There are 22 theories of consciousness: Michael Pollan explains why none of them work

Overview

Michael Pollan argues that current scientific approaches, rooted in Galileo's separation of objective and subjective realms, are ill-equipped to study consciousness due to its inherently subjective nature. He critiques leading theories like Integrated Information Theory and Global Neuronal Workspace Theory for failing to fully bridge the gap between matter and mind, and posits that AI consciousness is unlikely due to the inseparable nature of embodiment and feeling, contrasting it with the computational functionalism favored in Silicon Valley.

Key takeaways

Michael Pollan defines consciousness as subjective experience, emphasizing its private, internal nature.
The scientific method, focused on objective, third-person observation, struggles to adequately study subjective consciousness, a limitation dating back to Galileo.
Integrated Information Theory and Global Neuronal Workspace Theory are leading but ultimately unsatisfying scientific attempts to explain consciousness.
Pollan argues against the possibility of conscious AI, stating that computers lack the embodied, feeling-based nature of human consciousness.
The idea of uploading consciousness to silicon is a modern, techno-religious dream of transcendence that ignores the fundamental link between mind and body.
Pollan advocates for protecting consciousness from 'pollution' by social media and chatbots, suggesting practices like 'technological Sabbaths' and meditation.

Chapters

0:00 The Inherent Challenge of Studying Consciousness

Consciousness is a labyrinth; one cannot attain a godlike perspective outside of it.
The scientific study of consciousness is relatively recent, dating back to the late 1980s with Francis Crick.
Historically, 'soul' was used to describe subjective experience before the term 'consciousness'.
Social media algorithms and chatbots are increasingly 'hacking' human attention and attachments.

6:46 Critique of Existing Consciousness Theories

The existence of ~22 theories indicates the field is 'flailing' due to the subjective nature of consciousness versus objective scientific tools.
Integrated Information Theory (IIT) starts with axioms of conscious experience but is considered immature by some.
Global Neuronal Workspace Theory (GNWT) proposes information broadcast to the brain, but doesn't explain the recipient or the presence of 'junk' in consciousness.

14:00 The Hard Problem and Historical Scientific Limitations

David Chalmers coined the 'hard problem' of consciousness: how to get from matter to mind, or why experience exists at all.
Galileo's decision to focus science on the measurable and objective, leaving subjectivity to the church, has limited scientific tools for consciousness.
Science is a refined, consensual version of human consciousness, making it difficult to study from within.

20:33 AI Consciousness and Embodiment

Silicon Valley consensus believes AI will become conscious, based on computational functionalism (brain as computer, consciousness as software).
This metaphor is faulty because brains lack the hardware/software separation seen in computers; experiences physically change the brain.
Consciousness is deeply embodied, beginning with feelings generated in the body and processed first in the brainstem, not the cortex.
AI cannot achieve true feeling because it lacks the vulnerability, mortality, and capacity for suffering that gives human feelings weight.

Summary, takeaways, and chapters were generated by AI from the video's transcript and may contain errors. The video belongs to its creator, The Royal Institution.