Vocal learning is constrained by the statistics of sensorimotor experience

Area of Science:

• Neuroscience
• Animal Behavior
• Auditory Perception

Background:

• The brain uses sensory feedback to correct behavioral errors, crucial for motor control.
• Vocal control learning is complex, as auditory feedback can be noisy or inaccurate.
• Existing models often assume larger errors drive greater learning, but this may not hold true.

Purpose of the Study:

• To investigate how the magnitude of sensory feedback errors influences vocal learning and error correction.
• To test the hypothesis that vocal compensation is greatest for smaller errors and least for larger errors.
• To determine if the statistics of sensorimotor experience constrain learning.

Main Methods:

• Auditory feedback pitch was manipulated in singing Bengalese finches.
• Vocal compensation and learning rates were measured for different magnitudes of pitch error.
• The relationship between error magnitude and compensation was analyzed in relation to production variability.

Main Results:

• Learning was slower and compensation less complete for larger imposed pitch errors compared to smaller errors.
• The degree of vocal compensation was proportional to the overlap between baseline pitch production and experienced pitch shifts.
• Compensation approached zero when pitch shifts fell outside the normal production range.

Conclusions:

• Vocal learning and error correction are constrained by the statistics of sensorimotor experience.
• Sensory errors drive learning most effectively when they fall within the range of normal production variability.
• The brain may disregard aberrant auditory signals that fall outside expected ranges to ensure adaptive vocal corrections.