Single point estimation of a decision space.

Area of Science:

• Cognitive psychology
• Decision science
• Psychophysics

Background:

• Signal detection theory (SDT) provides measures of sensitivity and bias for discriminating signals from noise.
• $d^{\text{'}}$ is a standard sensitivity measure but is unstable with unequal variance Gaussian distributions.
• Existing alternatives for unequal variances often need multi-confidence level data or make prior assumptions.

Purpose of the Study:

• To propose an optimization approach for estimating decision space characteristics.
• To overcome limitations of existing sensitivity metrics in SDT.
• To provide a stable measure of sensitivity without multi-confidence level data.

Main Methods:

• Developed an optimization approach using a single false-alarm and hit rate pair.
• The method requires the bias parameter ($\beta$) for accurate estimation.
• Simulations were used to validate the approach's performance.

Main Results:

• The optimization approach accurately estimates decision space characteristics.
• It successfully recovers critical decision space parameters under specific conditions.
• The method is effective when $\beta$ is known or can be reasonably estimated.

Conclusions:

• The proposed optimization method offers a robust way to assess sensitivity in SDT.
• It provides a normative model for performance analysis within a decision-making framework.
• This approach enhances the understanding of observer performance in complex decision tasks.