What are the technical properties of the UFOV® assessment?

Presentation time tolerance

The UFOV assessment uses the computer’s internal clock to make sure that the images are presented accurately. Any trial that is outside the display tolerance is repeated until it meets the tolerance limitations.

Tolerance: +/- 5% of requested presentation time with a minimum of 2 ms.

If more than 50% of the presentations fail to meet the tolerance level, a warning will be shown. Inaccurate presentations slow down the assessment and if this warning is shown it is recommended that steps are taken to increase computer performance. The most common causes of a slow computer are that the browser is having a problem or the computer is doing something else in the background. Here are some things to do to troubleshoot.

• Close the browser and reopen it
• Only have one browser window open with the UFOV assessment
• Close all other programs
• Check to see if a back-up or automatic update is running and cancel it so that runs later
• If all else fails, restart the computer

The following warning is given if 50% or less of the presentations have been accepted (the calculation is performed for every 5 rejected presentations). The user has the option to Continue or to Exit. This option is required to avoid the possibility that the user will continue responding to trials without the test ever ending. If you need additional advice, please contact our support team at support@BrainHQ.com. 

Adaptive staircase algorithm

The assessments use a two-step size staircase algorithm to estimate thresholds. A "step" refers to a change in difficulty by either making the trial display time longer or shorter. To increase the difficulty the display time is decreased and to decrease difficulty, display time is increased. The algorithm starts at a 50 milliseconds (ms) step size and switches to 17 ms steps on the first reversal (first incorrect response) which generally occurs when the stimulus is below threshold. Bigger initial steps help home in on the user threshold quickly while the smaller steps allow for a more accurate final estimate. If the very first response is incorrect then that trial is ignored.

The assessment ends early if there are three consecutive trials at fastest (17 ms) or at slowest (500 ms) presentation times. To allow a greater chance for poor performers to be able to perform the tasks, the algorithm only exits early if at least 10 trials have been done.