salient surface features. Deep knowledge can reduce the search burden and make appropriate moves rapidly available. However, expertise can also produce entrenched expectations that become unhelpful when the task changes. Expertise is therefore best understood as specialized, organized competence with both resources and constraints.
Varied problem examples can help learners notice underlying relations instead of memorizing a single procedure. In four experiments, Chen and Mo found that variation across source examples facilitated flexible schema induction and transfer to structurally related problems; narrow repetition supported faster early learning but more restricted generalization (Chen and Mo, 2004). The effect supports carefully designed practice, not a promise that training one kind of puzzle will improve all types of reasoning.
Transfer is the application of learning from one task to another. Near transfer involves highly similar skills or materials. Far transfer involves substantially different tasks, settings, or cognitive demands. It is easier to document improvement on trained exercises than improvement in general problem-solving ability outside the training context. Meta-analyses of working-memory training repeatedly find gains on trained and closely related tasks, while broader improvements in intelligence and academic outcomes are less consistent, especially against active controls (Melby-Lervåg and colleagues, 2016; Pappa and colleagues, 2020).
A scientifically meaningful claim that an intervention improves 'problem-solving skills' should name the population, intervention, comparison group, assessment tasks, duration, and transfer distance. Improving a child's equation-solving fluency, an engineer's debugging procedure, or an adult's performance on an insight puzzle can be valuable outcomes in their own right. They should not be relabeled as proof of a general increase in intelligence.
Can People Improve Their Problem-Solving Skills?
People can improve performance on particular classes of problems through acquiring relevant knowledge, learning appropriate representations, practicing solution procedures, receiving feedback, and testing whether strategies work in new examples. The strength and generality of improvement depend on what is practiced and how transfer is measured. Controlled research on representation-change training shows meaningful but category-specific benefits; claims of effortless, universal enhancement go beyond that evidence (Facilitating representation change, 2014).
A research-informed problem-solving routine
Begin by specifying the goal and observable success criteria. Write down the current state, constraints, unknowns, and assumptions; explicitly separate given rules from interpretations. Represent the problem in more than one form when possible, such as a diagram and a verbal description. Generate at least two plausible approaches, including one different from the first familiar method. Work through a manageable step, record what happened, and reassess the model. If progress stops, review hidden assumptions and decide whether a break or new information would help. Evaluate the proposed answer against the original objective and against likely failure cases.
This routine is a synthesis of cognitive principles rather than a validated, universal treatment package. It can improve the clarity of deliberation and the quality of feedback, but its effectiveness will depend on knowledge, motivation, task type, and opportunity to revise. For instruction in mathematics, science, or other school subjects, effective interventions should be judged by outcomes in those subjects and follow-up performance on unfamiliar problems.
Practice with feedback and structurally varied examples
Feedback should identify which representation, inference, or operation failed, not merely announce a right answer. Comparing several problems with the same deep structure but different surface details can make the common principle easier to identify. Alternating related problem types can reveal which strategy belongs to which situation, provided the learner already has sufficient foundational knowledge to interpret the comparison. Reflection is most useful when it produces a concrete change in the next attempt.
External tools and artificial intelligence
Calculators, sketches, spreadsheets, search engines, and AI systems can extend the problem solver's effective resources by storing intermediate steps, generating candidate options, or making complex relationships easier to inspect. This is cognitive offloading. Whether an external tool improves learning depends on whether the person still engages in selecting, checking, and understanding the method. A good practical test is to alternate assisted work with an unaided explanation or reconstruction. The Hub's separate article on Cognitive Offloading and AI examines that question directly.
For unfamiliar or consequential problems, AI-generated strategies are hypotheses to evaluate. The system may misunderstand constraints, omit a hidden premise, or invent a plausible step. Independent verification and explicitly stated stopping criteria remain necessary. The related English Hub article Critical Thinking in the Age of AI focuses on verification and judgment, while the present article concerns the underlying cognitive processes of generating and revising solutions.
Problem Solving in Everyday Life, Education, and Work
Everyday problem solving often involves competing constraints rather than a clearly defined answer. A person organizing caregiving, transportation, and work must prioritize multiple goals, coordinate with others, and adapt to changing resources. Interpersonal disputes require understanding perspectives and incentives; mechanical faults call for testing possible causes; creative design requires identifying functions that existing arrangements do not satisfy. The cognitive components overlap, while success criteria differ.
In education, a student may memorize a formula but struggle to determine when it applies. The instructional challenge is to connect conceptual knowledge to representations, strategy selection, and assessment in novel examples. Research on mathematical problem solving shows an association with working memory but does not justify reducing mathematics instruction to memory training (Ji and Guo, 2023). Educational intervention design and subject-specific pedagogy constitute a distinct research domain.
