Timing Tasks

What you see

The screen shows a round dial with a short mark at the top, a hand, and a large Hold button beneath it; in the first part a shaded wedge sits on the dial's left side. You respond by pressing and holding the button with your pointer or finger, or by holding the Space bar. The time from press to release is your answer. A counter above the dial shows which interval you are on, such as Item 2 / 6.

One trial has two parts, always in this order.

Part 1: copy from the dial. The dial demonstrates an interval. The hand starts at the top mark and sweeps clockwise three-quarters of the way round, leaving a filled trail, and arrives in the middle of the wedge exactly when the interval ends. The wedge brightens as the hand passes through it, because the wedge spans 10% either side of the interval. The finished dial is held for a moment and then resets. As soon as the sweep finishes, the Hold button becomes active: press and hold for as long as the sweep took, then release. The next demonstration starts on its own.

Part 2: stated intervals. There is no demonstration. The question names a duration, such as 3 seconds, and the still dial shows it in its centre. Hold the button for that long and release.

Two details matter. First, the hand does not move while you hold. The button changes to read Holding, but there is no sweep to watch; you are timing from memory. Second, a hold ends the instant you release, and it also ends if your pointer slides off the button or the window loses focus. Keep your finger planted.

What changes with level

The demonstrated intervals are random to the millisecond, so you will see values like 2.44 s rather than round numbers. Each set always includes at least one interval from the bottom third of the level's range and one from the top third, and two intervals in a row never fall within 10% of each other. So every trial mixes short and long, and you can never simply repeat your last hold. Stated durations are picked at random from the list for that level, and the same one can come up twice.

What it measures

Timing Tasks feeds the Perceptual precision facet of the Attention dimension, alongside Lengths. It is scored mainly for processing speed, with a secondary weight on working memory, because in the first part you have to hold the demonstrated interval in mind while you produce it.

The two parts are the classic reproduction and production tasks of timing research. In reproduction you experience an interval and then recreate it. In production you are given a duration in conventional units and must generate it. Wearden and Lejeune's review groups both among the classical human timing tasks.

The scalar property

The best-known account of interval timing is scalar expectancy theory, introduced by Gibbon and usually described as an internal clock: a pacemaker emits pulses, an accumulator counts them, and the count is compared with a remembered count for the target duration. Its signature prediction is the scalar property: the spread of your timing errors grows in proportion to the duration being timed. It is Weber's law applied to time.

Scalar property

The finding that the variability of timed responses is proportional to the duration being timed, so the relative error stays roughly constant whether you time 1 second or 8.

Reviews find the scalar property in many studies, alongside some systematic violations. Jazayeri and Shadlen's reproduction experiment is a clean example: for all their subjects, the standard deviation of the reproduced times was roughly proportional to the mean.

That is why this game scores in percentages, not milliseconds. A hold 300 ms off a 1-second interval is a 30% error; the same 300 ms on a 6-second interval is only 5%.

Regression to the mean

Jazayeri and Shadlen also found a systematic bias: reproduced times gravitate toward the average of the intervals in the set. Short intervals get reproduced too long and long ones too short. The effect is old; it is associated with Karl von Vierordt's 1868 studies of the time sense. Their model explains it as the brain sensibly combining a noisy measurement with its knowledge of which intervals are likely, which makes average error smaller at the price of a predictable bias.

Counting and attention

Two more findings shape the method below.

• Counting changes the rules. Hinton and Rao asked participants to reproduce intervals of 8, 16 and 24 seconds either by timing or by counting. Accuracy was similar, but timing variability showed the scalar property and counting variability did not; it grew more slowly with duration.
• Attention feeds the clock. In the attentional-gate model reviewed by Block and Gruber, prospective timing (when you know you will have to judge a duration) depends on attending to time. In that model, attention pulled elsewhere lets pulses slip past the gate, so the interval seems shorter than it is.

How it is scored here

Each hold is scored against its target by relative error, the size of the miss as a share of the target:

• within 10%: full credit (1.0)
• 50% or more: no credit (0)
• in between: credit falls in a straight line, so a 30% miss earns 0.5

In formula form, for an error e between 10% and 50%, credit = 1 − (e − 0.10) ÷ 0.40.

The trial score is the average credit across every hold in both parts. Holding too long and too short are treated alike.

Difficulty is adaptive. If every hold is within 10%, the trial is perfect and you move up one level. If some credit is earned but not all, you stay at your level. If every hold misses by 50% or more, you drop a level. Your 0 to 100 score compares your trial scores with what the level you work at predicts; 50 means you are doing what that level predicts.

