Late Bayesian inference in mental transformations.

Area of Science:

• Cognitive Neuroscience
• Computational Neuroscience
• Human Perception

Background:

• Human skills often involve processing noisy sensory data and performing mental computations.
• Bayesian models propose that humans use prior expectations to reduce variability from measurement noise.
• It remains unclear if similar strategies apply to noise in downstream mental and sensorimotor computations.

Purpose of the Study:

• To investigate whether humans employ biasing strategies to compensate for noise in downstream mental and sensorimotor computations.
• To determine if the complexity of mental transformations influences the use of biasing strategies.

Main Methods:

• Participants completed a battery of tasks designed to assess performance under noisy conditions.
• Tasks varied in the complexity of mental transformations required.
• Behavioral variability and perceptual biases were measured.

Main Results:

• Increased task complexity correlated with increased perceptual bias.
• Findings suggest humans mitigate noise in both sensorimotor and mental transformations.
• Behavioral data indicate a strategy of delaying inference to account for multiple noise sources.

Conclusions:

• Humans actively compensate for noise not only in sensory input but also within internal cognitive and sensorimotor processes.
• The complexity of mental operations influences the degree to which prior expectations are used to reduce uncertainty.
• This research highlights the adaptive nature of human inference in the face of pervasive noise.