Modeling the dynamics of evaluation: a multilevel neural network implementation of the iterative reprocessing model

Area of Science:

• Cognitive Neuroscience
• Social Psychology
• Computational Neuroscience

Background:

• Attitudes and social evaluations are dynamic and change with processing.
• Existing models lack a mechanistic explanation for this evolution.
• The iterative reprocessing (IR) model proposes a hierarchical neural basis for developing social evaluations.

Purpose of the Study:

• To present a neural network implementation of key components of the IR model.
• To simulate the dynamic evolution of social evaluations through iterative processing.
• To provide computational support for the IR model and offer insights into attitude theory.

Main Methods:

• Developed a multilevel, bidirectional feedback neural network.
• Simulated the processing of social stimuli over repeated iterations.
• Integrated initial perceptual and later semantic processing stages within the network.

Main Results:

• The neural network demonstrated evolving evaluations of stimuli over iterations.
• Higher levels of semantic processing led to changes in stimulus evaluation.
• The network's performance supported the core tenets of the IR model.

Conclusions:

• The neural network implementation successfully models the IR framework for social evaluation.
• This work provides a computational basis for understanding how attitudes and stereotypes change.
• The findings offer new insights into the mechanisms underlying social cognition and attitude formation.