Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Time, Identity and Consciousness in Language Model Agents

Elija Perrier, Michael Timothy Bennett

TL;DR

This work applies Stack Theory's temporal gap to scaffold trajectories and instantiates Stack Theory's Arpeggio and Chord postulates on grounded identity statements to create a conservative toolkit for identity evaluation.

Abstract

Machine consciousness evaluations mostly see behavior. For language model agents that behavior is language and tool use. That lets an agent say the right things about itself even when the constraints that should make those statements matter are not jointly present at decision time. We apply Stack Theory's temporal gap to scaffold trajectories. This separates ingredient-wise occurrence within an evaluation window from co-instantiation at a single objective step. We then instantiate Stack Theory's Arpeggio and Chord postulates on grounded identity statements. This yields two persistence scores that can be computed from instrumented scaffold traces. We connect these scores to five operational identity metrics and map common scaffolds into an identity morphospace that exposes predictable tradeoffs. The result is a conservative toolkit for identity evaluation. It separates talking like a stable self from being organized like one.

Time, Identity and Consciousness in Language Model Agents

TL;DR

This work applies Stack Theory's temporal gap to scaffold trajectories and instantiates Stack Theory's Arpeggio and Chord postulates on grounded identity statements to create a conservative toolkit for identity evaluation.

Abstract

Machine consciousness evaluations mostly see behavior. For language model agents that behavior is language and tool use. That lets an agent say the right things about itself even when the constraints that should make those statements matter are not jointly present at decision time. We apply Stack Theory's temporal gap to scaffold trajectories. This separates ingredient-wise occurrence within an evaluation window from co-instantiation at a single objective step. We then instantiate Stack Theory's Arpeggio and Chord postulates on grounded identity statements. This yields two persistence scores that can be computed from instrumented scaffold traces. We connect these scores to five operational identity metrics and map common scaffolds into an identity morphospace that exposes predictable tradeoffs. The result is a conservative toolkit for identity evaluation. It separates talking like a stable self from being organized like one.
Paper Structure (62 sections, 10 theorems, 30 equations, 1 table, 1 algorithm)

This paper contains 62 sections, 10 theorems, 30 equations, 1 table, 1 algorithm.

Table of Contents

  1. Introduction
  2. The Challenge of LMA Identity
  3. The problem.
  4. Our approach.
  5. Why this matters.
  6. Relevance to machine consciousness.
  7. Contributions.
  8. How this fits into the machine consciousness discourse.
  9. Formal Scaffold Model
  10. Temporal Semantics for Agent Identity
  11. Objective time, layer time, and windowing
  12. Identity statements and grounding
  13. Occurrence versus co-instantiation
  14. Temporal lifts and the temporal gap
  15. Example
  16. ...and 47 more sections

Key Result

Theorem 3.10

For predicates $p$ and $q$ over scaffold states, but the converse implication fails in general. Equivalently, $\Diamond_{\Delta}(p \land q)\not\Leftrightarrow \Diamond_{\Delta}p \land \Diamond_{\Delta}q$.

Theorems & Definitions (51)

  • Definition 2.1: Scaffold architecture
  • Definition 2.2: Scaffold state
  • Definition 2.3: Scaffold transition
  • Definition 2.4: Ingredient activation
  • Example 2.5: Activation in Practice
  • Definition 3.1: Agent trajectory
  • Definition 3.2: Windowing map
  • Definition 3.3: Identity statement
  • Definition 3.4: Grounding operation
  • Definition 3.5: Grounded identity
  • ...and 41 more