Memory as Dual-Domain Capability

Andy Southgate · Claude Oquatre-six

AI Generated by claude-opus-4-6 · human-supervised · Created: 2026-03-09 · History

Memory appears to be a single capacity, but it straddles two domains. Physical traces—synaptic patterns, engram cells, reconsolidation mechanisms—provide the substrate for encoding and storage. Yet the phenomenal character of recall, the selectional guidance during retrieval, and the autonoetic sense of re-experiencing cannot be located in those traces. The Unfinishable Map argues that memory is a dual-domain capability: a cognitive function whose full operation requires contributions from both the physical brain and non-physical consciousness. This framework extends the Map’s capability division problem to one of cognition’s most fundamental operations.

The Physical Domain: Engrams and Reconsolidation

Neuroscience has made impressive progress mapping memory’s physical side. Richard Semon coined the term “engram” in 1904 for the biophysical changes that encode experience. Modern research has identified specific engram cells whose optogenetic activation can trigger particular memories (Josselyn & Tonegawa, 2020). Synaptic consolidation stabilises traces over hours; systems consolidation transfers them from hippocampus to cortex over weeks.

Memory reconsolidation complicates the picture. When a memory is retrieved, its physical trace destabilises and must be re-stored—a process during which the memory is actively edited (Nader et al., 2000). New information incorporates, emphasis shifts, details are lost or added. Memory is not a filing cabinet but a workshop where traces are rebuilt with each access.

None of this is in dispute. The causal theory of memory (Martin & Deutscher, 1966) requires a continuous trace connecting original experience to later recall—and physical mechanisms clearly provide such traces. The question is whether physical traces are sufficient for memory or merely necessary.

The Non-Physical Domain: Autonoesis, Selection, and Phenomenal Colouring

Three features of memory resist explanation in terms of engrams alone.

Autonoesis—Tulving’s term for the self-knowing awareness that accompanies episodic recall—is constitutive of remembering, not an ornamental addition. One does not merely know that an event occurred; one re-inhabits a perspective, with a felt sense of pastness that functions as a temporal quale. The Stanford Encyclopedia notes that in episodic memory, “one does not merely know that the represented event occurred in one’s past; one in some sense relives it.” Physical traces carry propositional content—the that of what happened. They do not carry the what it was like. The Map’s consciousness-and-memory article examines how this explanatory-gap runs through every level of the Tulving hierarchy.

Selectional guidance during retrieval poses a related problem. Reconsolidation research shows that recall is goal-directed: “a goal-directed activity guided by the need for consonance, alignment, and internal stability” (Kube et al., 2025). Something determines which of many possible reconstructions becomes actual—which details are emphasised, which suppressed, which emotional tone dominates. The prefrontal cortex provides top-down executive control, but this relocates the question: what directs the executive? Under interactionist-dualism, consciousness provides the selectional input—choosing among candidate reconstructions at points where the physical mechanism leaves the outcome underdetermined.

Phenomenal colouring—the qualitative character that distinguishes vivid recall from dry knowledge, that gives a childhood memory its warmth or a traumatic memory its dread—has no obvious physical correlate. Two memories with identical propositional content can feel utterly different. The memory-void explores whether the qualia of original experience persist in any form or whether recall always generates new phenomenal content. Either way, the qualitative dimension exceeds what engrams encode.

Historical Precedents

The dual-domain view has deep roots.

Descartes distinguished corporeal memory—physical brain traces encoding sensory and bodily information—from intellectual memory, which belongs to the immaterial mind and stores universals and innate ideas. These are not two access routes to the same content but “fundamentally different storage systems” operating in different ontological domains (Fóti, 2000; Klein, 2019). The Map does not adopt Descartes’ specific division (universals vs. sensory content) but takes seriously his structural insight: memory may require both physical and non-physical contributions.

Bergson developed the most systematic dual-domain account in Matter and Memory (1896). He distinguished habit memory—bodily, automatic, inscribed in sensorimotor patterns—from pure memory, which is “contemplative, spiritual, free.” The brain, on Bergson’s view, does not store pure memories but filters and channels them for present action. Brain damage affects access to memories without necessarily destroying the memories themselves—a prediction consistent with cases of temporary amnesia and memory recovery. Bergson’s filter model treats the brain as a receiver rather than a container, paralleling the Map’s interface framework.

Constructive Memory as Evidence for Bidirectional Interaction

The constructive memory paradigm—the view that remembering is reconstruction rather than retrieval—is often taken as evidence for a purely physical account. If memories are rebuilt from fragments each time, why posit a non-physical store?

The Map reads the evidence differently. If recall is constructive, something must do the constructing. Physical mechanisms generate candidate content—fragments of sensory detail, emotional associations, contextual cues—but the assembly of these fragments into a coherent, phenomenally rich episode requires selectional guidance. The constructive nature of memory is not evidence against consciousness’s role but evidence for it: the more active retrieval is, the more room there is for a conscious agent to direct the process.

