Synthetic Null Hypothesis
A statistical control method that artificially constructs null-hypothesis datasets to simulate false positives a probe may produce due to input artifacts or confounding effects, used to judge whether information read out by a probe is genuine.
Timeline (last 90 days)
The authors propose a testing protocol that introduces a synthetic null to simulate generic mixing effects and test whether the probe reads a computed value or an input copy/mixing artifact.
In a pre-registered study of two-step problems, this protocol captures a weak signal about the unwritten intermediate value a+b before Ouro writes the last digit of the answer, and the sign of the signal follows the type of operation.
For research that relies on probe conclusions to infer what a model "knows," minimal-pair designs and synthetic nulls should become standard configuration.
All Facts (3)
The authors propose a testing protocol that introduces a synthetic null to simulate generic mixing effects and test whether the probe reads a computed value or an input copy/mixing artifact.
50%UnverifiedIn a pre-registered study of two-step problems, this protocol captures a weak signal about the unwritten intermediate value a+b before Ouro writes the last digit of the answer, and the sign of the signal follows the type of operation.
50%UnverifiedFor research that relies on probe conclusions to infer what a model "knows," minimal-pair designs and synthetic nulls should become standard configuration.
50%