Attention
Contrast Sensitivity Test
Also called: contrast sensitivity function, CSF, Gabor contrast threshold, grating contrast sensitivity
Measure contrast sensitivity on the participant's own screen: tilted Gabor patches at five spatial frequencies, each with its own QUEST staircase, after a card and blind-spot calibration.
8-10 minUntimed
Free to start: 10 free participants per account, then $0.99 per participant for academic research, or any yearly plan. Preview any test free after you sign up.
- Papers, 2021 to 2025
- 463
- Evidence grade
- A
- Registered clinical studies
- 119
What it measures
A research measure of contrast sensitivity on the participant's own screen. After a short screen calibration, faint striped patches tilt left or right at five stripe widths, and participants press an arrow key for the tilt. A staircase for each stripe width finds the faintest patch the person sees.
Constructs
- Contrast sensitivity
- Spatial vision
- Visual perception
Research uses
- Vision and perception research
- Within-study group comparison
- Individual differences in contrast sensitivity
How the paradigm works
The contrast sensitivity test is a research measure of how faint a pattern can be and still be seen, on the participant's own screen. Contrast sensitivity is one over the lowest contrast at which a pattern is seen. Measured from coarse to fine stripes, it traces the contrast sensitivity function, which Campbell and Robson (1968) measured with sine-wave gratings. Here patches of stripes in a soft round window (Gabor patches) tilt 45 degrees left or right on a mid-gray screen, and participants press the left or right arrow key for the tilt.
More about the paradigm
Each of five spatial frequencies, 0.75, 1.5, 3, 6 and 12 cycles per degree, has its own QUEST staircase (Watson and Pelli, 1983), and the staircases are interleaved in a seeded order, as Pelli and Bex (2013) describe for measuring contrast sensitivity. Each patch is shown at the most likely threshold given the answers so far, and the threshold at 81.6% correct is estimated at the end by maximum likelihood. Contrasts finer than one step of an 8-bit display are drawn with the noisy-bit method (Allard and Faubert, 2008), with the screen assumed to follow the sRGB curve.
The patch sizes come from a screen calibration. The participant matches a box on the screen to a bank card, which gives the pixels per millimeter, and the viewing distance comes from the blind spot of the left eye (Li and colleagues, 2020) or from a distance the researcher sets. A frequency the screen cannot draw with at least 4 pixels per cycle is skipped and recorded. The primary score, log contrast sensitivity at 3 cycles per degree, lies near the peak of the curve and can be drawn on any screen the test allows, so it compares across screens.
Scores
- Log Contrast Sensitivity at 3 c/degscore
- Log10 of one over the Michelson contrast seen at 81.6% correct at 3 cycles per degree, from that frequency's answers by maximum likelihood.
- Higher means fainter patches were seen
- Area Under the Log CSFscore
- Area under log contrast sensitivity over log10 spatial frequency from 0.75 to 6 cycles per degree, by the trapezoid rule. Null unless all four frequencies have a threshold.
- Higher means better sensitivity across frequencies
- Log Contrast Sensitivity at 0.75 c/degscore
- Log contrast sensitivity at 0.75 cycles per degree.
- Higher means fainter patches were seen
- Log Contrast Sensitivity at 1.5 c/degscore
- Log contrast sensitivity at 1.5 cycles per degree.
- Higher means fainter patches were seen
- Log Contrast Sensitivity at 6 c/degscore
- Log contrast sensitivity at 6 cycles per degree.
- Higher means fainter patches were seen
- Log Contrast Sensitivity at 12 c/degscore
- Log contrast sensitivity at 12 cycles per degree. Null when the screen cannot draw that frequency with 4 pixels per cycle.
- Higher means fainter patches were seen
- Peak Log Contrast Sensitivityscore
- Highest log contrast sensitivity measured at any frequency.
- Higher means fainter patches were seen
- Peak Frequencyc/deg
- Spatial frequency of the highest log contrast sensitivity.
- Informational
- Staircase Reversals by Frequencycount
- Changes of direction in each staircase's contrasts, keyed by cycles per degree.
- Informational
- Frequencies Showncount
- Spatial frequencies the screen drew with at least 4 pixels per cycle.
