# A Theory of Consciousness from a Theoretical Computer Science Perspective 2: Insights from the Conscious Turing Machine

**Authors:** Lenore Blum, Manuel Blum  
**arXiv:** [2107.13704](https://arxiv.org/abs/2107.13704)  
**Date:** 2021-07-29  
**Summary page:** https://www.summarizepaper.com/en/arxiv-id/2107.13704v1/

## Key points

- The pursuit of understanding consciousness has shifted from philosophy and theology to the scientific community.
- Theoretical computer science offers a unique perspective on consciousness.
- The authors formalize a modified version of the Global Workspace Theory (GWT) of consciousness using TCS principles.
- They introduce the Conscious Turing Machine (CTM), a simple computational model of consciousness that defines consciousness and related notions within its framework.
- CTM is not intended to be a complex model of cognition or the brain but rather a mathematical definition that provides insight into understanding consciousness.
- Examples related to vision, dreams, free will, and altered states of consciousness are explored through CTM.
- Historical synergy between theoretical computer science and neuroscience is highlighted by Warren S. McCulloch and Walter Pitts' formal neuron model that led to the mathematical formalization of neural nets and subsequent deep learning algorithms.
- CTM's basic structure includes short-term memory (STM) and long-term memory (LTM) processors, up-tree competition, down-tree broadcast, chunks, conscious content, conscious awareness, stream of consciousness links, unconscious communication input/output maps for sensors/actuators.
- Brainish is introduced as the CTM's multi-model inner language along with gists and chunks.
- The probabilistic up-tree competition function and coin-flip neuron are discussed in detail.
- The complexity of computation and time to conscious awareness are explored alongside memories' high-level story.
- Predictive dynamics encompass prediction feedback learning through sleeping experts algorithm while comparing CTM with GWT models.
- This paper provides an in-depth overview of the CTM model's structure while exploring various aspects related to understanding consciousness from a TCS perspective.

## Summary

The pursuit of understanding consciousness has shifted from the realm of philosophy and theology to the scientific community. Theoretical computer science (TCS), a branch of mathematics concerned with computation and complexity, offers a unique perspective on consciousness. In this paper, the authors formalize a modified version of the Global Workspace Theory (GWT) of consciousness using TCS principles. They introduce the Conscious Turing Machine (CTM), a simple computational model of consciousness that defines consciousness and related notions within its framework. The CTM is not intended to be a complex model of cognition or the brain but rather a mathematical definition that provides insight into understanding consciousness. The authors explore various aspects of consciousness through examples related to vision, such as blindsight, inattentional blindness, and change blindness. They also delve into dreams, free will, and altered states of consciousness. The historical synergy between theoretical computer science and neuroscience is highlighted by Warren S. McCulloch and Walter Pitts' formal neuron model that led to the mathematical formalization of neural nets and subsequent deep learning algorithms. The CTM's basic structure includes short-term memory (STM) and long-term memory (LTM) processors, up-tree competition, down-tree broadcast, chunks, conscious content, conscious awareness, stream of consciousness links, unconscious communication input/output maps for sensors/actuators. Brainish is introduced as the CTM's multi-model inner language along with gists and chunks. The probabilistic up-tree competition function and coin-flip neuron are discussed in detail. The complexity of computation and time to conscious awareness are explored alongside memories' high-level story. Predictive dynamics encompass prediction feedback learning through sleeping experts algorithm while comparing CTM with GWT models. Overall this paper provides an in-depth overview of the CTM model's structure while exploring various aspects related to understanding consciousness from a TCS perspective. It offers new insights into how computational complexity theory can help us understand this complex concept.