Sequence Memory Test

Following a series of flashes requires remembering both the locations and their order after the display goes dark. A visual sequence or block-tapping task examines this brief retention of spatial information in working memory. Remembering all the locations is not enough if their order changes. An order error can leave every square familiar while the recalled sequence is still wrong.

Digit span asks a related question using the identities and order of numbers. Spatial span uses positions instead, so the arrangement of locations and the route between them become part of the task. The two exercises give different kinds of practice, and a score from one cannot simply replace a score from the other.

Forward recall follows the sequence from its first location to its last. Backward recall starts with the last location and works back to the first, requiring the remembered order to be reversed. Both involve the same locations, but the response requirement changes. Their results are best kept separate rather than treated as interchangeable scores.

A commonly reported span is the longest sequence reproduced correctly under the task's rules. It describes a successful performance in that session. Starting length, the number of chances at each length and the stopping rule all affect which lengths get tested. Reaching a particular span does not establish an absolute limit on what someone could remember on another attempt.

Equal sequence lengths can still make different demands. Several conditions matter when interpreting a change in span:

• Layout and paths. The number and arrangement of possible locations change the spatial task. Some paths have memorable structure, while others cross or change direction in less familiar ways.
• Presentation timing. Flash duration and the gaps between locations determine the sequence's pace. Changing that pace changes the practice conditions; it does not provide a fixed conversion from one score to another.
• Grouping strategies. Remembering a short path or shape as a group is a form of chunking. It can change how a sequence is retained, so using a new strategy can change the result even when the sequence length stays the same.

Practice can improve performance on a block-tapping task. Fatigue and distraction can also affect working-memory performance, so a later score can reflect both practice and the conditions of that session. For self-comparison, similar conditions and a consistent approach make the results more useful. Improvement establishes better performance on the exercise; it does not by itself establish a wider change in everyday memory.

An informal visual sequence exercise provides practice and personal comparison. Its score alone cannot diagnose memory problems or establish a general level of cognitive ability, and it should not be treated as a clinical assessment.

How to Use This Tool:

Choose a recall direction before starting. The standard setup uses Forward recall on a 4×4 board, a starting length of 2, 1000 ms per tile and one allowed miss at each length.

1. Choose Forward to repeat the order shown, or Backward to begin with the last tile shown.
2. Open Advanced if you want a different board, timing, starting length or miss allowance. Leaving the numeric overrides blank keeps their standard values.
   If Check settings appears, correct any value outside the allowed whole-number ranges or clear its override. Clearing an override restores its standard value and allows Start once all settings are valid.
3. Select Start and watch the fixation mark followed by the tile flashes. Wait for Recall before entering an answer. The settings stay fixed for that session.
4. Tap or click the tiles in the required order, then select Submit. Recall has no time limit. Undo removes the last tap, and an incomplete answer can be submitted after at least one tap. Filling every position does not submit automatically.
5. Select Next after feedback to continue. Use Stop to finish an active, unpaused session early and keep its submitted results.
   If presentation is interrupted and Voided · paused appears, select Continue for a fresh sequence at the same length. The void does not use an attempt.
6. Read the span and ending message under Last session. Check the Trial ledger for the attempt that ended the run before comparing it with another session.

With the board area focused, arrow keys move between tiles, Space selects a tile, Backspace undoes and Enter submits. A fresh Enter or Space press can start or advance a sequence when the corresponding action is available.

Interpreting Results:

The headline span is the longest sequence recalled exactly in the required direction. Correct / counted describes whole submitted trials, not the proportion of individual tiles remembered. A response with one wrong tile is a missed trial even when its other locations are correct. These results have no age-based benchmark or population ranking.

• No exact span yet means no tested sequence was recalled completely. It is not a zero-memory score. If the starting length was high, shorter lengths remain untested.
• Ceiling reached means the board's maximum test length was recalled. No longer sequence was tested, so the result cannot locate an upper personal limit.
• Stopped early: partial session preserves earlier submitted results. An unfinished trial is voided, and the session has not necessarily reached the length at which recall would fail.

The Length vs correct chart shows counts of successful and missed trials at each tested length. Voids are excluded. The Board error map and Recall positions help locate mismatches, but a single session may expose some tiles or positions much more often than others. More recorded errors at one location do not by themselves establish a persistent weakness.

For a useful Session comparison, match direction, board size, time per tile, starting length and misses allowed. Keep the device, input method and surroundings similar too. Read the stopping reason alongside the span, and compare several sessions with matching settings before treating a small change as a consistent pattern.

Technical Details:

Spatial sequence scoring has two separate jobs: deciding whether an entire recall is exact, and describing how an incorrect answer differs from the expected sequence. Only exact recall establishes a span. Error alignment explains a response without awarding partial credit.

Formula Core

Span is the greatest sequence length among the counted trials recalled exactly:

Here, S is span in tiles, C is the set of exactly correct counted trials, and L_i is the sequence length of trial i. If C is empty, span remains unestablished. All lengths and trial counts are whole numbers, with no rounding or partial-tile score.

