Skip to main content
All pymetrics games
pymetrics · Core game

pymetrics Lengths Practice: Spot the Short or Long Mouth

Practise a pymetrics-style Lengths game: a face flashes, you call its mouth short or long, and some answers win $0.20. See your bias and accuracy.

Practice: 90 faces in 3 blocks of 30, about 4 minutesAlso called “pymetrics lengths game”Also called “pymetrics faces mouth game”

How the practice module works

  1. 01A cross appears, then a face. Its mouth shows for a split second (0.1 seconds).
  2. 02Decide whether the mouth was short or long: press ← for short or → for long, or tap Short or Long.
  3. 03Some correct answers win $0.20. Other answers show nothing, and nothing tells you whether you were right.
  4. 04There are 3 blocks of 30 faces, with a short break between blocks.

Controls

  • ← for short, → for long
  • Tap the Short and Long buttons on a phone

Included in 1 Game Platform Prep (when you choose Harver), All Games Platforms Prep and Full Access. See plans

Task
A “Reward Learning Task”: cartoon faces with a short or a long mouth that differ only slightly; → means long, ← means short (pymetrics patent, official)
Rewards
Correct answers are paid on an uneven schedule: one length pays more often (pymetrics patent, official)
Screen
Some correct answers show money, about $0.20 on one prep site; other answers show nothing, with no right or wrong marks (prep sites)
Faces
60 or 90 faces, depending on the prep site; pymetrics does not publish a count (prep sites)
Research task
The published version pays the rich length 3 times as often, shows the mouth for 0.1 seconds and uses 3 blocks of 100 trials (Pizzagalli et al. 2005, published research; not confirmed for pymetrics)
Game length
Each pymetrics game takes about 1 to 3 minutes (pymetrics FAQ, official)
Practice version
Candidate Falcon’s choice: 90 faces in 3 blocks of 30, research timings and ratio, our own face, the rich length picked at random each run

What the pymetrics Lengths game looks at

Lengths is pymetrics’ version of the probabilistic reward task, a research task about how rewards shape decisions. pymetrics’ patent describes it as a reward learning task: a cartoon face appears with a short or a long mouth, the two lengths differ only slightly, and you press the right arrow for long or the left arrow for short. Correct answers are paid on an uneven schedule, so one length (researchers call it the rich one) pays more often than the other (the lean one). Prep sites add that a rewarded answer shows a small amount of money and that nothing ever tells you whether you were right.

Two things come out of the game. Response bias, usually written log b, measures how far you start answering with the rich length, especially on faces you were unsure about. Discriminability, log d, measures how well you told the two lengths apart regardless of that lean. A lean toward the rich length is the normal pattern, so the game reads more like a measure of how rewards shape your choices than a test with a pass mark. pymetrics does not publish how it scores Lengths: its games feed a model that compares your play with people who do well in the job.

Candidate Falcon’s practice uses the published research timings (a 0.1-second mouth) and its 3:1 reward ratio, with our own simple face and 90 faces in three blocks. Your results reveal which length paid more, then show your accuracy, your response bias and your discriminability, overall and block by block, in plain words.

How the Lengths practice works

  1. Press Start. A cross appears, then a face without a mouth.
  2. The mouth shows for 0.1 seconds, then disappears. Decide whether it was short or long.
  3. Press ← for short or → for long, or tap Short or Long on a phone. There is no time limit, but answer as soon as you can.
  4. A rewarded answer shows +$0.20. Every other answer, right or wrong, shows nothing.
  5. After each block of 30 faces, take the break and press Start block. After 90 faces, see the results.

Lengths tips

  • Fix your eyes on where the mouth will appear before it shows: the lower middle of the face.
  • Answer on your first impression; staring at the empty face will not bring the mouth back.
  • Expect to be unsure on many faces. The difference is meant to be hard to see.
  • Notice if one answer starts to feel “safer”. That pull is exactly what the game watches for.
  • Play on a screen you can see clearly, at normal brightness, and keep the distance steady.

Common Lengths mistakes

  • Taking “nothing happened” as “I was wrong”. Most correct answers are not rewarded either.
  • Trying to count rewards to work out a pattern instead of watching the faces.
  • Looking at the buttons or the keyboard when the mouth appears.
  • Answering only one length for a whole block when unsure.
  • Playing on a small phone screen when a computer is available; prep sites advise a keyboard.

A short Lengths practice routine

  1. Do one run without thinking about rewards, to get a baseline.
  2. Read the results: which length paid more, and how far your answers leaned toward it.
  3. In a second run, focus only on the mouth and see whether your discriminability rises.
  4. Compare the block-by-block table across runs rather than a single number.

Lengths FAQ

What is the pymetrics Lengths game?

It is a version of the probabilistic reward task. pymetrics’ patent describes cartoon faces with a short or a long mouth: you press the left arrow for short and the right arrow for long, and correct answers are paid on an uneven schedule, so one length pays more often.

Why do I only win money sometimes in the Lengths game?

Only some correct answers are rewarded, and one mouth length is rewarded more often than the other. Prep sites say other answers show nothing, so you never see whether an answer was right. That is by design: the game looks at how rewards pull your choices.

What are response bias and discriminability?

They are the two signal-detection measures used with this task. Response bias (log b) shows how far you lean toward answering with the length that pays more. Discriminability (log d) shows how well you tell the two lengths apart. Candidate Falcon’s results explain both in plain words.

How many faces are in the pymetrics Lengths game?

pymetrics does not publish a count. Prep sites report 60 or 90. The published research task uses 300 trials in three blocks. Candidate Falcon’s practice uses 90 faces in three blocks of 30.

Is this the real pymetrics game?

No. It is independent practice by Candidate Falcon, built from the task in pymetrics’ public patent, the published research and prep-site descriptions, with an original face and screens. It is not affiliated with pymetrics or Harver.

Independent practice by Candidate Falcon, based on pymetrics’ public patents and research and on prep-site descriptions. Not affiliated with pymetrics or Harver, and not their scoring. The game, its art and its wording are original, and practice results are not pymetrics scores. For all 12 games, timing, results and retakes, read the free pymetrics test guide.