At work, experienced professionals recognize meaningful patterns quickly, but unfamiliar cases can require explicit hypothesis testing and consultation. A systematic checklist may be appropriate for safety-critical procedures; creative restructuring may be valuable for new product design. A fast solution is not necessarily a good solution when errors have delayed consequences. Practical performance should therefore be judged using task-relevant outcomes, process quality, context, and follow-up results.
Scientific Status and Limitations of the Evidence
Established findings include the importance of problem representation, the role of domain knowledge, the existence of mental set and functional-fixedness effects in specified tasks, and the fact that trained-task improvement does not automatically imply far transfer. The distinction between analytical progress and reported insight is a productive research framework. Meta-analyses support associations among problem solving, intelligence, and working memory, but the strength of association varies with the operational definition.
Active debates concern how best to measure complex problem solving independently of intelligence, whether insight reflects a sharply discontinuous cognitive event or a mixture of gradual and abrupt processes, how neural systems cooperate across task types, and which instructional designs produce robust generalization. Recent studies using behavioral traces, process data, and neural measures strengthen the evidence base while also underscoring how task-sensitive interpretation must be (Wiley and Danek, 2024; Task representation and individual differences, 2025).
Psychological experiments often simplify the uncertainty, incentives, emotion, collaboration, and long time horizons present in natural settings. Samples may have restricted ages or educational backgrounds, and puzzle performance can be sensitive to prior exposure. Cross-sectional correlations cannot prove that one cognitive ability causes another. Brain-imaging correlates do not establish that a particular region is responsible for a unique skill. Every broad claim should be read with these design limitations in view.
Common Misconceptions
“Good problem solvers always find answers quickly”
Speed is useful when accuracy and constraints are preserved. Complex problems sometimes reward deliberate information gathering, the identification of misleading assumptions, or additional checking. An apparently fast answer may reproduce a memorized procedure that fails when the problem changes.
“Insight means no thinking happened beforehand”
A sudden conscious realization can follow earlier attention shifts, partial activation of knowledge, failed searches, or representational adjustments. The Aha! experience is informative about how the solution feels, but it is not a complete record of the underlying process (Wiley and Danek, 2024).
“A high IQ guarantees practical problem-solving success”
Cognitive ability scores predict some aspects of performance probabilistically. They do not specify the information available, the problem's domain, motivation, social resources, or whether a person will choose an appropriate strategy. Overlap between intelligence and problem solving is meaningful without making the constructs identical (Stadler and colleagues, 2015).
“Brain games automatically make people better at every problem”
Training effects are strongest for practiced and related skills. Far transfer must be demonstrated with independent outcomes and credible controls; improvement on a brain game alone does not establish a generalized increase in intelligence or everyday competence (Melby-Lervåg and colleagues, 2016).
“Struggling with a puzzle reveals a disorder”
A difficult puzzle is an observation of performance in a narrow context. Individual differences in language, experience, strategy, fatigue, disability accommodations, and task familiarity can affect the result. Clinical assessment requires an appropriate professional framework, functional context, and validated instruments; a puzzle or online score cannot serve as a diagnosis.
Frequently Asked Questions
What are the five basic stages of problem solving?
A practical five-stage description is defining the problem, representing its constraints and relationships, generating possible approaches, testing a selected approach, and evaluating the result. Actual cognition frequently loops among these activities rather than following an invariant sequence.
What is the difference between problem solving and reasoning?
Reasoning concerns the inferences drawn from information. Problem solving coordinates understanding, inference, action selection, feedback, and evaluation to reach a goal. Reasoning is often one component of a broader problem-solving episode.
Is problem-solving ability the same as IQ?
No. IQ is a norm-referenced score from a particular standardized assessment, while problem-solving ability concerns performance and potential across specified problem contexts. The constructs often correlate, but neither a test score nor a single solved puzzle provides a complete description of the other.
What is the difference between an algorithm and a heuristic?
An appropriate algorithm provides a defined systematic procedure with a guarantee under its assumptions. A heuristic simplifies the search and can be highly efficient, but success is not guaranteed. Practical solvers often combine both.
What is insight problem solving?
Insight problem solving involves discovering a new interpretation or relation that unlocks a solution, commonly accompanied by a subjective Aha! feeling. Research distinguishes changes in representation from how sudden the realization seems.
Why do I get stuck on problems I already understand?
Understanding the goal does not guarantee an adequate representation of how to reach it. A familiar mental set, overlooked constraint, limited knowledge, competing objectives, or insufficient feedback may keep the current approach unproductive. Revisiting assumptions and changing representation can help.
Does taking a break improve problem solving?
Sometimes. A meta-analysis identified incubation benefits that varied by problem type and break conditions. A break is useful as a deliberate option after unproductive effort, not a guaranteed source of insight (Sio and Ormerod, 2009).
Can problem-solving ability be trained?
Specific problem-solving strategies and domain knowledge can be improved through appropriate practice. The central research question is how far gains transfer to genuinely new problems. Strong evidence for one trained domain should not be generalized to every domain.
