Ethology of Latent Spaces

Type de record:

paper

Créateur:

Boi

Hôte:

This study challenges the presumed neutrality of latent spaces in vision language models (VLMs) by adopting an ethological perspective on their algorithmic behaviors. Rather than constituting spaces of homogeneous indeterminacy, latent spaces exhibit model-specific algorithmic sensitivities, understood as differential regimes of perceptual salience shaped by training data and architectural choices. Through a comparative analysis of three models (OpenAI CLIP, OpenCLIP LAION, SigLIP) applied to a corpus of 301 artworks (15th to 20th), we reveal substantial divergences in the attribution of political and cultural categories. Using bipolar semantic axes derived from vector analogies (Mikolov et al., 2013), we show that SigLIP classifies 59.4% of the artworks as politically engaged, compared to only 4% for OpenCLIP. African masks receive the highest political scores in SigLIP while remaining apolitical in OpenAI CLIP. On an aesthetic colonial axis, inter-model discrepancies reach 72.6 percentage points. We introduce three operational concepts: computational latent politicization, describing the emergence of political categories without intentional encoding; emergent bias, irreducible to statistical or normative bias and detectable only through contrastive analysis; and three algorithmic scopic regimes: entropic (LAION), institutional (OpenAI), and semiotic (SigLIP), which structure distinct modes of visibility. Drawing on Foucault's notion of the archive, Jameson's ideologeme, and Simondon's theory of individuation, we argue that training datasets function as quasi-archives whose discursive formations crystallize within latent space. This work contributes to a critical reassessment of the conditions under which VLMs are applied to digital art history and calls for methodologies that integrate learning architectures into any delegation of cultural interpretation to algorithmic agents. 23. pages, 14 figures, presented Hyperheritage International Symposium 9 ( paragraphe.univ-paris8.fr ) and accepted for publication in double-blind peer review in French in 2026-2027

Visit

arxiv.org

Similaires

mcgwilmo/Leveraging-Portuguese-Latent-Spaces-for-Kriolu-to-English-Translation-with-Monolingual-LM-PriorsCrafting Spaces of ValueComputational ethology for primate sociality: a novel paradigm for computer-vision-based analysis of animal behaviourLatent class multinomial logistic regression of predictors of latent class membership, Addis Ababa, Ethiopia,2023.Swahili in Spaces of WarBayesian inference of Latent Spectral Shapes

mcgwilmo/Leveraging-Portuguese-Latent-Spaces-for-Kriolu-to-English-Translation-with-Monolingual-LM-Priors

Experiments in low-resource Kriolu→English neural MT. We build on “Language Model Prior for Low-Reso

Crafting Spaces of Value

This ethnographic investigation of the rise of the artisanal oil refining industry in the Niger Delt

Computational ethology for primate sociality: a novel paradigm for computer-vision-based analysis of animal behaviour

Research in the biological and wildlife sciences is increasingly reliant on video data for measuring

Latent class multinomial logistic regression of predictors of latent class membership, Addis Ababa, Ethiopia,2023.

Latent class multinomial logistic regression of predictors of latent class membership, Addis Abab

Swahili in Spaces of War

Bayesian inference of Latent Spectral Shapes

This paper proposes a hierarchical spatial-temporal model for modelling the spectrograms of animal c