Recent math anxiety research is pushing practice routines in a useful direction: reduce threat, but do not remove thinking. Students need low-pressure chances to retrieve math facts, try procedures, correct errors, and notice that discomfort is not the same thing as inability.
The most direct new signal comes from a 2025 paper in the Journal of Experimental Psychology: Learning, Memory, and Cognition. Researchers studied how adults with high and low math anxiety learned unfamiliar multiplication problems across two sessions. The pattern matters for classrooms and home study: high-anxiety learners struggled more with unrepeated problems that required procedural learning, but they showed stronger learning when the same problems were repeated and could be retrieved from memory.
That does not mean anxious learners should only memorize. It means practice should be sequenced carefully. If every session is a stream of novel, timed problems, a learner with math anxiety may experience the task as threat plus uncertainty. If practice includes repeated retrieval, worked feedback, and gradual variation, the same learner gets evidence that knowledge can become more accessible with repetition.
A June 2025 University of Virginia education article made a related point from the developmental side. Cognitive neuroscientist Tanya Evans argued that the idea that many people are simply not math people is misleading, and the article noted that parents’ and teachers’ discomfort with math can influence children. Evans also described research connecting memory skills with later math performance and recommended making everyday math more familiar through simple counting, comparing prices, and memory games.
Meanwhile, classroom practice conversations are increasingly treating retrieval as a normal routine rather than a mini-test. In February 2026, Edutopia hosted a teacher discussion on making retrieval practice a classroom habit, with attention to how routines can support students who struggle with anxiety or processing speed. The timing is important: retrieval practice is most helpful when it feels like part of learning, not a surprise judgment of ability.
There is also a new technology wrinkle. A 2026 arXiv preprint, revised on July 13, analyzed millions of ALEKS math-learning interactions and reported that students spent less time on AI-susceptible math problems after ChatGPT’s release, with lower odds of correct answers on later proctored retention items. The paper is a preprint, so it should be read cautiously, but its practical warning fits the anxiety discussion: tools that make answers arrive faster can also reduce the effortful retrieval and error correction that build confidence.
What Happened
The current research picture is not saying that math anxiety is solved by confidence slogans. It is showing a more specific mechanism. Anxiety can change how learners approach math, which strategies they avoid, and how much productive effort they invest when a problem feels uncertain.
Earlier classroom intervention work also supports this shift. In a study published in npj Science of Learning, high school students received either emotion-regulation support or study-skills support. The study-skills condition emphasized spaced studying and retrieval practice through self-testing, practice problems, quizzes, and flashcards. The important practical detail is that the study-skills approach did not simply tell anxious students to work harder. It gave them a structure for approaching math more often, with better learning techniques.
Together, the newer arithmetic-learning paper, the UVA reporting, the classroom retrieval discussion, and the AI-study warning all point to the same design problem: math practice should create repeated contact with ideas without turning every contact into a high-stakes performance event.
Why It Matters
Math anxiety often creates a loop. A student feels threatened, avoids practice, loses fluency, then feels more threatened the next time math appears. Long worksheets, public speed drills, and vague reminders to “study more” can make that loop worse because they add pressure without improving the learner’s strategy.
The research suggests a better interpretation of confidence. Confidence is not the starting requirement for practice. It is an outcome of practice that is frequent, specific, and survivable. A learner becomes more confident after repeatedly retrieving a fact, solving a familiar problem type, seeing an error corrected, and then handling a slightly changed version.
This also changes how teachers, tutors, parents, and adult learners should use assessment. Low-stakes retrieval can prepare students for tests by making recall familiar. High-stakes testing without enough retrieval practice can make the test feel like the first real attempt, which is exactly when anxiety is most likely to consume attention.
The Practical Learning Conclusion
Design math practice as a confidence-building routine: lower the threat, keep retrieval active, and increase difficulty through small variations.
- Start with a short success set. Open with three to five problems the learner can probably retrieve or solve. The point is not to inflate confidence; it is to make the first contact with math feel manageable.
- Use retrieval before rereading. Ask the learner to write a formula, fact, step, or example from memory before looking at notes. Then check and correct it immediately.
- Repeat, then vary. Practice a small set until it becomes easier, then change one feature: numbers, context, order of steps, or wording. Avoid jumping straight from one example to a completely new problem type.
- Remove unnecessary speed pressure. Fluency matters, but early practice should prioritize accurate recall and clear steps. Timed work belongs later, after the learner has a stable method.
- Make errors routine. Keep an error log with the missed problem, the reason for the miss, and the corrected step. This turns mistakes into information instead of evidence of failure.
- Use AI as a coach, not a substitute. Ask for hints, similar practice problems, or explanations after an attempt. Do not let the tool replace the retrieval attempt itself.
- End with one confident repetition. Close by redoing one corrected problem without help. That final retrieval gives the learner a concrete memory of improvement.
The lesson for practice routines is simple but demanding: anxious learners need more contact with math, not more threat. Retrieval practice, spacing, feedback, and gradual variation can turn that contact into evidence that math gets easier when practice is built to make thinking visible and repeatable.