- Five on fine screens
- Frequencies Droppedcount
- Spatial frequencies left out because the screen could not draw them with 4 pixels per cycle.
- Informational
- Accuracy%
- Share correct over answers at or after 100 ms.
- The staircases keep it high
- Viewing Distancecm
- Distance from the eyes to the screen, from the blind spot or the set distance.
- Informational
- Blind-spot Pressescount
- Blind-spot presses accepted for the viewing distance. Null with a set distance.
- Five are planned
- Blind-spot Spreadscore
- Standard deviation of the blind-spot distances over their mean.
- Above 0.2 is flagged for review
- Screen Pixels per mmpx/mm
- Screen pixels per millimeter from the card match.
- Outside 2.5 to 15 is flagged for review
- Screen Pixels per Degreepx/degree
- Screen pixels in one degree of visual angle at the viewing distance.
- Informational
- Device Pixel Ratioscore
- Screen pixels per CSS pixel, as the browser reported it.
- Informational
- Dropped Frame Rate%
- Share of display frames the browser dropped while the patches showed.
- Above 0.05 is flagged for review
Use and evidence
463 papers from 2021 to 2025 name this task in their title or abstract, up 6% compared with 2016 to 2020. That ranks it 55 of 395 test families in our census. See the counts for every test.
Show the numbers
| Year | Papers |
|---|---|
| 2025 | 91 |
| 2024 | 110 |
| 2023 | 91 |
| 2022 | 85 |
| 2021 | 86 |
| 2020 | 92 |
| 2019 | 95 |
| 2018 | 85 |
| 2017 | 79 |
| 2016 | 85 |
| 2015 | 97 |
| 2014 | 102 |
| 2013 | 87 |
| 2012 | 71 |
| 2011 | 70 |
| 2010 | 124 |
| 2009 | 68 |
| 2008 | 70 |
| 2007 | 54 |
| 2006 | 46 |
| 2005 | 56 |
| 2004 | 49 |
| 2003 | 41 |
| 2002 | 53 |
| 2001 | 38 |
| 2000 | 35 |
Use grade B
300 to 999 papers from 2021 to 2025.
Evidence grade A
At least 25 papers report its test-retest reliability and at least 25 registered clinical studies use it as an outcome.
- Papers on test-retest reliability
- 81
- Papers on reliability or validity
- 204
- Registered clinical studies using it as an outcome
- 119
How the grades work
Use: A is 1,000 or more papers from 2021 to 2025, B 300 to 999, C 100 to 299, D 30 to 99 and E fewer than 30. Evidence: A means at least 25 papers report its test-retest reliability and at least 25 registered clinical studies use it as an outcome; B, at least 8 test-retest papers, and either 5 registered clinical studies or 150 reliability or validity papers; C, at least 2 test-retest papers, 30 reliability or validity papers or 3 registered clinical studies; D, little published measurement evidence found. The counts measure use. They include every paper that names the task in its title or abstract. Grades describe the published research on this task, not ConductCognition's version of it.
Source: ConductCognition test census 0.14.0, counted 31 October 2026 from OpenAlex and ClinicalTrials.gov.
Task design
The trial sequence as the test runs in ConductCognition.
- Patches
- Gabor patches, a sine grating in a Gaussian window with a standard deviation of 1 degree, tilted 45 degrees left or right on a mid-gray screen (pixel value 187).
- Spatial frequencies
- 0.75, 1.5, 3, 6 and 12 cycles per degree. A frequency the screen cannot draw with at least 4 pixels per cycle is skipped and recorded.
- Staircases
- One QUEST staircase on log contrast for each frequency, interleaved in a seeded order, with slope 3.5, guess rate 0.5, lapse rate 0.01 and a prior at 1% contrast.
- Threshold
- The contrast at 81.6% correct, by maximum likelihood from each frequency's answers.
- Contrast
- Computed in linear luminance with the screen assumed to follow the sRGB curve, and drawn with the noisy-bit method between display steps.
- Calibration
- A box matched to a bank card (85.60 by 53.98 mm) gives the pixels per millimeter. The viewing distance comes from 5 blind-spot presses at 13.5 degrees, or from a set distance (standard 57 cm; 30 to 150).
- Fixation
- 500 ms.
