Testing Visual Functions in Patients with Visual Prostheses - Artificial Sight: Basic Research, Biomedical Engineering, and Clinical Advances - page 105

Figure 5.8. Psychometric function. The rate of correct responses to a visual test in
response to increasing demands like higher spatial frequency is shown. The curve demon-
strates that there is no sharp transition between seeing and not seeing but rather an area
of uncertainty. The inflection point of the curve where the slope is highest should be
defined as the psychometric threshold.

the most sensitive point of the test. For this reason this point should be used to
determine a psychometric threshold, as it ensures a good reliability of this test.
The psychometric threshold can be estimated as the mean between guessing
probability and 100% correct responses. It can be calculated as follows:

Psychometric Threshold % = 100/n c + 100 − 100/n c /2

where n c = Number of choices

Accordingly the psychometric threshold for an eight-alternative forced choice,
like in the grating test, is 56.25%; for a four-alternative forced choice it would
be 62.5%.
As the psychometric threshold is approached, there is increasing uncertainty
and the patient has to guess. This is rather uncomfortable for the tested individual
and he will tend to give up before actually reaching the threshold. Canceling the
test at this point would lead to an underestimation of the psychometric threshold.
To avoid this, the BaLM test applies the forced choice principle, in which the
patient is required to make a decision even if uncertain. In this way the influence
of different criteria (“personalities”) can be minimized and the interindividual
reproducability of the test can be increased.

Test Battery Pillar Two: Testing Activities of Daily Living
and Orientation
Visual prostheses will be available in first clinical trials only to severely
visually impaired or completely blind subjects. These people have adopted