No benefit of subsampling in ensemble coding
- 1New York University
Abstract
Our visual system compensates for its limited information processing capacity by creating summary representations of stimulus ensembles. A longstanding debate concerns whether summary statistics are computed by integrating information across the entire ensemble or by sampling only a subset of its items. Subsampling is intuitively appealing because 1) integrating over a small subset presumably reduces processing resource expenditure and 2) observer models employing this strategy can effectively reproduce ensemble coding performance. Here, we systematically evaluate both aspects. We first show that the apparent success of subsampling models depends on a problematic assumption: that the sensory encoding noise of each item remains constant as set size increases. This assumption entails that total coding resources, quantified as Fisher information, grow unconstrained with set size. Under the realistic assumption that total coding resources are limited, such that the encoding precision for each item decreases with increasing set size, we demonstrate that an ideal observer integrating the entire ensemble achieves the same accuracy with a strictly smaller total resource budget for every set size. We further show that this resource-constrained full-integration model provides a more parsimonious, yet equally good, account of ensemble coding behavior across multiple existing datasets as well as newly collected data. Our findings call into question the necessity of the subsampling hypothesis, as restricting integration to a subset of ensemble items confers neither a theoretical nor an explanatory benefit over resource-rational full integration.
Related
- Comparing subsampling strategies for efficient pairwise analysis of large pathogen genomic and spatial datasets: An application to Mycobacterium tuberculosis transmission
- 36 Baseline systemic immune profiles predict prognosis and differential benefit from ipilimumab intensification in relapsed mesothelioma (Bio-IFCT-1501 MAPS2 trial)
- Informing an mpox stockpile investment case: epidemiological modelling and cost–benefit analysis
- Re-evaluating the benefit of pretraining scale in cell foundation models
- Examining hearing-aid benefit during story listening in older adults using large language models
BibTeX
@misc{ni2026benefit,
title = {{No benefit of subsampling in ensemble coding}},
author = {L. Ni and M. S. Landy},
howpublished = {bioRxiv},
year = {2026},
month = sep,
doi = {10.64898/2026.09.11.750959},
url = {https://www.biorxiv.org/content/10.64898/2026.09.11.750959v1},
note = {CC BY-ND},
}
RIS
TY - UNPB
TI - No benefit of subsampling in ensemble coding
AU - L. Ni
AU - M. S. Landy
PY - 2026
DA - 2026/09/11
PB - bioRxiv
DO - 10.64898/2026.09.11.750959
UR - https://www.biorxiv.org/content/10.64898/2026.09.11.750959v1
AB - Our visual system compensates for its limited information processing capacity by creating summary representations of stimulus ensembles. A longstanding debate concerns whether summary statistics are computed by integrating information across the entire ensemble or by sampling only a subset of its items. Subsampling is intuitively appealing because 1) integrating over a small subset presumably reduces processing resource expenditure and 2) observer models employing this strategy can effectively reproduce ensemble coding performance. Here, we systematically evaluate both aspects. We first show that the apparent success of subsampling models depends on a problematic assumption: that the sensory encoding noise of each item remains constant as set size increases. This assumption entails that total coding resources, quantified as Fisher information, grow unconstrained with set size. Under the realistic assumption that total coding resources are limited, such that the encoding precision for each item decreases with increasing set size, we demonstrate that an ideal observer integrating the entire ensemble achieves the same accuracy with a strictly smaller total resource budget for every set size. We further show that this resource-constrained full-integration model provides a more parsimonious, yet equally good, account of ensemble coding behavior across multiple existing datasets as well as newly collected data. Our findings call into question the necessity of the subsampling hypothesis, as restricting integration to a subset of ensemble items confers neither a theoretical nor an explanatory benefit over resource-rational full integration.
ER -