More effort, more variability, but greater relative precision: Distributional modeling of effort-performance coupling
Abstract
Ratings of perceived exertion (RPE) are widely used to prescribe exercise intensities. When used this way, researchers and practitioners may implicitly or explicitly rely on assumptions about the coupling between perceived effort and performance. These assumptions refer to two dimensions: the average shape of how effort translates to performance and the variability of performance across effort levels. Straightforwardly, effort-performance coupling can be thought of as linear, with constant variability across the intensity spectrum. However, classic psychophysical scaling laws (e.g., Fechner’s law) and motor control research (e.g., signal-dependent noise) predict that such coupling is non-linear, with intensity-contingent increases in variability. These assumptions are rarely made explicit and have not been jointly tested. To address this, we fitted and compared theory-driven Bayesian multilevel models reflecting distinct assumptions about the mean and the variability structure, using leave-one-out cross-validation for model selection. Data were collected from 101 climbers performing an isometric handgrip task with effort prescribed using a RPE 1-10 scale. We found robust evidence for a heteroscedastic variability structure, with a systematic increase in absolute force variability with prescribed RPE. Additionally, we found modest evidence favoring an exponential mean shape over a linear mean shape, suggesting that force increases disproportionately with effort. These effects were gender unspecific and are broadly consistent with Fechner’s psychophysical law and signal-dependent noise theory. Together, these findings suggest that a direct translation from effort to performance systematically misrepresents the actual relationship. Instead, practitioners and researchers should expect a disproportionate increase in performance with increasing absolute performance variability.
Related
- Sub-millisecond precision and variability in a cricket calling song pattern
- Interindividual Variability in Functional Food Clinical Trials: Statistical Challenges, Responder Phenotypes, and Implications for Precision Nutrition
- Multiscale predictive cellular modeling: integrating hypothesis grammars, digital twins, and multi-omics for In silico oncology and precision theranostics
- Variability in Clinical Performance of the FDA-Cleared Lumipulse P-Tau217/β-Amyloid 1-42 Plasma Ratio
- Seasonal rainfall impacts the relative performance of maize-legume intercropping systems in semi-arid Zimbabwe
BibTeX
@misc{hack2026more,
title = {{More effort, more variability, but greater relative precision: Distributional modeling of effort-performance coupling}},
author = {Lukas Hack and Katja Rewitz and Silas Wagner and Wanja Wolff},
howpublished = {PsyArXiv},
year = {2026},
month = sep,
doi = {10.31234/osf.io/34pjh_v2},
url = {https://osf.io/preprints/psyarxiv/34pjh_v2/},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - More effort, more variability, but greater relative precision: Distributional modeling of effort-performance coupling
AU - Lukas Hack
AU - Katja Rewitz
AU - Silas Wagner
AU - Wanja Wolff
PY - 2026
DA - 2026/09/25
PB - PsyArXiv
DO - 10.31234/osf.io/34pjh_v2
UR - https://osf.io/preprints/psyarxiv/34pjh_v2/
AB - Ratings of perceived exertion (RPE) are widely used to prescribe exercise intensities. When used this way, researchers and practitioners may implicitly or explicitly rely on assumptions about the coupling between perceived effort and performance. These assumptions refer to two dimensions: the average shape of how effort translates to performance and the variability of performance across effort levels. Straightforwardly, effort-performance coupling can be thought of as linear, with constant variability across the intensity spectrum. However, classic psychophysical scaling laws (e.g., Fechner’s law) and motor control research (e.g., signal-dependent noise) predict that such coupling is non-linear, with intensity-contingent increases in variability. These assumptions are rarely made explicit and have not been jointly tested. To address this, we fitted and compared theory-driven Bayesian multilevel models reflecting distinct assumptions about the mean and the variability structure, using leave-one-out cross-validation for model selection. Data were collected from 101 climbers performing an isometric handgrip task with effort prescribed using a RPE 1-10 scale. We found robust evidence for a heteroscedastic variability structure, with a systematic increase in absolute force variability with prescribed RPE. Additionally, we found modest evidence favoring an exponential mean shape over a linear mean shape, suggesting that force increases disproportionately with effort. These effects were gender unspecific and are broadly consistent with Fechner’s psychophysical law and signal-dependent noise theory. Together, these findings suggest that a direct translation from effort to performance systematically misrepresents the actual relationship. Instead, practitioners and researchers should expect a disproportionate increase in performance with increasing absolute performance variability.
ER -