run record~/research/cliste/runsAll Cliste runs

Type a section name and press Enter. Use “help” for the list.

Figures: softened for dark viewingDisplay only; image files are unchanged.
← All Cliste runs

Cliste experiment

Remembered continuation from partial visual perception

Registered 9 Oct 2026 · Snapshot 12 Oct 2026

Execution: failedScientific verdict: No final scientific verdict

The question

Connect partial VSA recall to retained temporal context and actual working continuation across new experience, growing memory and restart; compare baseline and support-scaled acceptance with visible errors.

Why this experiment?

A partial cue can resemble the wrong memory. Recall needs to recover useful content while recognizing when the evidence is insufficient.

How it works

Compare partial-cue recall with familiar and unfamiliar inputs, retaining incorrect accepts and refusals. Cleanup and confidence rules are qualified before relying on them for a remembered continuation.

Follow this part of the investigation: More answers are useful only if the extra confidence is justified →
Implementation and comparison details

Baseline and support-scaled partial recall drive actual temporal continuation. The latest life grows from79 to103 to127 encounters; fixed widening, causal context controls and restart expose correct, wrong and missed continuations.

What happened

Execution was externally interrupted during the reported daemon restart. The last after-stage log records query 95/480 at 2026-10-09T20:03:20.001514+00:00; host inspection found no surviving study worker or launcher. Before and middle source-denied receipts, ordinary acquisition, integrated acquisition, and integrated continuation receipts passed. The final after panel, restart/context controls and independent audit did not complete. This is an incomplete execution, not evidence against the scientific hypothesis. All 445 frozen source/input hashes and 55 dependency hashes remain unchanged; partial outputs and the preliminary visual are preserved. Any repeat must use a separately registered recovery attempt and disclose already exposed inputs.

Results and visual artifacts

Open a figure for its input, comparison, and result. Click an image to enlarge it.

This attempt stopped before its final evaluation completed. The displayed response survives in the completed recovery experiment, where it can be checked against the source and intended outcome. Following a remembered sequence is a separate ability from choosing the right starting memory.

The preliminary response was a false recall
The preliminary response was a false recall
The 80 × 80 input region came from an image that was never stored. With 79 encounters in memory, the original acceptance rule nevertheless returned the room, window, and airplane sequence. All three outputs were decoded from stored vectors, but none makes this the correct response to the cue. The recovery experiment later established this outcome; the original attempt remains incomplete.

Original recorded figure ↗

Replotted from saved outputs with clarified labels. The original recorded figure remains available.

Limits and next question

Supplied local visibility and replay control; fresh-photo comparisons and correlated windows from two previously examined trailers. Remembered encounter order is not prediction or learned attention. No default promotion; full milestone 1.3 and the sealed final panel remain open.

Next question. Does recall help remembered continuation, what limits its usefulness, and which quality or resource issue should the planned optimization round address?

Inspect the evidence

Download the methods and measurements for this experiment. Large training and saved-state files are not included.