Equation 10 · Running an Interpretability Investigation That Holds Up
What does this equation mean?
Read the formula alongside the article passage below. Each part has a deeper page with its role in the equation, the supporting passage and nearby citations.
This equation states an equality: the expressions on both sides have the same value under the article’s assumptions. Read the equation part by part below; each part has a contextual explanation and a link to its mathematical background.
Read it piece by piece
Symbol S
S is part of the quantity the equation computes from the expression on the right.
Symbol C
C is an argument of the function-like quantity on the left; its role is set by that function’s stated inputs.
Symbol Δ
Δ is one of the signed contributions combined to compute the quantity on the left.
=
The expressions on both sides represent the same quantity under the stated assumptions.
See an illustrated explanation →subtraction
Subtract the following term or group from the preceding one. A leading minus marks a negative quantity.
How to interpret it
Read it with the definitions, units, and assumptions supplied by the article.
What the article says around this equation
The defence against this is a control condition run alongside the target one, not after it: intervene identically on a component set ' matched to the target set in size, depth and baseline activation statistics, but which the hypothesis explicitly predicts should have no effect, and compare . A specificity margin S that is small relative to the run-to-run variance means the intervention is detecting generic sensitivity to perturbation somewhere in the network, not the mechanism under test — which is exactly the failure mode the subspace illusion exhibits. Reporting alone, without the matched control alongside it, is reporting half…
Read the full surrounding passage
The defence against this is a control condition run alongside the target one, not after it: intervene identically on a component set ' matched to the target set in size, depth and baseline activation statistics, but which the hypothesis explicitly predicts should have no effect, and compare . A specificity margin S that is small relative to the run-to-run variance means the intervention is detecting generic sensitivity to perturbation somewhere in the network, not the mechanism under test — which is exactly the failure mode the subspace illusion exhibits. Reporting alone, without the matched control alongside it, is reporting half an experiment.
Sources cited in the article section
- [3] Towards Best Practices of Activation Patching in Language Models: Metrics and Methods ↗
- [4] Causal Scrubbing: A Method for Rigorously Testing Interpretability Hypotheses ↗
- [5] Is This the Subspace You Are Looking for? An Interpretability Illusion for Subspace Activation Patching ↗
- [6] Towards Automated Circuit Discovery for Mechanistic Interpretability ↗
These citations give research context. Read each source to check which claims it supports.
Return to Running an Interpretability Investigation That Holds Up