Signal but not noise changes with perceptual learning.

Area of Science:

• Neuroscience
• Cognitive Psychology
• Perception

Background:

• Perceptual learning, the improvement of sensory discrimination with practice, is often stimulus-specific.
• Changes in cortical sensory neurons are implicated, but the link to behavior remains unclear.
• Theories propose learning enhances signal, reduces noise, or both.

Purpose of the Study:

• To differentiate between signal enhancement and noise reduction as mechanisms of perceptual learning.
• To investigate how practice-specific changes in sensory processing relate to behavioral improvements.

Main Methods:

• Developed stimuli (faces, textures) with adjustable signal strength embedded in noise.
• Trained observers on identification tasks over multiple sessions.
• Assessed performance changes and response consistency using an ideal discriminator model and double-pass technique.

Main Results:

• Observer performance improved by up to 400% with training.
• Learning increased the efficiency of encoding task-relevant information.
• Response consistency did not change, indicating no reduction in internal noise.

Conclusions:

• Perceptual learning enhances discrimination primarily by increasing signal strength.
• Findings support signal enhancement as a key mechanism in perceptual learning.
• Results provide constraints for computational models of learning and sensory processing.