Associative account of self-cognition: extended forward model and multi-layer structure

Area of Science:

• Cognitive Neuroscience
• Neuroimaging
• Psychology

Background:

• Neuroimaging studies reveal diverse neural correlates for different aspects of the self.
• Understanding the neural basis of self-representation is crucial for cognitive neuroscience.
• Existing research often focuses on specific self-related constructs without a unifying framework.

Purpose of the Study:

• To propose a neurobiologically grounded categorization of the self into three distinct types.
• To identify the neural substrates and cognitive processes supporting each self-category.
• To explore a shared computational mechanism underlying different aspects of self.

Main Methods:

• Review and synthesis of neuroimaging findings related to self-representation.
• Evaluation of cognitive processes associated with distinct self-related functions.
• Theoretical integration of findings into a hierarchical model of self.

Main Results:

• Three categories of self identified: physical (sensory-motor cortices, right hemisphere), interpersonal (dorsomedial frontal, lateral posterior cortices), and social (ventral medial prefrontal cortex, posterior cingulate cortex).
• All self-categories likely utilize a shared 'forward model' computation based on internal schemas and associative learning.
• A hierarchical structure of self is proposed, evolving developmentally and updated via prediction error attribution.

Conclusions:

• The self is not a monolithic entity but comprises distinct, hierarchically organized neural and cognitive components.
• Association cortices play a critical role in self-representation through associative learning.
• A unified computational framework, the forward model, can account for diverse self-related neural processes.