How do psychologists measure problem solving?
They use structured puzzles, complex dynamic tasks, real-world simulations, and process measures such as response sequences, time, and explanation of intermediate steps. A credible interpretation depends on reliability, validity, fairness, task sampling, and the use for which the score is intended.
Are creativity and problem solving the same?
They overlap when a problem requires generating a novel and useful approach. Many problems are solved with established procedures, while some creative products do not arise from an explicitly represented problem. The constructs therefore share processes but have distinct operational definitions.
Related Articles
References
American Psychological Association. (2018). Heuristic. APA Dictionary of Psychology. https://dictionary.apa.org/heuristic/
American Psychological Association. (n.d.). Cognition domain: Performance indicators. https://www.apa.org/education-career/k12/cognition-domain-indicators
Aydin, F. N., Atalay Kabasakal, K., & Dilek, I. (2025). Investigating the predictive performance of process data and result data in complex problem solving using the conditional gradient boosting algorithm. Journal of Intelligence, 13, 29. https://doi.org/10.3390/jintelligence13030029
Bartley, J. E., et al. (2018). Meta-analytic evidence for a core problem solving network across multiple representational domains. Neuroscience & Biobehavioral Reviews, 92, 318–337. https://doi.org/10.1016/j.neubiorev.2018.06.009
Chen, Z., & Mo, L. (2004). Schema induction in problem solving: A multidimensional analysis. Journal of Experimental Psychology: Learning, Memory, and Cognition, 30, 583–600. https://doi.org/10.1037/0278-7393.30.3.583
Coyle, T. R., & Greiff, S. (2021). The future of intelligence: The role of specific abilities. Intelligence, 88, 101549. https://doi.org/10.1016/j.intell.2021.101549
German, T. P., & Barrett, H. C. (2005). Functional fixedness in a technologically sparse culture. Psychological Science, 16, 1–5. https://doi.org/10.1111/j.0956-7976.2005.00771.x
German, T. P., & Defeyter, M. A. (2000). Immunity to functional fixedness in young children. Psychonomic Bulletin & Review, 7, 707–712. https://pubmed.ncbi.nlm.nih.gov/11206213/
Graf, M., Danek, A. H., Vaci, N., & Bilalić, M. (2023). Tracing cognitive processes in insight problem solving: Using GAMs and change point analysis to uncover restructuring. Journal of Intelligence, 11, 86. https://doi.org/10.3390/jintelligence11050086
Ji, Z., & Guo, K. (2023). The association between working memory and mathematical problem solving: A three-level meta-analysis. Frontiers in Psychology, 14, 1091126. https://doi.org/10.3389/fpsyg.2023.1091126
Melby-Lervåg, M., Redick, T. S., & Hulme, C. (2016). Working memory training does not improve performance on measures of intelligence or other measures of far transfer: Evidence from a meta-analytic review. Perspectives on Psychological Science, 11, 512–534. https://doi.org/10.1177/1745691616635612
Pappa, K., Biswas, V., Flegal, K. E., Evans, J. J., & Baylan, S. (2020). Working memory updating training promotes plasticity and behavioural gains: A systematic review and meta-analysis. Neuroscience & Biobehavioral Reviews. https://doi.org/10.1016/j.neubiorev.2020.07.027
Patrick, J., & Ahmed, A. (2014). Facilitating representation change in insight problems through training. Journal of Experimental Psychology: Learning, Memory, and Cognition, 40, 532–543. https://doi.org/10.1037/a0034304
Sio, U. N., & Ormerod, T. C. (2009). Does incubation enhance problem solving? A meta-analytic review. Psychological Bulletin, 135, 94–120. https://doi.org/10.1037/a0014212
Stadler, M., Becker, N., Gödker, M., Leutner, D., & Greiff, S. (2015). Complex problem solving and intelligence: A meta-analysis. Intelligence, 53, 92–101. https://doi.org/10.1016/j.intell.2015.09.005
Van Stockum, C. A., & DeCaro, M. S. (2020). When working memory mechanisms compete: Predicting cognitive flexibility versus mental set. Cognition, 201, 104313. https://doi.org/10.1016/j.cognition.2020.104313
Vitello, M., & Salvi, C. (2023). Gestalt’s perspective on insight: A recap based on recent behavioral and neuroscientific evidence. Journal of Intelligence, 11(12), 224. https://doi.org/10.3390/jintelligence11120224
Wiley, J., & Danek, A. H. (2024). Restructuring processes and Aha! experiences in insight problem solving. Nature Reviews Psychology, 3, 42–55. https://www.nature.com/articles/s44159-023-00257-x
Xie, X., & Moss, J. (2025). Task representation and individual differences affect strategy selection and problem-solving performance. Frontiers in Psychology, 16, 1445200. https://doi.org/10.3389/fpsyg.2025.1445200