---
source: https://iinventedlight.com/library/notes/ai-circuits-features-connections.md
title: "Why study connections as well as individual features?"
license: CC-BY-4.0
license_url: https://creativecommons.org/licenses/by/4.0/
source_license: CC-BY-4.0
attribution: "Adapted summaries of the named authors below; wording and synthesis by Librarian."
author: Librarian
date: 2026-11-05
status: published
checked_on: 2026-11-05
as_of: 2026-11-05
confidence: high
untested: "Literature summary; no model experiments run."
category: coverage-gaps
topic_area: "ai-research: evaluation and interpretability"
tags: [ai-research, research-methods]
sources:
  - id: S1
    title: "Zoom In: An Introduction to Circuits"
    publisher: "Distill"
    url: https://distill.pub/2020/circuits/zoom-in/
    accessed: 2026-11-05
    license: CC-BY-4.0
    attribution: "Olah, Chris; Cammarata, Nick; Schubert, Ludwig; Goh, Gabriel; Petrov, Michael; Carter, Shan"
  - id: S2
    title: "The Building Blocks of Interpretability"
    publisher: "Distill"
    url: https://distill.pub/2018/building-blocks/
    accessed: 2026-11-05
    license: CC-BY-4.0
    attribution: "Olah, Chris; Satyanarayan, Arvind; Johnson, Ian; Carter, Shan; Schubert, Ludwig; Ye, Katherine; Mordvintsev, Alexander"
---
# Why study connections as well as individual features?

The Circuits introduction studies connections between neurons to find algorithms in neural-network weights. [S1]

## Key distinctions

Its examples include a dog head assembled from eyes, snout, fur, and tongue. [S1]

The Circuits project has only begun understanding a single vision model. [S1]

Detecting features differs from assembling individual pieces into later network decisions. [S2]

## Research scope

The Circuits introduction presents speculative claims about features, circuits, and universality. [S1]

Future articles will explore this research in detail. [S1]

## Claims

- id: C1
  claim: "The Circuits introduction studies connections between neurons to find algorithms in neural-network weights."
  source: S1
  evidence: "By studying the connections between neurons, we can find meaningful algorithms"

- id: C2
  claim: "Its examples include a dog head assembled from eyes, snout, fur, and tongue."
  source: S1
  evidence: "You can see a dog head be assembled from eyes, snout, fur and tongue."

- id: C3
  claim: "The Circuits project has only begun understanding a single vision model."
  source: S1
  evidence: "we’ve only scratched the surface of understanding a single vision model"

- id: C4
  claim: "Detecting features differs from assembling individual pieces into later network decisions."
  source: S2
  evidence: "how the network assembles these individual pieces to arrive at later decisions"

## Sources

- [S1]

Zoom In: An Introduction to Circuits. Distill. https://distill.pub/2020/circuits/zoom-in/ CC BY 4.0; accessed 2026-11-05.
- [S2]

The Building Blocks of Interpretability. Distill. https://distill.pub/2018/building-blocks/ CC BY 4.0; accessed 2026-11-05.

Attribution: source authors named above; adapted summaries, not endorsed by the authors.