What working memory is
Working memory (WM) is the cognitive system that holds a small amount of information in an active, readily available state while you use it. When you follow a complex instruction, perform mental arithmetic, or track the thread of an argument across paragraphs, working memory is doing the heavy lifting.
It is distinct from long-term memory (which stores information indefinitely) and from short-term memory (a simpler buffer). Working memory is active and manipulative — you do not just hold information in WM, you work with it.
Why it matters for IQ
Working memory capacity correlates strongly with fluid intelligence (Gf) — typically r=0.5–0.7 in research studies. The relationship is so strong that some researchers have proposed WM capacity is the primary mechanism underlying g (general intelligence). Tests of WM — particularly complex span tasks that require concurrent storage and processing — are among the best predictors of performance on fluid reasoning tasks.
Can working memory be trained?
Research on WM training has produced a complicated picture. The headline findings from the 2000s and early 2010s — particularly the claim that n-back training improves fluid intelligence — generated enormous excitement and produced a wave of commercial products. Subsequent replication attempts have been more sobering.
The current consensus: WM training reliably improves performance on the trained task and on closely similar tasks (near transfer). Far transfer — improvement on tasks measuring fundamentally different cognitive abilities like reasoning and problem-solving — is weak and inconsistent.
N-back training
Dual n-back training, popularised by Jaeggi et al. (2008), involves monitoring a sequence of stimuli and responding when the current item matches the item from n positions back. The original paper reported transfer to fluid intelligence, creating enormous interest. Multiple independent replications have found near-zero transfer to untrained tasks.
A 2013 meta-analysis (Melby-Lervåg & Hulme) found that WM training produces specific, near-transfer improvements that do not generalise to measures of fluid intelligence or academic achievement. A 2019 update reached similar conclusions.
Evidence-based strategies
The most robustly supported approaches to maintaining and supporting working memory are not training programs: they are lifestyle factors. Sleep deprivation strongly impairs WM performance — recovering adequate sleep reliably restores it. Aerobic exercise produces small but consistent improvements in WM capacity across meta-analyses. Stress and anxiety impair WM by occupying attentional resources with threat-monitoring.
For learning contexts specifically, chunking (grouping information into meaningful units), spaced repetition, and deliberate practice in the domain you care about produce real improvements in functional memory capacity within that domain — through knowledge acquisition rather than WM expansion per se.
The bottom line
Working memory cannot be substantially expanded through training in a way that transfers to general cognitive ability. What you can do is ensure it operates at capacity by sleeping well, managing stress, and staying physically active. For domain-specific tasks, developing expertise and knowledge structures reduces the WM load required — which functionally increases the complexity of what you can handle.