5
$\begingroup$

The traditional SART task requires key presses to all digits appearing on the screen. However, the key press should be withheld if the digit is a 3

When aggregating the collected data, what is considered an error?

Is it simply the number of times the key was pressed when a 3 is on the screen? Or does it also include incorrectly withheld key presses (i.e. when the digit is anything other than 3)?

$\endgroup$

1 Answer 1

2
$\begingroup$

Both are errors indeed. Pressing when a 3 is shown is called a false alarm, and not pressing when a three is shown is a miss. Correct responses are (1) hits (pressing when anything but a three is shown) or (2) correct rejections (not pressing with a three).

http://gregstevens.com/2012/06/28/minimum-wage-part-2-signal-detection-theory/

These classifications are used in signal detection theory (Wiki). More specifically, they are used to calculate sensitivity and bias.

Sensitivity, or d-prime (see also other questions on CogSci, e.g. Calculating d-prime), is the metric for how well one can distinguish targets from distractors. Bias, on the other hand, is:

the extent to which one response is more probable than another (Wiki)

$\endgroup$
2
  • 1
    $\begingroup$ I realise that they're both errors, but are they both used in the error calculation that is typically published? The original Robertson paper isn't explicit about this, but I get the sense they are only considering false alarms in their results $\endgroup$
    – Simon
    Commented Feb 5, 2017 at 9:13
  • $\begingroup$ I made some minor changes to my answer in hope to clarify my answer. I don't work with it myself, so I don't know of any recent literature and their practices, but I hope my answer may give you some starting point for your search. $\endgroup$ Commented Feb 5, 2017 at 9:59

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.