Related Concept Videos
Working Memory
Language and Cognition
Reason and Intuition
High-Level and Low-Level Awareness
Higher Mental Functions of the Brain: Language
Language formation and comprehension take place in the dominant hemisphere. The dominant hemisphere is responsible for understanding the meaning of spoken, written, or sign language, as well as the ability to communicate. For most people, the left hemisphere is the dominant one. The right hemisphere, then, gives tone and emotional context to the...
Metacognition
You might also read
Related Articles
Articles linked to this work by shared authors, journal, and citation graph.
Active Inference with Dynamic Planning and Information Gain in Continuous Space by Inferring Low-Dimensional Latent States.
Modeling autonomous shifts between focus state and mind-wandering using a predictive-coding-inspired variational recurrent neural network.
How deep will you go? Hierarchy in predictive coding and transformers.
Related Experiment Video
Updated: Jan 13, 2026

The Adventures of Fundi Intervention Based on the Cognitive and Emotional Processing in Attention Deficit Hyperactive Disorder Patients
Published on: June 12, 2020
Working Memory and Self-Directed Inner Speech Enhance Multitask Generalization in Active Inference.
Jeffrey Frederic Queißer1, Jun Tani2
1Okinawa Institute of Science and Technology, Okinawa 904-0412, Japan jeffrey.queisser@oist.jp.
This study demonstrates how stacked recurrent neural networks (RNNs) with two working memory modules and inner speech can acquire complex tasks. This approach enhances task performance and generalization through structured interplay between memory and language.
More Related Videos
06:57Utilizing Electroencephalography Measurements for Comparison of Task-Specific Neural Efficiencies: Spatial Intelligence Tasks
Published on: August 9, 2016
06:45Task Interruption and Resumption Paradigm for Testing the Activation and Pursuit of an Abstract Thinking Goal
Published on: April 18, 2017
Area of Science:
- Computational neuroscience
- Cognitive modeling
- Artificial intelligence
Background:
- Working memory is crucial for cognitive tasks.
- Recurrent neural networks (RNNs) are used to model sequential data.
- Active inference (AIF) provides a framework for goal-directed behavior.
Purpose of the Study:
- To investigate simultaneous acquisition of working memory tasks using a stacked RNN and multiple working memories.
- To explore the role of inner speech in task performance and generalization.
- To analyze the internal dynamics of the model under optimal conditions.
Main Methods:
- Simulated a stacked RNN interacting with multiple working memory modules.
- Employed supervised training by minimizing free energy.
- Utilized the active inference (AIF) framework for goal-directed task execution.
- Incorporated self-directed inner speech during task performance.
Main Results:
- Optimal task performance was achieved with two working memory modules and inner speech.
- A temporal hierarchy developed in the stacked RNN under optimal conditions.
- The model demonstrated generalization capabilities across novel task configurations.
Conclusions:
- The interplay between working memory and self-directed language generation supports generalization.
- This model provides insights into the neural mechanisms underlying working memory and language.
- The findings suggest a potential architecture for artificial general intelligence.