Materialism › Neurobiological
Hawkins’s Thousand-Brain Remembered Modeling
Consciousness as Remembered States of Cortical Models
Jeff Hawkins argues that the neocortex builds thousands of parallel models of the world, one per cortical column, and that these models reach a single perception by voting. Awareness, on his account, is the brain continuously remembering and recalling its own recent model states, while qualia reflect what each patch of uniform cortex happens to be wired to.
Jeff Hawkins
What consciousness is
Consciousness is a brain-generated simulation produced by neocortical modeling. Awareness is identified with the continuous storage and recall of the brain's recent model states; qualia with the connectivity of cortical columns.
Mind and body
Mind is what the neocortex does. Experience is the output of cortical models, voting, prediction and action, with no extra ingredient beyond the functional architecture.
Mechanism
Thousands of column-level models vote to reach a consensus percept, and awareness is the continuous memory and recall of the brain's recent model states.
Core claims
- Following Mountcastle, Hawkins holds that the neocortex runs one common algorithm, with regions differing in connections rather than structure.
- Each cortical column learns complete models of objects using reference frames derived from grid-cell and place-cell mechanisms.
- Columns vote to produce unified perception from fragmentary sensory input.
- An action erased from memory would be judged unconscious, which Hawkins takes to show that awareness consists in remembered states.
- Any system that learns a world model and continuously stores and recalls its states will be conscious.
Criticisms
Reviewers sympathetic to the neuroscience object that identifying qualia with connectivity restates the correlation without explaining why any pattern is felt. Neuroscientists also attack the sharp old-brain/new-brain split as a revival of the discredited triune-brain schema, since the mammalian neocortex closely resembles the bird and reptile pallium.
Implications
- AI consciousness: Yes
- Hawkins holds that machines built on Thousand Brains principles would be conscious, though affectless, since they lack the evolved drives of the old brain.
- Survival beyond death: No
- Since consciousness here is generated by neocortical modeling, nothing survives the brain's death.
- Meaning and purpose: Humanistic only
- The theory explains experience as brain function and posits no cosmic purpose.
- Virtual immortality: Possible
- If the right functional architecture suffices for consciousness, a system reproducing a person's cortical models could in principle carry it, though Hawkins does not address uploading.
Related theories
- Predictive Theories (Top-Down) (Overlap)
- Budson’s Consciousness as a Memory System (Overlap)
- Chalmers's Panpsychism (Contrast)
- Functionalism (Compatible)
Sources
- A Thousand Brains: A New Theory of Intelligence, Jeff Hawkins (2021)
- On Intelligence, Jeff Hawkins and Sandra Blakeslee (2004)
- A framework for intelligence and cortical function based on grid cells in the neocortex, Jeff Hawkins, M. Lewis, M. Klukas, S. Purdy and S. Ahmad (2019)
- A theory of how columns in the neocortex enable learning the structure of the world, Jeff Hawkins, S. Ahmad and Y. Cui (2017)
- Why neurons have thousands of synapses, a theory of sequence memory in neocortex, Jeff Hawkins and S. Ahmad (2016)
- An organizing principle for cerebral function: The unit module and the distributed system, Vernon B. Mountcastle (1978)
- Facing up to the problem of consciousness, D. J. Chalmers (1995)