Quick Summary

  • O-Span stands for Operation Span.

  • Automated means the task runs on a computer, scores automatically, and does not require manual calculations.

  • The task involves remembering sequences of letters while solving simple math problems.

  • The math problems are very easy.

  • There are practice blocks before the real test starts.

  • The final O-Span score ranges from 0 to 75 points.

Introduction

The Automated O-Span task is a well-known test for measuring working memory (Unsworth and colleagues, 2005). It takes about 20 minutes to complete. It is based on the original O-Span test by Turner and Engle (1989). The automated version runs on a computer and calculates the score automatically.

The task is of interest for a number of reasons. In part, it correlates with fluid intelligence. For example, in the study of Unsworth and colleagues (2005), table 2, a correlation between the automated O-span task and Raven’s matrices was reported (p=.448, p<.01).

In this task, you need to remember a sequence of letters (for example, Q, N, P, Y). This is similar to a simple memory test. However, before each letter appears, you must solve a simple math problem, such as:

3 × 2 + 1 = ?

This makes it harder to remember the letters because your attention is split between solving math problems and memorizing letters. You can try the task yourself using the further link below (note that it takes 20 minutes to complete).

ospantrial

Before the real test, there are practice blocks to help you get used to the task:

  1. A short block with only letters (sequences of 2 or 3 letters).

  2. A short block with only math problems (15 simple math problems).

  3. A practice session with both letters and math problems (6 trials).

  4. The real test (15 trials).

The number of letters in a sequence varies. Some sequences contain only 2 letters, while others contain up to 7 letters. Letters are chosen from the following set: F, H, J, K, L, N, P, Q, R, S, T, Y. A letter never repeats within a single sequence.

How well do people do?

The task is actually quite difficult. In the original study by Unsworth and colleagues (2005), they started with 296 participants aged 18 to 35 year old. They were recruited via a community newspaper advertisements but no further demographics were reported. Therefore, this is difficult to generalize to the general population.

15% of their participants were excluded from further analyses because they did not manage to have less than 15% mistakes in their math problems. In total, the task has 33 math problems (15 untimed practise trials, and then a further 18 in the practise and real block). That means you could make only 5 mistakes in any of the math tasks, which seems quite a strict criterion.

The study reports for people who are that good in the math problems managed an O-Span score of just 40 points (median score).

If you think about it, if you can manage all the easiest sequences of 3 and 4 letters, you already can get 35 points, and if you can then get one more correct in the harder ones, you are already at the average.

In the original study, there were people who could do much better. 25% of people who managed the math criterion of less than 15% mistakes scored 51 points or more.

Task Structure

The table below shows the number of trials in each block and the sequence lengths:

Block

Total Trials

Letter Sequence Lengths

Practice - Letters

4

2 or 3

Practice - Math

15

N/A

Practice - Full Task

3

2

Real Test

15

3 to 7

In the final block (real test), there are 15 trials for each sequence length (3, 4, 5, 6, or 7 letters). You earn points for each completely correct trial. A 3-letter sequence earns 3 points, a 4-letter sequence earns 4 points, and so on.

Timing of a Trial

Each trial follows these steps:

  1. A math problem appears.

    • In the practice blocks, you have unlimited time.

    • In the real test, the time limit is based on your response speed in the earlier math practice block.

  2. You press a button when you are ready to answer.

  3. A possible answer (true or false) is shown, and you must decide if it is correct. You have unlimited time for this decision.

  4. A letter appears for 800 milliseconds (ms).

  5. A new math problem appears, and the cycle repeats.

  6. After all math problems and letters are shown, you must recall the letters in order by clicking on them.

    • If you make a mistake, you can use the backspace button on the key pad.

    • If you are unsure about a letter, you can enter a question mark (?).

Implementation Details

Some details in Unsworth et al. (2005) potentially ambiguous. Professor Unsworth provided additional clarifications to PsyToolkit:

  • Timing of Math Problems: The paper says math timing is based on performance in the practice block. However, it was unclear which response was used (first response or final decision). Prof. Unsworth confirmed that only the first response is timed, while the true/false decision has unlimited time.

  • Math Timing Calculation: The time limit in the real test is based only on correct math responses in the practice block.

Other issues you might want to reflect on when using this task:

  • Timing Strategy Issue: If someone rushes through the math practice and makes errors, they may get less time in the real test, which could hurt their score. On the other hand, if someone takes a long time, they may get more time in the real test, giving them an advantage. At the same time, a timing restriction is necessary to prevent people from using the math task to extensively rely on rehearsing the memorized letter sequence.

  • Feedback Timing: The paper states that participants see trial feedback for only 2 seconds. This seems too short. A longer or self-paced feedback display would make more sense.

  • Feedback Content: The feedback for letters and math is inconsistent. Letter recall feedback shows the number of correct letters, while math feedback shows the number of errors.

  • Math Accuracy Feedback: The paper says that math accuracy is shown during letter recall, but the figures in the paper do not display this. We assume it was intended to be shown after letter recall.

Scoring

The O-Span score is based on performance in the final block (15 trials). You earn points only for completely correct trials (where both the math and letter sequence are 100% correct).

You get between 3 and 7 points for a sequence. For example, a sequence with three math problems and three letters gets you 3 points; a sequence with four math problems and four letters gets you four points; etc.

The highest possible score is 75 points.

Additionally reported scores:

  • O-Span percentage: The percentage of trials completed correctly.

  • Math accuracy: The percentage of correctly solved math problems.

  • Letter recall accuracy: The percentage of sequences remembered correctly (even if math answers were incorrect).

Run the task

This is the full official version of the task which takes 20 minutes. At the end you get your score.

Click here to run a demo

Data Output File

For each trial, there is one line.

Column

Description

1

Block number

2

Block name

3

O-Span score so far (0–75), the last line of the data file contains the total O-Span score for the participant

4

Letters correct so far (0-75), the last line of the data file contains the total letters correct score for the participant

5

Math problems correct so far (0-75), the last line of the data file contains the total math problems correct score for the participant

6

Reaction time for true/false decision

7

Reaction time for completing letter recall

8

Max allowed reaction time for math problem (when pressing "ready")

9

Percentage math problems correct in current block (this is shown to participants)

10

The trial’s set size (2-7). Set size two only in practise blocks

11

The trials' number of math problems correctly solved

12

The trials' number of math problems incorrectly solved (math errors)

13

The number of letters correctly remembered in the current trial

Check out the source code

The PsyToolkit code

Download

The PsyToolkit code zip file, including stimulus file

If you have a PsyToolkit account, you can upload the zipfile directly to your PsyToolkit account.

If you want to upload the zipfile into your PsyToolkit account, make sure the file is not automatically uncompressed (some browsers, especially Mac Safari, by default uncompress zip files). Read here how to easily deal with this.

Related tasks

The PsyToolkit experiment library also has other memory span tasks, which you can find here

References

  • Turner, M. L., & Engle, R. W. (1989). Is working memory capacity task-dependent? Journal of Memory & Language, 28, 127-154.

  • Unsworth, N., Heitz, R. P., Schrock, J. C., & Engle, R. W. (2005). An automated version of the operation span task. Behavior Research Methods, 37, 498–505. https://doi.org/10.3758/BF03192720. Link to paper