Mathematics Anxiety, Working Memory, and Mathematics Performance in Secondary-School Children

Mathematics anxiety is not simply a dislike of numbers. Passolunghi and colleagues define it precisely as "a feeling of tension and anxiety that interferes with the manipulation of numbers and the solving of math problems in a wide variety of ordinary life and academic situations." This definition captures something real: apprehension, frustration, and fear, which arise specifically around math, not learning in general. It matters because higher levels of this anxiety correlate with weaker calculation skills. Anxiety worsens performance, poor performance deepens negative beliefs about math ability, and those beliefs fuel further avoidance. It creates a cycle. Now, picture grades six through eight — a time when curricula shift toward abstract reasoning, faster fact retrieval, and multi-step procedures. For a child with math anxiety, a timed quiz or a public problem on the board isn’t just uncomfortable. It feels threatening. The consequences aren’t just momentary; they shape course choices and academic trajectories, influencing whether a student ever voluntarily picks up a math-heavy subject again. So, the question Passolunghi and colleagues wanted to answer is: what exactly is going wrong, cognitively, for these kids? To explore this, they looked at working memory. Working memory is the brain’s mental scratch pad — the limited pool of resources you draw on to hold and manipulate information while solving a problem. When you're doing long division in your head, working memory does most of the heavy lifting. Individual differences in working memory capacity predict a lot of academic outcomes, and two theoretical frameworks place it at the center of math anxiety. The first is Processing Efficiency Theory, proposed by Eysenck and Calvo. The core idea is simple: worrying consumes cognitive resources. If anxiety fills up your mental scratch pad with intrusive thoughts about failure, there’s less space left to actually do the math. The second framework, Attentional Control Theory, developed by Eysenck and colleagues, refines this idea by specifying which cognitive operations are disrupted. Anxiety, according to this theory, shifts the balance of attention away from the goal at hand toward threat-related thoughts. The specific casualties are inhibition — the ability to filter out irrelevant information — and shifting, the capacity to move flexibly between mental tasks. Both are essential for sustained mathematical thinking. Inhibitory control deserves special attention. When you’re solving a math problem, your brain needs to suppress distracting thoughts, candidates for wrong answers, and anxiety-related rumination. If inhibition is compromised, irrelevant material can intrude. You make more errors, you lose your place, and the task becomes harder not because you lack knowledge, but because your mental workspace is cluttered. Passolunghi and colleagues set out to test whether high math anxiety actually looks like this in real middle school students. They screened 135 children aged eleven to thirteen from state schools in Northern Italy, using an adaptation of the Abbreviated Math Anxiety Scale, which includes nine items on a five-point Likert scale and had a Cronbach's alpha of 0.81. They also used the Revised Children's Manifest Anxiety Scale, second edition, which measures general anxiety, and a verbal ability measure. The design was careful. To isolate math anxiety from general anxiety and verbal ability, the researchers selected only those students whose general anxiety and vocabulary scores were average. From that screened pool, they formed two groups: 32 students with high math anxiety, also referred to as HMA, whose math anxiety scores exceeded one standard deviation above the mean, and 34 with low math anxiety, or LMA. The groups were matched on age, gender, general anxiety, and vocabulary. The math anxiety gap between them was substantial — an F statistic of 144.91, with a p-value below 0.0001. Everything else was held constant. Both groups then completed a two-session battery. Collectively, they took the AC-MT 11-14 standardized mathematics battery, covering written calculation, magnitude judgment, place value, logical reasoning, approximate calculation, and a timed fact-retrieval task — which involved thirty-two simple calculations in two minutes. They also completed a reading comprehension task. Individually, they were tested on word reading, writing, verbal short-term memory using a Word Span Forward task, and working memory using a Listening Span Test, which required judging sentences as true or false while remembering the final word of each. This last task also tracked intrusion errors — cases where irrelevant words came into recall — serving as a direct measure of inhibitory failure. The results were clean and specific. Students with high math anxiety underperformed those with low math anxiety across four math measures: written calculation, magnitude judgment, logical reasoning, and fact retrieval, with the LMA group outperforming HMA by more than half a standard deviation on each. The only math subtest that showed no significant difference was approximate calculation. Critically, reading comprehension, word reading, and writing showed no significant group differences at all. This indicates a math-specific deficit, not a general learning difficulty, and not a language problem — it is math. On the cognitive side, HMA students recalled significantly fewer words on the short-term memory task and the listening span task. They also made more intrusion errors on the listening span — meaning irrelevant words bled into their recall more often. That pattern aligns exactly with what Attentional Control Theory would predict: anxiety degrading inhibitory control and cluttering the workspace with material that should have been filtered out. Then came the logistic regression, which posed a sharper question: which variables best predict whether a given child falls into the high- or low-anxiety group? Two predictors survived. Fact retrieval, with a Wald statistic of 7.16 and a p-value of 0.007. And intrusion errors on the listening span, with a Wald statistic of 7.00 and a p-value of 0.008. Together, these two measures correctly classified 79.4 percent of LMA children and 68.8 percent of HMA children, for an overall classification accuracy of 74.2 percent. The model explained roughly 24 percent of the variance in group membership, and a Hosmer-Lemeshow goodness-of-fit test confirmed that the model fit the data well. That pairing is telling. Fact retrieval is a timed, high-pressure math task — exactly the kind of situation that activates anxiety. Intrusion errors reflect a failure to suppress irrelevant information. One is a measure of math performance; the other is a measure of cognitive control. Together, they capture both sides of the problem: what these students struggle to do mathematically, and why their working memory lets them down when they try. This leads us to a question the study raises but cannot fully resolve: does anxiety drain working memory, or do weak working memory skills produce anxiety? The data fit both narratives. Students with high math anxiety show reduced working memory and inhibitory control, consistent with the notion that anxiety is consuming cognitive resources. However, they also demonstrate weaker math skills, including on tasks like fact retrieval that heavily depend on practice and fluency. Weak math competence and high anxiety co-occur so tightly that separating cause from effect requires longitudinal data — tracking students over time to see which comes first. Passolunghi and colleagues explicitly acknowledge this and call for follow-up work. What the cross-sectional design can confirm is that the two variables are entangled and that inhibitory control sits at the center of both, making the theoretical account provided by Attentional Control Theory a good fit for the results. The practical implications follow quite directly. If inhibitory control and fact retrieval distinguish high-anxiety students from low-anxiety students, then targeting math content alone won't suffice. The paper suggests several classroom-level steps: identify students with high math anxiety early, reduce unnecessary time pressure — since timed conditions activate the threat-focused attention shifts described by Attentional Control Theory — and provide positive corrective feedback. Anxiety tends to increase after negative feedback and decrease following positive feedback, as shown by Daniels and Larson. Working memory training and inhibitory control practice are also worth considering alongside standard math instruction. The deeper point is this: a child who struggles in math class may not be struggling because they haven’t tried, or because they lack raw ability. They may be struggling because anxiety has compromised the very cognitive machinery they need to succeed. Verbal working memory, inhibitory control, the ability to suppress irrelevant thoughts and hold the problem in mind long enough to solve it — these are the targets. Recognizing this is the first step toward genuinely helping. This lecture was created by ennepō. Go to https://ennepo.ai to Discover, Create and Follow the latest research in your field. Read when you can. Listen when you want to.