Adding noise can reduce response biases in addition to improving perceptual performance

Area of Science:

• Cognitive Psychology
• Neuroscience
• Perception

Background:

• Stochastic resonance (SR) is a phenomenon where optimal noise enhances sensory system performance.
• Reducing response biases is crucial in tasks involving infrequent stimuli, like medical diagnoses.
• Previous research has primarily focused on SR's performance enhancement, not bias reduction.

Purpose of the Study:

• To investigate if stochastic resonance (SR) can simultaneously improve perceptual accuracy and reduce response bias.
• To quantify the effects of controlled visual noise on performance and criterion shifts in a multi-alternative forced-choice task.
• To explore the potential of SR as a strategy for tasks with rare targets.

Main Methods:

• A visual perceptual task was designed with six possible letters presented at threshold levels.
• Response bias was induced by manipulating stimulus frequency (80% frequent, 20% rare).
• Stochastic resonance was induced using Gaussian visual luminance noise, with performance measured by accuracy and d'.

Main Results:

• SR led to improved accuracy and d' at low to intermediate noise levels across 22 observers.
• Response biases (criterion) were significantly reduced.
• The reduction in bias was associated with improved identification of rare stimuli.

Conclusions:

• Optimal levels of noise, via stochastic resonance, can enhance visual perception.
• SR effectively reduces response biases, particularly for infrequent stimuli.
• These findings suggest potential applications in real-world scenarios like medical screening for rare conditions.