Version: 0.1 · Build: 2026-08-19 13:10 UTC
Self from Coupled Predictive Loops Hypothesis¶
A structural framework for explaining consciousness based on predictive dynamics, temporal compression, and signal coupling Self from Coupled Predictive Loops Hypothesis
Overview¶
Intuitive Overview¶
We usually think of the “self” as something stable and persistent. This framework proposes instead that what we call the “self” is better understood as a trajectory that continuously extends through time.
The brain continuously engages in multiple interrelated processes: it predicts what may happen next, maintains multiple possible directions of future development, and compresses past experience into an integrated structure that can be continuously used.
In such a system, the future is not a single predetermined path. Instead, multiple possible predictive trajectories coexist. As the actual state continues to evolve, some predictions become more consistent with what actually occurs. The system tends to maintain the trajectory that is both most coherent with the past and most consistent with ongoing state evolution. It can be understood as the “smoothest path” continuously selected from among many possibilities.
This dominant trajectory, continuously selected and maintained over time, constitutes what we experience as the “self.”
Subjective experience can be understood through the same structure. Multiple processes—including prediction, matching, and state integration—operate simultaneously within the system. When their signals are highly synchronized in time, refer to the same state, share a representational space, and lack independent channels through which the system could distinguish their sources, they become internally indistinguishable.
As a result, what we experience is not a collection of independent signals, but a unified whole.
The sense of “I am thinking” or “I am acting” further involves attribution. The system typically does not generate only one prediction about the future; instead, multiple possible predictions coexist. When actual state evolution matches one of these predictions, and that match persists over time, the corresponding action or thought is attributed to the current dominant trajectory and is therefore experienced as “generated by me.”
If predictions repeatedly fail, or if continuity between actual evolution and the current trajectory substantially decreases, attribution may shift, causing certain actions or thoughts to appear as though they were not self-generated.
These multiple predictions do not require perturbations to exist. They may arise from the system’s intrinsic uncertainty, probabilistic sampling, or parallel modeling of different possible states, and they may also be influenced by internal or external perturbations. Perturbation is therefore only one factor that may affect the multi-prediction structure; it is not a necessary condition for the formation of self or subjective experience.
From this perspective, the self is not an entity hidden within the system, but a dynamic process that is continuously selected and maintained. Likewise, the unity of experience does not arise because only one process exists within the system, but because multiple processes are structurally unavailable for further differentiation by the system itself.
In one sentence:
What we experience as the “self” is a temporal trajectory continuously selected from among multiple possibilities; subjective experience is the unified state produced when multiple internal processes along that trajectory become indistinguishably coupled.
Formal Overview¶
This framework proposes that the self and subjective experience can be understood as structural products of a prediction-driven system. The system continuously generates multiple possible future state trajectories, compresses historical states into representations, and maintains a dominant trajectory within trajectory space through sustained matching between predictions and actual state evolution.
At any given moment, the system does not have only one possible future but implicitly maintains multiple predictive trajectories. As actual evolution unfolds, the system tends to select the trajectory with the lowest continuity error under a cross-temporal prediction–evolution consistency constraint. The stable continuation of this trajectory over time constitutes the “self.”
Subjective experience corresponds to the indistinguishable coupling of multiple internal signals along the dominant trajectory. When predictive, matching, integrative, and related signals are highly synchronized in time, co-refer to the same state within representational space, and lack independent channels or external references, they cannot be further distinguished within the system and therefore functionally manifest as a unified experience.
Agency is not simply the result of a single successful prediction, but a process of matching and attribution under conditions of multiple predictions. When actual state evolution persistently matches a particular prediction, and that match remains continuous with the current dominant trajectory, the corresponding action or thought is attributed to that trajectory and is therefore experienced as “generated by me.”
Sources of multi-prediction may include internal uncertainty, probability distributions or sampling processes, parallel expansion of different strategies and future states, modeling of potential perturbations, and the effects of perturbations themselves on prediction or actual evolution. Thus, multi-prediction does not depend on perturbation, and neither the self nor subjective experience requires perturbation as a necessary condition.
Formally, the framework can be compressed into three correspondences:
Self = trajectory with minimal continuity error
Experience = indistinguishable coupled signals on the selected trajectory
Agency = attribution based on sustained prediction–evolution matching
Theory¶
The core of the theory is not a set of three independent functional modules, but a continuously operating dynamic structure:
Multi-prediction generation
→ Actual state evolution
→ Prediction–evolution matching
→ Continuity evaluation
→ Dominant trajectory selection
→ Historical trajectory compression and updating
→ Constraints on subsequent predictions
Within this cycle, the self corresponds to the continuously maintained dominant trajectory; experience corresponds to indistinguishable signal coupling along that trajectory; and agency corresponds to source attribution following sustained matching between actual evolution and prediction.
→ View the full theoretical framework
Core Concepts¶
Terms such as “self,” “experience,” “agency,” “trajectory,” “continuity error,” “coupling,” and “indistinguishability” have specific definitions within this framework and do not fully correspond to their meanings in everyday language.
