The Corsi Block-Tapping Test: What It Measures and How to Improve

The apparatus is deliberately low-tech: nine blocks fixed at irregular, pseudo-random positions on a board — irregular so that blocks cannot be chunked into rows like a telephone keypad. The examiner taps a sequence at roughly one block per second; the participant reproduces it in the same order. Sequences lengthen until failure, and the span is the longest sequence reproduced correctly — classically averaged over a few trials at the boundary. Where the digit span test measures verbal short-term memory, Corsi measures its spatial twin: how much of the visual-spatial world you can hold and reproduce at once.

There is a backward variant — reproduce the sequence in reverse — which shifts the load from pure retention to manipulation, engaging working memory more heavily, exactly as backward digit span does. Performance drops by roughly one to two blocks in the backward direction, and the drop itself is informative: clinicians compare forward and backward spans to separate retention from manipulation difficulties.

The healthy adult Corsi span sits around five blocks — five to six is the normal band across most normative studies, with no meaningful average difference between men and women. If you expected more, you are in good company: people consistently overestimate their spatial span because they confound it with vision ('I saw all nine blocks'). You did see them; the test is about how many positions you can still act on seconds later, and five is the human standard.

The developmental curve is gentle by memory-test standards: spans rise through childhood from about three to four at age five, peak in early adulthood, and decline slowly with age — noticeably less steeply than the fast-capture abilities the chimp test probes. This slow decline is part of why Corsi is so clinically useful: a span far below five in a younger adult, or a rapid drop from a person's own baseline, is a signal worth investigating, while ordinary aging moves the number only gradually.

Clinically, Corsi-type tasks appear in the assessment of traumatic brain injury, stroke, dementia and ADHD, and in research on virtually anything that touches spatial cognition — because spatial and verbal short-term memory can dissociate: a patient can hold seven digits and only three blocks, or the reverse. The Wechsler Memory Scale's spatial span subtest is a direct Corsi descendant. The test's virtues are its brevity, its zero equipment needs and its resistance to language and education effects — it works the same whether you speak anything at all.

One honest caveat that comes with that clinical weight: a single Corsi score is a screen, not a diagnosis. Spans vary with sleep, attention, medication and motivation; clinicians interpret them against normative baselines and alongside everything else they know about the patient. If you are playing the browser version and scoring three, that is not a diagnosis — it is a number, and the next section is about what you can do with it.

The strategy that moves Corsi scores most is the one memory athletes use for everything: stop treating the sequence as blocks and start treating it as a route. Three taps become 'a walk: corner, hop to the middle, slide left.' Routes are how the memory palace technique encodes entire speeches, and the same machinery compresses block sequences the same way — you are trading seven loose positions for one path with seven stations, and paths are dramatically cheaper to store.

The second lever is chunking pairs into shapes: two consecutive blocks become a line, three become an L or a triangle. Most people spontaneously begin doing this around a span of five, which is arguably why five is where untrained spans sit — the natural chunking limit. Training pushes past it deliberately: encode two shapes of two blocks plus one anchor, and a span of six to seven is reachable for most people within a few weeks of short sessions.

• Trace the route physically. Let your eyes (or finger, on a touchscreen) draw the path between blocks during presentation — motor encoding adds a second memory channel.
• Name the shape. 'L, then diagonal, then corner' — verbal labels glue spatial sequences together.
• Chunk in twos. Two blocks per item halves the load; a span of six is three chunks.
• Practice the boundary. Growth lives at the length where you fail; drill sequences one longer than your current span, not comfortable ones.

Corsi Blocks reproduces the clinical task as a game — same irregular layout, same growing sequences, scored by blocks and streaks, so your span becomes a number you can watch move. Because the layout randomizes, you cannot memorize the board; every round is a fresh measurement, which keeps the score honest.

For a fuller picture, pair it with the tests that neighbor it: Sequence Memory for sequential order in a fixed grid, and the memory palace guide for the route technique at full size — encoding fifty items instead of five. The interesting thing about training Corsi specifically is how visible the strategy change feels: the week your score jumps, you will remember it as the week you stopped seeing blocks and started seeing walks.