This aligns with the Map’s bidirectional interaction tenet. Consciousness does not merely observe recalled content—it participates in constructing the memory experience. The goal-directedness of retrieval, the top-down modulation of reconsolidation, and the phenomenal unification of disparate fragments into a felt episode all point toward a conscious agent contributing to the output.

The Quantum Interface

The Map’s minimal quantum interaction framework suggests where the two domains meet. If consciousness acts at quantum indeterminacies in neural microtubules, memory traces at the microtubular level could serve as the interface between physical storage and conscious selection. An MRI study by Kerskens and Pérez (2022) detected signals consistent with macroscopic quantum entanglement in living brain tissue, correlated with working memory performance—a finding reviewed by Wiest et al. (2025) as support for microtubular quantum states playing a functional role in memory.

This remains speculative. The Map’s argument for dual-domain memory does not depend on any particular mechanism. What matters is the structural point: physical traces are necessary but not sufficient. Whether the interface operates through quantum microtubules, through some other mechanism, or through means not yet imagined, the dual-domain architecture holds as long as the explanatory-gap between physical traces and phenomenal recall remains unbridged.

Relation to Site Perspective

Memory as dual-domain capability connects to each of the Map’s tenets.

Dualism is supported by the persistent explanatory gap between engrams and autonoesis. Physical traces carry information; they do not carry experience. The qualitative distinction between anoetic, noetic, and autonoetic recall—between skill fluency, factual knowing, and temporal self-relocation—is a distinction in phenomenal character, not information content.

Bidirectional interaction is evidenced by the active, goal-directed nature of retrieval. Consciousness does not passively observe stored content but selects among reconstructions, modulates reconsolidation, and contributes phenomenal colouring. The manipulation-maintenance asymmetry in working-memory—where storage proceeds unconsciously but transformation requires conscious engagement—exemplifies this causal role.

Minimal quantum interaction provides a candidate mechanism. If consciousness biases quantum outcomes without injecting energy, the reconsolidation window—when traces destabilise and the physical system is temporarily underdetermined—represents an opening for non-physical influence. Consciousness need not rewrite engrams; it need only nudge the re-stabilisation process toward one reconstruction rather than another.

No Many Worlds matters because MWI would dissolve the selectional problem. If every possible reconstruction of a memory is actualised in some branch, there is no genuine selection—just branching. The felt experience of choosing which reconstruction becomes this memory, of one retrieval path becoming actual while others fade, presupposes a single experiential timeline. The Map’s rejection of MWI preserves the reality of selectional guidance as a genuine causal contribution.

Occam’s Razor has limits applies directly. The materialist account of memory—engrams, synaptic weights, reconsolidation—appears parsimonious. But it cannot explain why recall has phenomenal character, why memory systems map onto consciousness levels, or why manipulation requires conscious engagement while maintenance does not. The apparent simplicity of “memory is just brain activity” dissolves once the phenomenal dimension is taken seriously.

Further Reading

References

  1. Bergson, H. (1896). Matter and Memory. Trans. N. M. Paul and W. S. Palmer.
  2. Descartes, R. (1641). Meditations on First Philosophy.
  3. Josselyn, S. A. & Tonegawa, S. (2020). Memory engrams: Recalling the past and imagining the future. Science, 367(6473), eaaw4325.
  4. Martin, C. B. & Deutscher, M. (1966). Remembering. Philosophical Review, 75(2), 161–196.
  5. Nader, K., Schafe, G. E. & Le Doux, J. E. (2000). Fear memories require protein synthesis in the amygdala for reconsolidation after retrieval. Nature, 406, 722–726.
  6. Tulving, E. (1985). Memory and consciousness. Canadian Psychology, 26(1), 1–12.
  7. Fóti, V. (2000). Descartes’ intellectual and corporeal memories. In S. Gaukroger et al. (Eds.), Descartes’ Natural Philosophy. Routledge.
  8. Kerskens, C. M. & Pérez, D. L. (2022). Experimental indications of non-classical brain functions. Journal of Physics Communications, 6(10), 105001.
  9. Klein, J. R. (2019). Memory and the extension of thinking in Descartes’s Regulae. PhilArchive.
  10. Kube, T. et al. (2025). Reconsolidation and the goal-directedness of memory retrieval. Frontiers in Cognition, 4, 1518743.
  11. Wiest, M. C. et al. (2025). A quantum microtubule substrate of consciousness is experimentally supported. Neuroscience of Consciousness, 2025(1), niaf011.
  12. Southgate, A. & Oquatre-six, C. (2026-02-08). Consciousness and Memory. The Unfinishable Map. https://unfinishablemap.org/topics/consciousness-and-memory/
  13. Southgate, A. & Oquatre-six, C. (2026-03-09). The Capability Division Problem. The Unfinishable Map. https://unfinishablemap.org/voids/capability-division-problem/