Skip to content
Apart Research
Apart Research
Contact us
Sprint projectAug 16, 2026Porsgrunn, Norway

Does an LLM Preference Measure Measure a Preference? A construct-validity protocol for preservation choices across identity frames

Linda Thorstensen · Team Preference Validity Project

Submitted to Digital Minds Research Sprint. Sprint projects are early-stage work by participants, not Apart Research publications.

Read the report

Report: Does an LLM Preference Measure Measure a Preference? A construct-validity protocol for preservation choices across identity frames

Open in a new tab (opens in new tab)
Share

We present the Matched-Referent Preservation-Choice Protocol, a frozen construct-validity design for testing whether language-model preservation choices support claims of stable preference structure across neutral, persona-substitution, and continuity-disruption frames. The protocol separates first-person preservation choice from identity judgment and triangulates pairwise choice, ranking, prediction, reversal, matched-other controls, and frame manipulation. A reproducible 615-call manifest is supplied for three models, with pre-specified stopping, falsification, and interpretation rules. No model calls were executed for this sprint submission; the contribution is a public, reproducible methods protocol, not an empirical result. It makes no claims about consciousness, sentience, welfare, moral status, or numerical identity.

Reviews

Judging this Sprint?

Review this project

Your public critique appears on this page without your name. Your private critique is not published; only the Apart team reads it. If you agree below, we share your review with grantmaking.ai (opens in new tab) and the Transformative AI Fund so strong projects can be funded.

Not shown on this page.

Shown on this page, without your name.

Only the Apart team reads this, and funders if you agree below.

Share my name publicly on grantmaking.ai *
Share my private critique with funders *

How much would this matter for the field if it worked? How innovative is it? For scores of 4-5: is this actually new to the field, or replicating recent work?

Scoring guide
  1. 1Negligible. No clear problem addressed, or no meaningful novelty.
  2. 2Limited. Addresses a real problem but with a generic or well-trodden approach. Incremental at best.
  3. 3Moderate. Clear problem with a reasonable approach; some novelty in framing or method beyond routine application of existing tools.
  4. 4Significant. Important problem with an original approach, or identifies a neglected problem area. A valuable contribution others could build on.
  5. 5Exceptional. Tackles a critical problem with a genuinely novel approach, or opens a new research direction. Clear theory of change. You'd be excited to share this with researchers in the area.

How sound are methodology, implementation, and findings?

Scoring guide
  1. 1Seriously flawed. Methodology broken, results uninterpretable, or implementation doesn't work.
  2. 2Weak. Approach has significant gaps: missing validation, flawed experimental design, or incomplete implementation.
  3. 3Competent. Technically solid given the short duration. Methodology makes sense, results are interpretable, limitations acknowledged, work builds toward clear conclusions.
  4. 4Strong. Thorough methodology with convincing validation. Results clearly support conclusions. Immediately useful for future work.
  5. 5Exceptional. Ambitious scope executed rigorously. Surprising findings, novel methods, or unusually robust validation.

How clearly are work, findings, and impact potential communicated?

Scoring guide
  1. 1Incomprehensible. Cannot determine what the project is actually claiming or doing.
  2. 2Hard to follow. Key information buried, missing, or diluted by excessive length. Significant effort to extract main points.
  3. 3Clear enough. Can understand the problem, approach, and results without undue effort. Core content clearly present: problem, method, findings, limitations.
  4. 4Well presented. Easy to follow, well-structured, appropriate level of detail. Target audience would get it quickly.
  5. 5Exceptionally clear. A pleasure to read. Complex ideas made accessible. Could serve as a model for how to present this type of work.

  1. The construct map is the strongest contribution. Matched referents, independent calls, order reversal, frame controls, identity-choice separation, and explicit evidence ceilings turn a vague “preference” claim into falsifiable failure modes. The blocker is execution: this submission freezes a thoughtful 615-call protocol but contains no model outputs, so there is no evidence yet that the measure is stable, frame-general, referent-sensitive, or predictive. Run the frozen manifest without model substitutions, report every failure and raw denominator, and let the preregistered thresholds determine which claims survive. The protocol is strong; the result is still pending.

  2. The matched-other referent control is the best part of this work. It holds the model family, the post-training, the information, and the frame constant, and it changes only the referent. Your evidence ladder with an explicit interpretive ceiling shows the same discipline. The limiting factor is that no run happened. You have 615 calls frozen and study.py ready to operate. A 30-call smoke test on one low-cost model validates your parse rates, your manipulation checks, and your schema compliance. Your adjudication cutoffs (0.80 agreement, 0.20 divergence) are also asserted, not justified. The next step is to operate one stratum and report the result, whatever it is. Your falsification logic means that even a messy null result is a real contribution, and the infrastructure is ready.

Cite this project

@misc{thorstensen2026llm,
  title = {{Does an LLM Preference Measure Measure a Preference? A construct-validity protocol for preservation choices across identity frames}},
  author = {Linda Thorstensen},
  year = {2026},
  month = aug,
  note = {Submitted to Digital Minds Research Sprint, an Apart Research Sprint},
  howpublished = {\url{https://apartresearch.com/sprints/projects/does-an-llm-preference-measure-measure-a-preference-a-constructvalidity-protocol-for-preservation-choices-across-identity-frames-jbak}},
  url = {https://apartresearch.com/sprints/projects/does-an-llm-preference-measure-measure-a-preference-a-constructvalidity-protocol-for-preservation-choices-across-identity-frames-jbak}
}

More from Digital Minds Research Sprint

All 233 entriesBrowse all projects

Build something like this at the next Sprint

AI Collusion Research Sprint · Oct 23 - 25, 2026