Teaching Problem Solving in Math: Getting Students Beyond Answer-Getting

The most common complaint about high school math students is that they can execute procedures but fall apart on novel problems. They can follow an algorithm when the problem type is obvious, but the moment a problem requires deciding which approach to use — or recognizing that they need to combine strategies — they stop.

This isn't a student failure. It's an instruction failure. A curriculum built entirely on "here is the procedure, now practice the procedure" produces exactly what it's designed to produce: students who can execute procedures. Teaching problem solving is a different instructional project, and it requires deliberate design.

The Problem with Procedure-First Instruction

When students learn mathematics primarily through worked examples followed by practice problems of the same type, they're building pattern recognition, not mathematical thinking. They learn "when I see this format, I do this algorithm" — which works fine for procedural fluency but fails completely when the problem format is unfamiliar, the strategy isn't obvious, or the student needs to transfer what they know to a new context.

This is exactly why students who are fluent on routine practice tests fall apart on applied assessments, SAT math, or college courses. They've been trained to pattern-match procedures, not to think mathematically.

What Mathematical Problem Solving Actually Requires

Problem solving in mathematics involves several distinct capacities that are rarely taught explicitly:

Problem representation. Translating a problem from verbal or symbolic form into a representation that makes the structure visible. This might be a diagram, a table, a graph, or a different symbolic form. Students who can't represent a problem can't solve it — and this skill is almost never taught.

Strategy selection. Knowing what approaches are available and making a judgment about which is likely to be productive. This is expert knowledge that novices don't have, which is why teaching problem-solving strategies explicitly — with names and examples — is more effective than simply giving students hard problems and hoping they figure it out.

Monitoring and revision. Checking whether a strategy is working, recognizing when it's not, and being willing to back up and try something different. Students who have been trained to execute procedures often lack this metacognitive dimension entirely — they run the algorithm to completion whether or not the answer makes sense.

Sense-making. Checking whether the answer is reasonable, what it means in the context of the problem, and whether it fully answers what was asked. "Does this make sense?" is a question that many students never ask because they've been trained that getting an answer ends the task.

Teaching with Problem Types, Not Just Problems

One of the most underused tools in mathematics instruction is teaching students to recognize problem types — categories of structure that call for particular strategies. This is different from pattern-matching to procedure formats; it's learning to recognize the underlying mathematical structure of a problem.

Polya's four phases (understand, devise a plan, carry out the plan, look back) give students a problem-solving framework. The CUBES strategy and other elementary frameworks serve the same function for younger students. The value isn't in any particular framework but in giving students a structured process to follow rather than staring at the problem until something occurs to them or panic sets in.

Worked Examples vs. Problem Solving

Research on worked examples (Sweller, others) shows that for novices, studying worked examples is more efficient than attempting to solve problems independently. This seems to contradict problem-solving instruction, but it doesn't — it clarifies the sequence.

For new content, worked examples are efficient for initial acquisition. As students develop some competence, problem solving becomes more appropriate. The failure mode in most classrooms is not using worked examples — it's using worked examples exclusively, never reaching the stage where students are actually doing mathematical thinking. The gradual release should move toward student-generated problem solving; it just shouldn't start there.

Productive struggle in math requires that students have enough foundational knowledge to actually engage with the problem. A student who doesn't know what linear equations are can't productively struggle on a linear systems problem. The pre-requisite is knowing enough procedures and concepts to have tools to try.

The Classroom Culture Question

Students who have been trained to answer-get are often resistant to problem-solving instruction because it feels like punishment. They want the procedure; the productive struggle feels like withholding help.

This resistance is information about history, not inability. Naming it directly — "I know this feels harder than what we usually do, and that's because we're doing something different: we're learning to think mathematically, not just follow procedures" — is more effective than pretending the resistance isn't happening.

Creating a classroom where wrong attempts are visibly valued, where "what have you tried" is the standard response to "I'm stuck," and where multiple approaches to the same problem are compared and discussed — this changes what students believe problem solving is. It stops being a test of whether they have the right answer in their head and becomes a collaborative mathematical investigation.

That shift doesn't happen from one lesson. It's built from consistent culture across a semester, from a teacher who responds to wrong attempts with curiosity rather than correction, and from instruction that treats the thinking as the point rather than the answer.