Cognitively Guided Instruction: Letting What Kids Already Know Drive the Lesson

In 1985, a team of mathematics education researchers at the University of Wisconsin Madison began a project that would eventually become one of the most influential bodies of research in elementary mathematics education. Their starting question was both simple and radical: what do children actually know about mathematics before they are taught it?

The researchers, led by Thomas Carpenter and Elizabeth Fennema, spent years documenting the informal mathematical strategies that young children use to solve problems when left to approach them in their own way. They found something that surprised many educators of the time: children as young as five or six, without any formal instruction in addition or subtraction, could solve quite complex word problems using their own invented strategies, built from intuitive understandings of quantity and operation that they had developed through everyday experience.

From this research emerged an approach to mathematics teaching called Cognitively Guided Instruction, or CGI. It has since been validated through decades of classroom research, adopted in schools across the United States and internationally, and shown repeatedly to produce stronger mathematical understanding and greater mathematical confidence than traditional procedural approaches.

The core insight it rests on is as useful at a kitchen table as in a classroom: children do not arrive as empty vessels to be filled with mathematical knowledge. They arrive with knowledge already there, and the most effective teaching builds deliberately on what is already present.

The Core Principle: Children Think Before They Are Taught

Before formal instruction begins, children develop intuitive understandings of mathematical situations through everyday experience. They understand that adding more to a group makes it larger. They understand that sharing equally means everyone gets the same amount. They understand that if you take some away from a group, fewer remain.

These intuitive understandings are not yet expressed in mathematical notation, and they may not generalize to all problem types. But they are genuine mathematical knowledge, and they are the foundation on which formal mathematical understanding can be built most effectively.

CGI is built on the practice of presenting children with problems and then watching, listening, and asking questions about how they solved them, before doing any teaching. This practice reveals what the child already knows, how they are thinking about the problem, and precisely where their current understanding ends. Everything that follows is designed to extend that understanding from where it actually is, not from where the curriculum assumes it should be.

The Problem Types That Structure CGI

One of the most practically useful contributions of CGI research is the taxonomy of word problem types that Carpenter and colleagues developed through years of studying how children approach different mathematical situations.

They found that children's problem solving strategies were not random. The way a child approached a problem depended systematically on the structure of the problem: what was known, what was unknown, and what action or relationship the problem described. And the order in which different problem types became accessible to children was also systematic and predictable.

Joining problems involve adding to an existing quantity. "Maria had five apples. She got three more. How many does she have now?" These are typically the most accessible problems for young children because they map directly onto a physical action: put together and count.

Separating problems involve removing from a quantity. "Maria had eight apples. She ate three. How many does she have now?" Also highly accessible, and the action can be physically represented and counted.

Part part whole problems describe a situation with two parts that make up a whole, without a specific action. "There are five red apples and three green apples. How many apples are there?" These require understanding the relationship between parts and whole without a temporal action to guide the solution.

Comparing problems involve finding the difference between two quantities. "Maria has eight apples. Carlos has five apples. How many more apples does Maria have than Carlos?" These are typically harder for young children because the action of comparison is less physically obvious than joining or separating.

Multiplication and division problems appear in multiple forms: equal groups, arrays, and rate situations. Children develop different solution strategies for each, and their readiness for formal multiplication notation depends on their understanding of these underlying structures.

The reason this taxonomy matters for parents and homeschooling educators is practical. If you are presenting word problems to your child, presenting them in order from most to least accessible is not just considerate. It is how understanding develops. A child who cannot yet solve a comparison problem does not need more practice with comparison problems. They need more experience with joining and separating problems to build the numerical understanding that comparison problems require.

What CGI Looks Like in Practice

The practice of CGI is built around three teacher moves that translate directly into parent and homeschooling educator moves.

Posing a problem and standing back. Present a mathematical problem, ensure the child understands the situation, and then allow them to solve it in their own way without guidance or hints. The goal at this stage is to see what the child does, not to teach them what to do.

This is harder than it sounds for most adults. The impulse to help, to guide, to suggest a strategy when a child seems stuck is strong and well intentioned. But intervening before the child has had a genuine opportunity to think deprives you of the diagnostic information that watching their approach provides, and it deprives them of the experience of productive struggle that builds genuine understanding.

Listening to and understanding the child's solution. After the child has solved the problem, ask them to explain how they did it. Listen with genuine curiosity. What strategy did they use? What did they know that helped them? Where did they seem uncertain? What does their approach reveal about their current understanding?

This listening is not evaluation. It is the research that will guide your next instructional decision. A child who solves a joining problem by counting all objects from one is at a different place in their understanding than a child who counts on from the larger number, and a child who immediately recalls a known fact is at yet another place. Each of these positions calls for a different instructional response.

Extending the child's thinking. Based on what you have learned from watching and listening, pose a new problem that is designed to extend the child's current thinking. Not a problem they cannot solve, but a problem that is slightly harder than what they just did: one that invites them to apply their current strategy in a new context, or that creates a situation where their current strategy is less efficient and a more sophisticated one might emerge.

Why Invented Strategies Are Not Wrong

The most common concern that parents raise about CGI inspired practice is this: if I let my child solve problems their own way, will they develop habits that interfere with learning the standard methods?

The research answer is no. Children who develop their own strategies first, and who understand what they are doing at each stage, learn standard algorithms more quickly and with greater understanding than children who are taught the standard algorithm first.

The reason is that an invented strategy is a strategy the child understands. It is built from their own knowledge and their own reasoning. When the standard algorithm is introduced later, it can be connected to and compared with the strategy they already understand, which makes it meaningful rather than arbitrary.

A child who has spent time solving two digit addition problems by breaking the numbers into tens and ones and recombining them, their own invented strategy, arrives at the standard regrouping algorithm with a genuine understanding of what the regrouping represents. The algorithm is not a mysterious set of steps. It is a more efficient version of something they already understand.

Using CGI at Home Without a Classroom

You do not need a classroom, a teacher credential, or a deep familiarity with the CGI research literature to use its core practices at home. The essential moves are accessible to any parent.

Present word problems rather than computation exercises as the primary form of mathematical practice. Word problems force genuine mathematical reasoning in a way that bare computation does not, and they provide the context that children need to apply their intuitive understandings.

Ask "how did you do that?" every time your child solves a problem, whether or not the answer is correct. Make this question a genuine inquiry rather than an evaluation. What you learn from the answer will tell you more about your child's mathematical understanding than any assessment.

Resist the urge to teach before you understand what they already know. The time spent watching and listening at the beginning of any new mathematical topic is not wasted time. It is the research that makes everything that follows more efficient.

Follow the child's thinking forward rather than pulling them toward a standard method. If a child is using a strategy that works, the appropriate response is usually to extend that strategy to harder problems, not to replace it with a different one.