The three most central concepts are:
- Self: the dominant trajectory continuously selected within trajectory space
- Subjective Experience: the indistinguishable coupling of multiple signals along the dominant trajectory
- Agency: an attribution mechanism based on sustained matching between prediction and actual evolution
→ View core concepts and terminology
Core Mechanism¶
At any given moment, the system maintains multiple possible predictive trajectories simultaneously, while actual evolution proceeds along only one state path.
The system continuously compares actual evolution with predictive trajectories and maintains a dominant trajectory through cross-temporal continuity constraints. Historical states are simultaneously compressed into reusable trajectory representations, which in turn influence subsequent predictions.
This structure produces three interrelated outcomes:
Trajectory selection → Self
Indistinguishable coupling → Subjective experience
Sustained matching and attribution → Agency
Perturbations may alter the predictive distribution or actual evolution, but they are not necessary for this structure to operate.
Indistinguishability Principle¶
When multiple internal signals:
- Are highly synchronized in time
- Refer to the same state variables
- Share a representational space
- Lack independent channels
- Lack external references
the system cannot reliably distinguish the sources of these signals.
Therefore:
The unity of experience may arise from the internal indistinguishability of multiple signals, rather than from a single signal or a single consciousness module.
Explanatory Power¶
This framework attempts to explain the following phenomena using the same set of structural mechanisms:
- The unity of conscious experience
- The temporal continuity of the self
- The typical uniqueness of the self
- The sense of being the subject of “I am thinking” and “I am acting”
- Self-attribution of actions and thoughts
- Certain disturbances of self-attribution and agency
Rather than treating these as independent problems, the framework interprets them as different manifestations of the joint operation of multi-prediction, trajectory selection, continuity constraints, signal coupling, and attribution.
Scientific Background¶
This hypothesis is not constructed from scratch. It has structural correspondences with several existing research directions, including:
- Predictive Processing
- Forward Models
- Predictive Coding
- Efficient Coding / Information Compression
- Temporal Integration
- Sense of Agency / Comparator Models
- Global Workspace Theory
- Integrated Information Theory
- Self-Model Theory
Existing research provides varying degrees of support for mechanisms involving prediction, error processing, temporal integration, information compression, action attribution, neural synchronization, and self-modeling.
However, an important distinction must be maintained:
Evidence supporting individual mechanisms within the framework does not amount to validation of the complete “multi-prediction–trajectory selection–indistinguishable coupling” theory.
Predictions¶
To develop the framework from an explanatory model into a scientifically testable hypothesis, the project proposes a series of testable predictions.
The primary variables include:
- Continuity error
- Degree of prediction–evolution matching
- Structure of the multi-prediction distribution
- Trajectory stability
- Signal coupling strength
- Attribution stability
For example, the theory predicts that when continuity errors among multiple candidate trajectories become very similar, switching of the dominant trajectory should become more likely. Likewise, when sustained matching between prediction and actual evolution is disrupted, the sense of agency and self-attribution should decrease accordingly.
Limitations¶
This framework is currently at an early theoretical stage.
Major unresolved issues include:
- Continuity error has not yet received a rigorous mathematical definition
- The trajectory selection mechanism has not yet been expressed as a complete dynamical model
- The specific implementation of multi-prediction remains undetermined
- The agency attribution mechanism has not yet been mathematically formalized
- A theoretical explanatory gap remains between indistinguishability and subjective experience
- Direct experimental validation of the complete framework is currently lacking
- The framework does not yet provide sufficiently distinctive predictions to clearly differentiate it from other theories of consciousness
In addition, this model does not attempt to explain:
Why information processing itself is accompanied by subjective experience.
It is therefore not currently proposed as a solution to the “hard problem” of consciousness.
Status¶
This is an early-stage theoretical hypothesis.
Currently:
- A basic conceptual framework has been established
- Basic structural correspondences for Self / Experience / Agency have been defined
- Multi-prediction, trajectory selection, and attribution mechanisms have been proposed
- Relationships with existing scientific research have been outlined
- Preliminary testable predictions have been proposed
However:
- Rigorous mathematical formalization has not yet been completed
- A complete computational model has not yet been developed
- The framework has not yet undergone direct experimental validation
Future Directions¶
The next stage will primarily involve:
- Defining continuity error and its relevant timescales
- Developing a dynamical systems model of trajectory selection
- Formalizing prediction–evolution matching and attribution mechanisms
- Constructing a minimal multi-prediction simulation system (toy model)
- Defining observable variables and experimental metrics
- Developing critical experimental predictions capable of distinguishing this framework from related theories
Documentation¶
The Chinese-language workspace is used to record theoretical derivations, supplementary material, and open questions that have not yet been fully stabilized.
License¶
This project is licensed under:
Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 CC BY-NC-ND 4.0
https://creativecommons.org/licenses/by-nc-nd/4.0/
Permitted:
- Redistribution
- Citation
Prohibited:
- Commercial use
- Unauthorized modification or creation of derivative versions
For permission to modify, create derivative works, or use the material commercially, please contact the author via GitHub.
See the LICENSE file for full terms.
The Chinese-language documents primarily serve as a workspace for theoretical development, organization, and extension. The English-language version is intended to serve as the primary public-facing version.