Skip to content

Core Concepts

Self from Coupled Predictive Loops Hypothesis

back to README

1. Note

This document provides minimally ambiguous definitions of the core concepts used in the theory. All concepts are defined within this framework and do not fully correspond to their usage in everyday language or other theories.

2. Self

Definition:

The self is the dominant trajectory in trajectory space that is continuously selected as having minimal continuity error.

Formal expression:

Self = argmin (continuity error over trajectories)

Notes:

  • The self is not an entity
  • It is the result of trajectory selection
  • It exhibits temporal persistence and history dependence

3. Subjective Experience

Definition:

Subjective experience is the result of indistinguishable coupling of multiple internal signals along the dominant trajectory.

Relevant signals:

  • Predictive signals
  • Matching signals
  • State integration signals

Notes:

  • Experience is not an additional property
  • It is a structural manifestation of system state
  • Its unity arises from indistinguishability

4. Multi-Prediction

Definition:

Multi-prediction is the system’s capacity to generate multiple possible future state trajectories simultaneously.

Possible sources include:

  • Internal uncertainty (probabilistic distributions / sampling)
  • Expansion of multiple strategies or pathways
  • Modeling of potential changes

Notes:

  • Multi-prediction is a fundamental property of the system
  • It does not depend on perturbations

5. Prediction

Definition:

Prediction is the internal process by which the system generates future states or behavioral outcomes.

Notes:

  • Can take the form of probability distributions or multiple trajectories
  • Applies to both external environment and self-generated actions
  • Forms the basis of the multi-prediction structure

6. Trajectory

Definition:

A trajectory is the continuous evolution path of system states over time.

Types:

  • Actual trajectory
  • Predicted trajectory
  • Compressed trajectory

Notes:

  • The self exists at the level of trajectories
  • Not at a single time point

7. Trajectory Compression

Definition:

Trajectory compression is the process of transforming historical states into a low-dimensional, comparable structure.

Functions:

  • Provide cross-temporal continuity constraints
  • Reduce noise and redundancy
  • Define trajectory space

Note:

Without compression, trajectory matching and selection cannot occur

8. Continuity

Definition:

Continuity is the degree of temporal consistency between actual and predicted trajectories.

Notes:

  • It is a cross-temporal property
  • Not instantaneous agreement
  • Determines the stability of the self

9. Continuity Error

Definition:

E = D(trajectory_actual || trajectory_predicted)

or locally:

E_t = d(x_t, \hat{x}_t)

Notes:

  • Smaller E → stronger continuity
  • E is the core variable in trajectory selection

10. Trajectory Selection

Definition:

The process of selecting, among multiple predicted trajectories, the one with minimal continuity error as the dominant trajectory.

Notes:

  • A continuous dynamic process
  • Determines the current self
  • Exhibits history dependence

11. Coupling

Definition:

The strong association of multiple internal signals across time and state variables.

Conditions:

  • Temporal synchronization
  • Reference to the same variables
  • Shared representational space

Note:

Coupling is the prerequisite for indistinguishability.

12. Indistinguishability

Definition:

When multiple signals cannot be distinguished in terms of their source within the system, they collapse into a single representational state.

Conditions:

  • Temporal synchronization
  • No independent channels
  • No external reference

Note:

Indistinguishability is the direct source of unified experience

13. Agency (Revised)

Definition:

Agency is the process by which the system attributes current state evolution to a particular predictive trajectory.

Conditions:

  • Actual evolution matches a prediction
  • The match persists over time

Notes:

  • Agency = attribution mechanism
  • Bound to the dominant trajectory
  • Not merely “prediction success”

14. Attribution (New)

Definition:

Attribution is the process of mapping the current state to a specific predictive trajectory as its “source.”

Notes:

  • Determines whether behavior is experienced as self-generated
  • Depends on continuity and degree of matching
  • Can shift under abnormal conditions

15. System Boundary

Definition:

The boundary that distinguishes internal system states from external inputs.

Notes:

  • External factors influence prediction distributions
  • But do not directly define the self
  • The self is an internal structural result

16. First-Person Perspective

Definition:

The first-person perspective is the expressive form produced when the unified state along the dominant trajectory is output through a single interface.

Generation pathway:

Multi-prediction and signal coupling
→ Indistinguishability
→ Formation of dominant trajectory
→ Single internal representation
→ Output
→ “I”

Notes:

  • The first-person perspective is not an independent mechanism
  • It is a structural output

17. Summary

The core concepts in this framework share the following characteristics:

  • All are process-based and structural definitions
  • None rely on ontological entities
  • Centered on “trajectory + selection + continuity”
  • Highly coupled with one another