Cognitively Guided Instruction Math Problems

Working with Cognitively Guided Instruction Math Problems

I keep running into people who treat CGI as some kind of worksheet series. It isn't. Cognitive Guided Instruction is a framework for understanding how kids think through math, and the problems you pull from it are just entry points for that work. The actual method lives in how you respond to what a student says, not in the problem itself. The core idea comes out of research done at UW-Madison back in the mid-80s. Carpenter and Fennema spent years watching first through third graders solve word problems and noticed something that basically flipped conventional teaching on its head. Kids already had sophisticated strategies before any formal instruction. They were adding by counting all, then by counting on, then by decomposing. They understood part-part-whole relationships intuitively. Most classrooms were teaching procedures that bypassed those natural strategies entirely, which meant kids learned to follow steps without actually knowing what the steps meant. So the approach builds around observing a student's solution method and then scaffolding from there instead of jumping straight to the standard algorithm. In practice, you present a problem, let them work it, listen to their explanation, and then decide what to say next based on the reasoning you heard.

Cognitively Guided Instruction Math Problems

Here is what the problem types actually look like in the research taxonomy, and more importantly, how to recognize the thinking behind each one. Join problems are the starting point. Something happens, then something else happens. Three birds were on the fence, two more flew in, how many now? A kindergartener might count all the birds on their fingers. A second grader might say "3 and 2, that's 5" without counting. Both are correct. Your job is to note which strategy they used and then introduce a slightly harder version that pushes them past their current method without forcing a new one yet. Separate problems go the other direction. Seven frogs were on the log, three jumped in the water. How many stayed? These trip kids up more than joins because the result is unknown rather than the change. I had a fourth grader who could do multi-digit subtraction fluently but couldn't figure out this problem because she was looking for two numbers to subtract rather than understanding the situation. She kept trying 7 minus 3 when the question was actually asking about what remained after the change. We spent two weeks just talking about what the situation meant before I let her near any algorithm. That sounds excessive until you realize she was applying procedures mechanically to every problem and getting roughly half of them wrong regardless of the numbers involved.

Put-together problems involve two parts making a whole. No action verb, just a static relationship. Six stickers on my notebook are stars, the rest are dots, and there are nine stickers total. How many are dots? This is where part-part-whole thinking develops. Kids who struggle here usually haven't internalized that a whole can be broken apart in multiple ways. The workaround I use is physical manipulation. I give them ten blocks and ask them to split them into two piles however they want, then record each combination. After about eight variations, the structure starts clicking. The abstract version tends to click for them shortly after. Compare problems are the hardest type and usually appear later in the sequence. Tim has five marbles. Joe has eight. How many more does Joe have? Some kids subtract immediately. Some count up from five to eight. A problematic response I see constantly is kids answering thirteen because they see two numbers and default to addition. This is not a calculation problem. It is a language and situation comprehension problem, and drilling more subtraction facts does not fix it. The fix is having them physically model both quantities side by side and literally see the difference. From a teaching standpoint, the hardest part is resisting the urge to show the method. I used to move really fast through CGI problems because I wanted kids to reach the standard algorithm. Once I slowed down and actually listened to their reasoning, I noticed that the kids who got there fastest were the ones who had explored multiple strategies over months, not the ones I drilled on procedures. The tradeoff is real though. You will cover significantly less content in the first semester. Expect it. The payoff comes in the second semester when those same kids actually understand what multiplication means instead of just reciting facts.

One thing the literature glosses over is that this approach requires you to be comfortable with being wrong publicly. You have to sit there while a kid explains a strategy that leads nowhere and resist correcting them immediately. That is genuinely uncomfortable if you are used to being the person who knows the answer. The research shows it works, but the research also assumes a teacher who can tolerate that discomfort consistently. If you cannot do that, you will slip back into direct instruction after about three weeks regardless of what the papers say. There are also situations where CGI does not help much. If a student has a significant cognitive disability that affects working memory or numerical reasoning, the observational approach still applies but the pacing changes dramatically and you may need to pair it with more structured intervention. Same thing for students who are far behind grade level mathematically. The framework assumes a certain baseline of number sense. When that baseline is missing, you build it first through different activities, then bring in the CGI problem types later. If you want to actually use this, the original material comes out of the Everyday Mathematics curriculum and the Research Foundations documents from the Wisconsin Center for Education Research. The problem sets are not something you download and hand out. You select problems appropriate to your students' current thinking level, present them individually or in small groups, record what each student does, and then plan your next move based on the data. That record-keeping is the part nobody mentions enough. You need a simple tracking system that notes the problem type, the strategy observed, and what scaffolding you provided. Without that, you are just doing word problems with extra steps.