Re: Statistical foundation of jnd and critical bandwidth

[William Hartmann <hartmann@xxxxxxxxxx>]

AM,

        For a yes/no task such as you describe, you need to find the hit
rate and the false alarm rate. Then the JND is normally defined for
a d-prime of approximately 1.0.

Best,
Bill

Andrew Milne wrote:
>
> Dear List,
>
> When terms like critical bandwidth or just-noticeable-difference are used,
> what is their statistical foundation?
>
> For instance, if I say that the JND for pitch is 5 cents (at a given
> frequency, for a sample), for what percentage of responses would I expect
> would I expect to get a "yes the pitch has changed" answer?
>
> Andy Milne