- Patch
- 500 ms (standard) or until the answer, in whole screen frames.
- Keys
- Left and right arrows for the tilt, with no time limit.
- Length
- 30 trials per frequency (20 to 60), up to 150 trials, with a break after every 50.
- Practice
- 6 trials with feedback at 25% contrast.
- Scores
- Log contrast sensitivity at 3 cycles per degree, which any screen the test allows can draw, and the area under the log curve from 0.75 to 6 cycles per degree.
Laboratory studies measure contrast sensitivity on a calibrated display at a fixed distance. This version runs in the participant's browser: it assumes the screen follows the sRGB curve and measures pixel size and viewing distance with a card and the blind spot, as Li and colleagues (2020) did online.
Quality checks
Each session is marked valid, needs review or invalid, with the reason, before you see its scores.
Scored rows: contrast_sensitivity_test from the scored run, left or right arrow for the tilt of the patch (chance 50%, each tilt on half the trials of each spatial frequency), with no time limit. Below chance invalidates and not above chance is flagged. One key on 90% of answers invalidates when accuracy is not above chance. At least half the answers under 200 ms invalidates. A card match outside 2.5 to 15 screen pixels per millimeter, a blind-spot distance outside 25 to 100 cm, no viewing distance, or no threshold at 3 cycles per degree is flagged for review and withholds the percentile. Blind-spot distances that vary by more than 20% of their mean, or more than 5% of display frames dropped, are flagged for review.
Why these rules
Accuracy significantly below chance (one-sided exact test, p < .05) cannot come from guessing; it means the response mapping was reversed or the task was not followed (Slick, Sherman and Iverson 1999 treat below-chance forced-choice performance as definite response bias). Chance-level accuracy can be a real result in impaired participants, so it is only flagged. One option on at least 90% of answered trials, when the correct options are balanced and accuracy is not above chance, means the answers did not depend on the items (the task counterpart of long-string responding in surveys; Curran 2016). An answer under 200 ms is faster than a simple visual reaction, so it cannot reflect the item. Following response-time effort screening for untimed tests (Wise and Kong 2005), the session is invalid only when at least half the answers are that fast. The staircases place most trials near threshold, where accuracy stays well above chance, so chance-level answers mean the patches or the keys were not followed. The patch sizes depend on the card match and the viewing distance, so an implausible calibration changes what was shown.
Running this test
Settings you can change
| Setting | Standard | Allowed |
|---|---|---|
| Trials per spatial frequency | 30 | 20 to 60 |
| Viewing distance | Measured from the blind spot (standard) | Measured from the blind spot (standard), A set distance |
| Set viewing distance (cm) | 57 | 30 to 150 |
| Patch display | 500 ms (standard) | 500 ms (standard), Until the answer |
Yearly plans can change settings within these limits. Free and pay-as-you-go studies use the standard settings.
Need a setting this test does not have?
Devices and languages
- Untimed test: runs on desktops and laptops. Participant devices
- Languages: English, Spanish, French, German, Italian, Portuguese (Brazil), Chinese (Simplified) and Japanese.
- Available on: Every plan, including the free plan.
- Recruiting online? See the setup guides for Prolific, Sona Systems, CloudResearch Connect and Qualtrics.
Cite and describe
Participants completed the Contrast Sensitivity Test in a web browser (ConductCognition task version 1.0.0, scoring version 1.0.0; Campbell and Robson, 1968; Pelli and Bex, 2013), with the standard settings (trials per spatial frequency: 30; viewing distance: Measured from the blind spot; set viewing distance (cm): 57; patch display: 500 ms). The main score was log contrast sensitivity at 3 c/deg (Log10 of one over the Michelson contrast seen at 81.6% correct at 3 cycles per degree, from that frequency's answers by maximum likelihood).
ConductCognition versions: task version 1.0.0, scoring version 1.0.0. Every export records the versions each session used, for a methods section.
- Campbell, F. W., & Robson, J. G. (1968). Application of Fourier analysis to the visibility of gratings. The Journal of Physiology, 197(3), 551-566. https://doi.org/10.1113/jphysiol.1968.sp008574
- Pelli, D. G., & Bex, P. (2013). Measuring contrast sensitivity. Vision Research, 90, 10-14. https://doi.org/10.1016/j.visres.2013.04.015Gabor patches, a forced choice and a staircase for each spatial frequency.
