Memory Benchmark

Treat it as a playful snapshot, not a clinical measurement. Memory Benchmark is four short tasks in a browser — useful for self-discovery and comparison, not for diagnosis. Memory varies hour-to-hour with sleep, stress, mood, caffeine, time of day, and many other factors. For a more stable picture, run the battery two or three times across different days and look at the pattern, not the single number.

Chain (digit span) tests how many digits you can hold in working memory in order — the classic Miller 7±2 capacity. Trail (visual sequence) is the visuospatial counterpart, based on Corsi 1972. Wordlist tests verbal recognition memory with a short interference delay (Sternberg 1966 / Brown-Peterson 1958). Twin (N-back, n=2) tests working memory under continuous updating (Kirchner 1958). Together they cover all three components of Baddeley & Hitch's working-memory model.

We custom-built an on-screen number pad for the Chain sub-test specifically because the dominant memory-test site triggers the native OS keyboard, which covers half the screen during recall. That's the single biggest UX flaw in the category, and we close it. Other sub-tests are tap-based and work the same on both platforms. Avoid switching tabs mid-test — browsers may pause timers and void the round.

That's a balanced profile — either the Polymath shape (balanced-high) or the Dreamer shape (balanced-low) depending on where the composite sits. A flat profile is informative on its own: it tells you your memory works as a general-purpose system rather than as a set of specialized subsystems, and it predicts how you'll handle novel learning situations across formats.

Direct practice on a specific memory task yields some improvement that stabilizes after a few sessions and is largely task-specific (you get better at digit span without getting better at face recognition). Sleep, hydration, and ambient stress have much larger same-session effects. We do not claim that this test 'trains your brain' or improves real-world memory, and you should be skeptical of products that do — the science on transfer effects from memory-training apps to everyday cognition is mixed at best.

Because those framings overstate what a five-minute browser test can measure. 'Brain age' specifically is adjacent to language that the US Federal Trade Commission has policed in this product category. We report your raw scores and a percentile against published academic norms — that's honest. The archetype is a shareable interpretive frame; it isn't a diagnosis.

Free recall (saying or typing the words you remember) is fairly described as the gold-standard test in clinical neuropsychology — but it's a mobile UX nightmare (free typing on phones is slow and frustrating). Recognition memory (tapping seen words from a pool of 30) gives a robust signal-detection score (hits minus false alarms) that doesn't penalize slow typing. The Sternberg 1966 paradigm we use is well-validated and academically standard.

The four sub-tests in Memory Benchmark are built from the same public-domain paradigms that underlie those clinical instruments (Wechsler-lineage digit span, Corsi block-tapping, Sternberg recognition, Kirchner n-back). However, this is not a clinical test — clinical instruments come with standardized administration manuals, normed against specific populations, and interpreted by trained psychologists. Memory Benchmark is a playful self-discovery battery; we use the same paradigm shapes but our own UI, our own item generation, and our own scoring presentation. Treat your result as a conversation-starter, not as clinical assessment. If you have real concerns about memory function, see a qualified clinician.