Saved in:
| Main Author: | |
|---|---|
| Format: | Recurso digital |
| Language: | |
| Published: |
Zenodo
2025
|
| Online Access: | https://doi.org/10.5281/zenodo.18061097 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866901493046575104 |
|---|---|
| author | Park, Jiho |
| author_facet | Park, Jiho |
| contents | <p dir="ltr">The functional role of dreaming has long resisted unified explanation, with existing accounts distributed across memory consolidation, emotional processing, and epiphenomenal theories.</p> <p dir="ltr"> </p> <p dir="ltr">Building on Erik Hoel’s Overfitted Brain Hypothesis (2021), we propose that dreaming serves as a critical computational regularization mechanism for a biological system engaged in continuous, online learning. During waking hours, the brain is exposed to limited and context-biased data, creating a structural risk of overfitting.</p> <p dir="ltr"> </p> <p dir="ltr">We argue that dreams mitigate this risk through a biologically implemented form of generative data augmentation, producing distorted sensory scenarios that expand the effective training distribution.</p> <p dir="ltr"> </p> <p dir="ltr">We formalize this process mathematically and demonstrate that the characteristic bizarreness of dreams is not a cognitive defect but a functional requirement for maintaining generalization and cognitive robustness.</p> <p> </p> |
| format | Recurso digital |
| id | zenodo_https___doi_org_10_5281_zenodo_18061097 |
| institution | Zenodo |
| language | |
| publishDate | 2025 |
| publisher | Zenodo |
| record_format | zenodo |
| spellingShingle | Dreaming as a Generative Regularization Mechanism in an Overfitted Brain Park, Jiho <p dir="ltr">The functional role of dreaming has long resisted unified explanation, with existing accounts distributed across memory consolidation, emotional processing, and epiphenomenal theories.</p> <p dir="ltr"> </p> <p dir="ltr">Building on Erik Hoel’s Overfitted Brain Hypothesis (2021), we propose that dreaming serves as a critical computational regularization mechanism for a biological system engaged in continuous, online learning. During waking hours, the brain is exposed to limited and context-biased data, creating a structural risk of overfitting.</p> <p dir="ltr"> </p> <p dir="ltr">We argue that dreams mitigate this risk through a biologically implemented form of generative data augmentation, producing distorted sensory scenarios that expand the effective training distribution.</p> <p dir="ltr"> </p> <p dir="ltr">We formalize this process mathematically and demonstrate that the characteristic bizarreness of dreams is not a cognitive defect but a functional requirement for maintaining generalization and cognitive robustness.</p> <p> </p> |
| title | Dreaming as a Generative Regularization Mechanism in an Overfitted Brain |
| url | https://doi.org/10.5281/zenodo.18061097 |