Your review also reports the average signed error: late holds count as positive, early ones as negative. If the average is more than 150 ms either way, you are labelled as running long or running short; otherwise you are labelled balanced. The session report adds a headline, Early or late, and a measure of how consistent your timing was.

Where you will meet it

Pure duration reproduction is rare in graduate hiring. Timing in the broader sense, stopping or responding at exactly the right moment, is more common, and it draws on the same internal clock.

• Arctic Shores. Arctic Shores' own site describes a task-based assessment of about 45 minutes. Candidate guides describe two timing-flavoured tasks in it: a security door task where you stop a rotating dial on the correct number, and a rocket launcher task where you land a moving needle, which changes speed and rhythm, in the right area. These are timing-to-a-target tasks rather than duration reproduction: close cousins, not the same task.
• Schuhfried's Time/Movement Anticipation test (ZBA). A ball moves across the screen and then disappears; you press a button when you think it reaches a target line, and then indicate where it would cross. It measures how well you extrapolate time and motion, and Schuhfried lists personnel selection in manufacturing and safety among its uses.

The method

1. Calibrate a count during the demonstrations

   In the first part, count silently at a steady pace while the hand sweeps, for example one-and-two-and-three. Note where in your count the hand reached the wedge. Every demonstration is a free calibration of your counting speed against real seconds.

2. Hold for the same count, not the same feeling

   When you press, restart the same count at the same pace and release on the count where the sweep ended. In Hinton and Rao's study, the variability of counted intervals did not grow in proportion to duration the way timing variability did, which favours counting over longer intervals.

3. Use your calibration on the stated intervals

   If your demonstrations taught you that your counts run at about one per second, a stated 4 seconds means release on four. If they ran fast, add a beat.

4. For very short intervals, rely on a single beat

   Below about a second and a half, there is no room for a count. Treat the interval as one movement: press, feel one beat, release. Here a slip of 80 ms is already 10% of a 0.8-second interval, so a smooth release matters more than anything else.

5. Correct for the pull toward the middle

   Expect to run long on the shortest intervals in a set and short on the longest. When an interval is clearly the shortest you have seen, release a touch earlier than feels natural; when it is clearly the longest, hold a touch longer.

6. Keep your attention on the count

   Do not look around the screen or think about your last hold while holding. In attentional-gate terms, attention pulled away from time lets pulses slip by, which pushes you toward holding too long.

Worked example 1: the tolerance band

The dial demonstrates an interval of 2.4 seconds. What range of holds earns full credit, where does credit reach zero, and what do you score if you hold for 2.75 seconds?

Show the solution

Full credit: within 10% of 2.4 s, so from 2.16 s to 2.64 s. That is a window of just under half a second.

Zero credit: a miss of 50% or more, so at or below 1.2 s and at or above 3.6 s.

Your hold: 2.75 − 2.4 = 0.35 s late. Relative error = 0.35 ÷ 2.4 = 14.6%. Credit = 1 − (0.146 − 0.10) ÷ 0.40 = 1 − 0.115 = 0.885.

Being 110 ms past the edge of the band cost you about 11.5% of that hold's credit, and it also means this trial cannot be perfect, so you will not move up a level from it.

Worked example 2: a balanced label that hides a bias

At level 5 you reproduce four intervals, 1.3, 4.6, 2.2 and 3.8 seconds, holding for 1.55, 4.1, 2.3 and 3.7 seconds. Then you produce 3 seconds and 5 seconds, holding for 3.2 and 4.4 seconds. What is your trial score, what bias label do you get, and what does the pattern tell you?

Show the solution

Target	Hold	Miss	Credit
1.3 s	1.55 s	+19.2%	0.769
4.6 s	4.1 s	−10.9%	0.978
2.2 s	2.3 s	+4.5%	1.000
3.8 s	3.7 s	−2.6%	1.000
3 s	3.2 s	+6.7%	1.000
5 s	4.4 s	−12.0%	0.950

Score = (0.769 + 0.978 + 1 + 1 + 1 + 0.950) ÷ 6 = 0.95, about 95%. Not every hold was within 10%, so you stay at level 5.

Bias label. Signed errors are +250, −500, +100, −100, +200 and −600 ms, which average about −108 ms. That is inside ±150 ms, so you are labelled balanced.

