Measuring performance efficiency and consistency in visual discriminations with noisy images

Area of Science:

• Visual perception
• Human factors engineering

Background:

• Assessing observer performance efficiency is crucial for understanding human-computer interaction and visual task design.
• Quantifying observer consistency provides insights into the reliability of perceptual judgments under noisy conditions.

Purpose of the Study:

• To measure the efficiency of disk discrimination performance relative to an ideal observer.
• To compare performance consistency across two visually dissimilar tasks with identical noisy image stimuli.
• To investigate the impact of disk size on observer efficiency and consistency.

Main Methods:

• Experiments measured disk discrimination performance efficiency against an ideal observer model.
• Performance consistency was assessed by analyzing correlations in observer judgments across repeated trials with identical stimuli.
• Two distinct visual tasks were employed using identical noisy image stimuli.

Main Results:

• Larger disk sizes in stimuli reduced observer performance efficiency by a factor of 10 in both tasks.
• Even at very low efficiency levels (≤5%), performance consistency remained high (around 50%).
• Approximately half of the observed inefficiency was attributed to consistent, albeit suboptimal, perceptual misinterpretations of stimulus information.

Conclusions:

• Disk size is a critical factor negatively impacting visual discrimination efficiency.
• High performance consistency despite low efficiency suggests systematic biases or errors in perceptual processing.
• Understanding these consistent perceptual miscalculations is key to improving visual task design and human performance models.