- Watson, A. B., & Pelli, D. G. (1983). QUEST: A Bayesian adaptive psychometric method. Perception & Psychophysics, 33(2), 113-120. https://doi.org/10.3758/BF03202828QUEST staircase and its psychometric function.
- Allard, R., & Faubert, J. (2008). The noisy-bit method for digital displays: Converting a 256 luminance resolution into a continuous resolution. Behavior Research Methods, 40(3), 735-743. https://doi.org/10.3758/BRM.40.3.735Contrasts finer than one display step.
- Li, Q., Joo, S. J., Yeatman, J. D., & Reinecke, K. (2020). Controlling for participants' viewing distance in large-scale, psychophysical online experiments using a virtual chinrest. Scientific Reports, 10, 904. https://doi.org/10.1038/s41598-019-57204-1Card and blind-spot calibration of pixel size and viewing distance.
Source papers
- Campbell FW, Robson JG (1968). Application of Fourier analysis to the visibility of gratings. Journal of Physiology, 197(3):551-566.Paradigm source, Adults
- Pelli DG, Bex P (2013). Measuring contrast sensitivity. Vision Research, 90:10-14.Paradigm source
- Watson AB, Pelli DG (1983). QUEST: a Bayesian adaptive psychometric method. Perception & Psychophysics, 33(2):113-120.Paradigm source
- Allard R, Faubert J (2008). The noisy-bit method for digital displays: converting a 256 luminance resolution into a continuous resolution. Behavior Research Methods, 40(3):735-743.Paradigm source
- Li Q, Joo SJ, Yeatman JD, Reinecke K (2020). Controlling for participants' viewing distance in large-scale, psychophysical online experiments using a virtual chinrest. Scientific Reports, 10:904.Paradigm source, Adults
Questions
How long does the Contrast Sensitivity Test take?
About 8 to 10 minutes, as listed in the test library.
What device does a participant need?
Untimed test: runs on desktops and laptops. The browser must be Chrome, Edge, Firefox or Safari.
What scores does it produce?
Log Contrast Sensitivity at 3 c/deg, Area Under the Log CSF, Log Contrast Sensitivity at 0.75 c/deg, Log Contrast Sensitivity at 1.5 c/deg, Log Contrast Sensitivity at 6 c/deg, Log Contrast Sensitivity at 12 c/deg, Peak Log Contrast Sensitivity, Peak Frequency, Staircase Reversals by Frequency, Frequencies Shown, Frequencies Dropped, Accuracy, Viewing Distance, Blind-spot Presses, Blind-spot Spread, Screen Pixels per mm, Screen Pixels per Degree, Device Pixel Ratio and Dropped Frame Rate. Each session is also marked valid, needs review or invalid, with the reason.
How widely is the Contrast Sensitivity Test used in research?
463 papers from 2021 to 2025 name it in their title or abstract, up 6% compared with 2016 to 2020. That ranks it 55 of 395 test families in the ConductCognition test census (use grade B). Source: ConductCognition test census 0.14.0, counted 31 October 2026 from OpenAlex and ClinicalTrials.gov.
How much reliability evidence does the Contrast Sensitivity Test have?
In the published research, 81 papers report its test-retest reliability and 119 registered clinical studies use it as an outcome (evidence grade A in the ConductCognition test census).
Which settings can a researcher change?
Trials per spatial frequency (20 to 60), viewing distance (Measured from the blind spot (standard), A set distance), set viewing distance (cm) (30 to 150) and patch display (500 ms (standard), Until the answer). Settings can be changed on yearly plans; free and pay-as-you-go studies use the standard settings.
Which languages can participants use?
Participants can take this test in English, Spanish, French, German, Italian, Portuguese (Brazil), Chinese (Simplified) and Japanese. English is on every plan, and the other languages come with yearly plans. Have a native speaker read the machine-translated text before launch.
Is this a clinical test?
No. ConductCognition is for research use only. It runs, scores and exports the test; clinical interpretation is outside the platform.
Every test: the full test list.
Run this test in your next study
Start free with 10 free participants on any test, then use participant credits or a yearly plan.