The pattern. Balanced overall hides a clear tilt. Your shortest target (1.3 s) ran 19% long, and your two longest (4.6 s and 5 s) ran 11 to 12% short. That is exactly the pull toward the middle that Jazayeri and Shadlen describe. The errors cancelled in the average, but they cost you three holds' worth of credit. The fix is the correction step in the method: release early on the clearly short ones, hold on for the clearly long ones, and use a calibrated count for anything over about two seconds.

Tips and traps

Levelling up

A perfect trial needs every single hold within 10%, and the number of holds grows from 3 at level 1 to 10 at level 10. At the same time the range widens at both ends: demonstrated intervals go down to 0.8 seconds and up to 8 seconds, and stated durations of 2, 3, 4 and 5 seconds can all appear from level 7. The pull toward the middle gets stronger the wider the range, because the shortest and longest intervals sit furthest from the average.

The adaptations that pay off most at higher levels are:

• A stable count. From level 5, intervals reach 5 seconds and beyond. Pure feel drifts; a calibrated count drifts less.
• Deliberate correction at the extremes. Know which interval in the set was the shortest and longest and adjust.
• Clean motor habits for the shortest intervals. At 0.8 to 1 second, your finger's release matters as much as your clock.

Be realistic about what practice does. In Wright and colleagues' study of timing discrimination, ten days of training at a 100 ms interval improved performance at that interval but did not carry over to 50, 200 or 500 ms. Improvement can be specific to the durations you practise, which is one reason to work across the whole range rather than only the intervals you find easy. On a more encouraging note, Meegan and colleagues found that training on a perceptual timing task transferred to a motor timing task, so judging and producing durations seem to draw on shared machinery.

1. Week 1: build the count
   • Three sessions of about 10 minutes at your natural level
   • Count along with every demonstration and note where the hand lands in your count
   • After each session, write down your average count per second
2. Week 2: fight the middle
   • Three sessions; before each hold, ask whether this is one of the shortest or longest intervals so far
   • Adjust early on the shortest and late on the longest
   • Check the Early or late headline and compare it with week 1
3. Week 3: full range
   • Three sessions at your highest stable level
   • Aim for zero holds outside 20% before chasing 10%
   • Add one Lengths session to train the other half of the Perceptual precision facet

Check yourself

5 practice questions

1. The target is 5 seconds. Which hold earns full credit?
2. You hold for 3.9 seconds on a 3-second target. What credit does that hold earn?
3. What does the scalar property of timing predict?

and 2 more

Create a free account to answerSign in

Sources

1. Gibbon, J. (1977). Scalar expectancy theory and Weber's law in animal timing. Psychological Review, 84(3), 279–325. doi:10.1037/0033-295X.84.3.279
2. Wearden, J. H., & Lejeune, H. (2008). Scalar properties in human timing: Conformity and violations. Quarterly Journal of Experimental Psychology, 61(4), 569–587. doi:10.1080/17470210701282576
3. Jazayeri, M., & Shadlen, M. N. (2010). Temporal context calibrates interval timing. Nature Neuroscience, 13(8), 1020–1026. PMC2916084
4. Lejeune, H., & Wearden, J. H. (2009). Vierordt's The Experimental Study of the Time Sense (1868) and its legacy. European Journal of Cognitive Psychology, 21(6), 941–960. doi:10.1080/09541440802453006
5. Hinton, S. C., & Rao, S. M. (2004). One-thousand one... one-thousand two...: Chronometric counting violates the scalar property in interval timing. Psychonomic Bulletin & Review, 11(1), 24–30. doi:10.3758/BF03206456
6. Block, R. A., & Gruber, R. P. (2014). Time perception, attention, and memory: A selective review. Acta Psychologica, 149, 129–133. doi:10.1016/j.actpsy.2013.11.003
7. Wright, B. A., Buonomano, D. V., Mahncke, H. W., & Merzenich, M. M. (1997). Learning and generalization of auditory temporal-interval discrimination in humans. Journal of Neuroscience, 17(10), 3956–3963. PMC6573705
8. Meegan, D. V., Aslin, R. N., & Jacobs, R. A. (2000). Motor timing learned without motor training. Nature Neuroscience, 3(9), 860–862. doi:10.1038/78757
9. Arctic Shores. How it works. arcticshores.com
10. GraduatesFirst. Arctic Shores game-based assessments (candidate-reported task descriptions). graduatesfirst.com
11. Schuhfried. ZBA: Time/Movement Anticipation. Vienna Test System Marketplace. marketplace.schuhfried.com

Put it into practice: Timing TasksSign in to try it