hypothesis

active

hypothesis:we-hypothesize-that-high-low-frequency-detectors-if-predicted-by-artificial-neural-network-universality-might-be-found-in-biological-neural-networks

We hypothesize that high-low frequency detectors, if predicted by artificial neural network universality, might be found in biological neural networks.

Specific cross-domain prediction mentioned by neuroscientists in conversation with the authors

Source paper

extracted_from

Zoom In: An Introduction to Circuits

(2020) · Chris Olah · Nick Cammarata · Ludwig Schubert · Gabriel Goh +2

Neighborhood — ranked by edge-count

Claims (2)

claim

Analogous features and circuits form across models and tasks.
associated_with
Third of three speculative claims asserting that learned features are not model-specific but represent universal solutions to learning problems
If the universality hypothesis is broadly true, it raises the exciting possibility that artificial neural networks could predict features previously unknown in biological neural networks.
supports
Speculative extension of universality to neuroscience, with high-low frequency detectors as a candidate prediction

Related by similarity (8)

cosine ≥ 0.65 · no typed edge

Entities in the same semantic neighborhood but without a typed relation to this one — candidates for new edges or unrecognized duplicates.

High-low frequency detectors found across AlexNet, InceptionV1, VGG19, and ResNetV2-50finding0.820
Second low-level feature type demonstrating cross-architecture universality
Neural networks show substantial alignment with biological representations in the brain, driven by shared task and data constraintsclaim0.785
Extends convergence argument to brain-machine alignment
Bioelectric networks discovered by evolution ~time of bacterial biofilms; served as ideal medium for scaling computation and information synthesis before neural systems.finding0.775
Evidence that pre-neural bioelectric infrastructure predates and likely precedes neurobiology; supports continuity of intelligence across substrates.
Superposition hypothesis: neural networks represent more features than dimensions using almost-orthogonal directions.hypothesis0.774
Explanation for why dictionary learning can recover many more features than dimensions.
High-Low Frequency Detectorconcept0.770
A less intuitive feature family detecting low-frequency patterns on one side of the receptive field and high-frequency on the other; used as example of non-obvious but understandable features
Biological networks exhibit the lowest EI among real networks and show the most significant causal emergence after coarse-graining.finding0.768
Finding from Klein & Hoel (2020) on real network analysis.
Linear representation hypothesis: neural networks represent meaningful concepts as directions in their activation spaces.hypothesis0.762
Foundation for interpreting features as linear directions.
Different neural network models trained on different objectives and modalities are converging to a shared statistical model of reality in their representation spaceshypothesis0.761
The central hypothesis of the paper; the platonic representation hypothesis itself