For forward recall, the expected answer is the displayed sequence. For backward recall, the expected answer is that complete sequence reversed. An answer passes only when its length and every tile in order match the expected answer.

Progression and stopping

Misses are counted separately at each length. Let M be the chosen miss allowance and m the counted misses at the current length. A retry remains available while m is at most M; the next miss ends the session when m exceeds M.

An allowance of 0 ends the session on the first miss. Allowances of 1, 2 and 3 end it on the second, third and fourth misses at that length respectively. Exact recall advances immediately, even if retries remain. Voids stay in the ledger but are excluded from counted trials and all error tallies.

Board limits and sequence construction

The number of available locations and the longest tested sequence are different quantities on the largest board. Starting length must be a whole number from 2 through the selected ceiling.

Each displayed sequence uses a location at most once. Randomly generated paths exclude three consecutive tiles that move in the same direction by the same distance on both steps. This still permits straight portions with unequal spacing. Within a session, the next sequence differs from the immediately preceding one, including after a retry or void. These rules do not guarantee that all permitted paths are equally likely or equally memorable.

Responses can contain repeated taps, but cannot contain more taps than the current sequence length. A nonempty shorter answer can be submitted, allowing missing locations to appear explicitly in the feedback.

Presentation timing

Every sequence begins with a scheduled 500 ms fixation. Time per tile is the interval from one tile's onset to the next, including a fixed 200 ms dark gap. Whole-number intervals from 400 to 3000 ms are accepted, including both endpoints.

T is the selected interval in milliseconds and t_lit is the scheduled lit duration in milliseconds. At the standard 1000 ms interval, the tile is scheduled to stay lit for 800 ms, followed by 200 ms dark. At the minimum interval, those durations are 200 ms and 200 ms. Recall starts after the final dark gap and is untimed.

Hiding the tab, losing window focus or skipping a scheduled presentation phase during fixation or flashing voids that trial. These checks apply to presentation, not the untimed recall period. They detect interruptions but do not measure actual screen illumination or calibrate display timing, so the selected milliseconds are requested intervals rather than verified physical measurements.

How recall errors are aligned

Feedback compares the expected order with the submitted order using optimal string alignment, or OSA. It chooses a lowest-cost alignment in which the same sequence segment is not edited more than once. This allows later correct locations to stay aligned after an earlier omission or extra tap.

For equally short explanations, the comparison favours a correct match, then an adjacent swap, substitution, omission and extra tap, in that order. An adjacent swap costs one edit but marks two recall positions as transposed. Alignment retains every submitted tap and expected location; it describes a mismatch without determining whether forgetting, a mistaken tap or another cause produced it.

A near miss is a substitution into an immediately neighbouring tile, including a diagonal neighbour. Near misses are already included in substitutions and still make the trial a miss. Extras before places an extra tap before the next expected recall position; a position beyond the sequence means after its final tile. In backward recall, position 1 refers to the last tile shown.

Location counts and recall-position counts

Coordinates use letters from left to right and row numbers from top to bottom, so A1 is the upper-left tile. They identify locations in feedback and records; the presentation and recall board does not print these coordinates on its tiles.

• Seen counts how often a tile was expected in counted trials.
• Miss counts expected appearances at that tile that were omitted, substituted or transposed.
• Wrong counts incorrect actual taps on that tile, including extra taps and taps involved in a swap.

A substitution can therefore add one Miss at the expected location and one Wrong at the tapped location. Adding those columns would count the same mismatch twice. Recall-position counts instead group events by their place in the required order. Tested equals Correct plus Substituted plus Omitted plus Transposed; near misses and extras are not additional tested positions.

Privacy Notes:

Recall and scoring happen in your browser. Sequences and answers are not sent to a scoring service, and taking the test requires no camera, microphone or location permission. Page and chart resources may still load over the network.

• The most recent 20 completed or stopped session summaries are kept in this browser when storage is available. If saving fails, comparisons remain in the current tab and the storage message explains the problem.
• Detailed trial records are not restored from saved summaries. Starting a new session or reloading loses the current detailed record. Download session JSON can retain it after completion or Stop, including the settings, sequences, answers, results and random seed.
• The page address can retain setup choices, but does not contain trial answers or results. Sharing it does not share a completed session.
• Clear saved sessions removes this test's saved summaries after confirmation and leaves the current result visible. A storage error can prevent the saved data from being cleared, so check the resulting message.

Copy actions write the selected result to the clipboard when requested. Downloaded records remain wherever you save them, independently of the browser's saved-session history.

Worked Examples:

References:

• eCorsi: implementation and testing of the Corsi block-tapping task for digital tablets, Frontiers in Psychology, 2014.
• Spatio-Temporal Structure, Path Characteristics, and Perceptual Grouping in Immediate Serial Spatial Recall, Frontiers in Psychology, 2016.
• Efficacy of Corsi Block Tapping Task training for improving visuospatial skills: a non-randomized two-group study, Experimental Brain Research, 2022.
• The Role of Motivation as a Factor in Mental Fatigue, Human Factors, 2019.