Materialism › Computational
Sejnowski’s Computational Two Tiers and Traveling Waves
Consciousness as Self-Generated Thinking in a Spike-Timed Cortical Tier
Sejnowski proposes that consciousness arises from a second, cognitive tier of cortical processing that rides on top of the ordinary sensorimotor tier. The sensorimotor tier codes information in firing rates, while the cognitive tier uses millisecond-precise spike timing, organized by traveling waves and temporary synaptic plasticity, to hold context over time and to keep thinking going without outside input. He expects that once these mechanisms are understood, the problem of consciousness may disappear ‘like the Cheshire cat.’
Terrence J. Sejnowski
What consciousness is
Consciousness is the global dynamic state of a brain thinking about thinking: self-generated activity in a cognitive cortical tier that builds internal representations from past experience.
Mind and body
Mental phenomena are neural mechanisms, and Sejnowski expects philosophical questions about mind to receive scientific answers as those mechanisms are understood.
Mechanism
Traveling waves organize precisely timed spikes into a spacetime population code, which triggers spike-timing-dependent plasticity lasting hours and forms a temporary cognitive network on top of the sensorimotor network.
Core claims
- Mainen and Sejnowski found that cortical neurons initiate spikes with millisecond precision when driven by fluctuating ‘frozen noise’ input, with timing set by post-inhibitory rebound from basket cells.
- The flash-lag effect shows that perception of a flash depends on what happens after it, which Eagleman and Sejnowski interpret as postdiction.
- Traveling waves observed across many frequency bands could trigger spike-timing-dependent plasticity lasting hours, supporting long-term working memory.
- Sejnowski proposes that the BOLD fMRI signal reflects the metabolic cost of synaptic plasticity and attention more than spiking.
- Unlike large language models, whose feedforward activity stops when a prompt ends, the brain sustains self-generated activity through recurrent loops.
Implications
- AI consciousness: Possible
- Current large language models lack self-generative activity, but Sejnowski suggests that machines built with self-generative ‘System 2’ dynamics organized by traveling waves and plasticity could gain cognition closer to human experience.
- Survival beyond death: No
- Since consciousness is the global dynamic state of cortical and thalamic circuits, it follows from the view that it ends when those dynamics stop.
- Meaning and purpose: Humanistic only
- Sejnowski frames self-knowledge as understanding the mechanisms of thinking, with no cosmic purpose involved.
- Virtual immortality: Uncertain
- The view is computational and allows machine cognition in principle, but Sejnowski does not address copying an individual mind.
Related theories
- Crick and Koch's Neural Correlates of Consciousness (Foundation)
- Baars's and Dehaene's Global Workspace Theory (Overlap)
- Eagleman's Possibilianism (Overlap)
- Ward's Thalamic Dynamic Core and Min's Thalamic Reticular Gate (Compatible)
Sources
- Reliability of Spike Timing in Neocortical Neurons, Z. F. Mainen and T. J. Sejnowski (1995)
- Motion Integration and Postdiction in Visual Awareness, D. M. Eagleman and T. J. Sejnowski (2000)
- Regulation of Spike Timing in Visual Cortical Circuits, P. Tiesinga, J. M. Fellous, and T. J. Sejnowski (2008)
- Consciousness, Terrence J. Sejnowski (2015)
- The Deep Learning Revolution, Terrence J. Sejnowski (2018)
- Transformers and Cortical Waves: Encoders for Pulling in Context Across Time, L. Muller, P. S. Churchland, and T. J. Sejnowski (2024)
- Dynamical Mechanisms for Coordinating Long-term Working Memory Based on the Precision of Spike-timing in Cortical Neurons, Terrence J. Sejnowski (2026)