bioRxiv Science⌕ Search

bioRxiv · 10.64898/2026.10.19.755340

Infantile amnesia results from differential memory retrieval in humans

Abstract

Infantile amnesia, the inability to recall memories from our earliest years, is the most common form of forgetting in mammals. Whether early memories never form, decay, or persist but become inaccessible remains unknown. Here, we investigated the nature of forgetting during the infantile amnesia period by longitudinally testing toddlers aged 18-24 months across the complete memory cycle from encoding, initial retrieval one day later, and delayed retrieval up to six months later, in a novel naturalistic episodic memory paradigm. Computational modelling revealed that forgetting during the infantile amnesia period was heterogeneous. Toddlers were as likely to retain the full memory as to lose it completely, while nearly a quarter retained only part of the memory. Encoding accuracy predicted which toddlers retained the full memory at the test one day later, but not at the delayed test up to six months on. Instead, a retrieval-specific EEG signature predicted complete, but not partial, memory retrieval, distinguishing which toddlers had retained the full memory before their behavioural response. Finally, leveraging the COVID-19 pandemic as a natural experiment, we found that maternal immune activation during pregnancy, as in rodents, alters vulnerability to infantile forgetting in humans. Together, these findings show that early memories formed during the infantile amnesia period are not simply erased but persist in an increasingly inaccessible form, a transformation whose trajectory can be shaped by both the neural retrieval mechanism and the prenatal environment.

Explore related subjects

Keep this discovery

Explore connections, maps & timelines

BibTeXRIS

Power, S. D., Werwach, A., Peikert, A., Fey, M., Forbes-Mitchell, T., Hild, J., Buss, C., Lindenberger, U., Ryan, T. J., Werkle-Bergner, M.. 2026-10-20. Infantile amnesia results from differential memory retrieval in humans. https://doi.org/10.64898/2026.10.19.755340

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related preprints

Human iEEG recordings during audiovisual speech violations support and extend predictive routing models

Rather than constantly reconstructing the world solely through sensory input, predictive processing theories posit that the brain utilizes predictions maintained as internal representations to guide sensory processing. An emerging model, predictive routing, proposes that an interplay between neuronal gamma (40-150 Hz) and alpha-beta (10-30 Hz) rhythms implements a dynamic mechanism for predictive processing. Namely, predictions enhance alpha-beta oscillations that suppress gamma activity during the processing of predicted inputs. While this model is supported by animal studies, its validity has not been fully explored within the human brain. The present study addresses this by leveraging human intracranial EEG (iEEG) recordings across superior temporal and prefrontal/premotor areas of seven presurgical epilepsy patients during an audiovisual speech paradigm. We calculated iEEG power in the theta, alpha-beta, and high gamma bands during simple auditory deviance, multisensory integration, and a condition where an incongruent visual cue (lip movements) clashed with an auditory stimulus (a syllable), creating audiovisual incongruence that can elicit the McGurk illusion. Results revealed increased high-gamma and decreased alpha-beta power in the superior temporal cortex in response to auditory deviance and the McGurk condition, while frontal areas, despite reflecting a similar decrease in alpha-beta power, did not present this high gamma prediction error signal. Instead, the theta band revealed a strong prediction error signal in the McGurk condition, consistent with theories on the role of theta in reorganizing cortical networks. These results both support and extend existing predictive routing models for human-specific forms of predictive processing.

neuroscience↗

A Unified World Model in Human Posterior Parietal Cortex

How does the brain understand the physical world, predict future events, and plan actions? Are these problems solved by a unified world model, as hypothesized by cognitive scientists and artificial intelligence researchers? Posterior parietal cortex has been implicated in all these functions, but whether the same neurons support each of them is unknown. We recorded from the superior parietal lobule (SPL) of two tetraplegic participants implanted with microelectrode arrays. SPL units preferentially engaged by intuitive physics, judging block-tower stability and viewing naturalistic physical events, also predicted whether moving objects would collide hundreds of milliseconds before contact, generalizing between different physical scenarios. These populations also encoded which hand action the participant intended to perform. These findings identify SPL as a substrate for physical world modeling.

neuroscience↗

Cortical long-range dynamics track excitation-inhibition balance across cognitive states and organizational principles

A fundamental question in systems neuroscience is how the brain's excitation-inhibition (E/I) balance shapes the temporal structure of neural activity and whether we can non-invasively recover this relationship from neural scaling properties. We show that MRS-derived E/I balance is expressed in long-range temporal dynamics and constrained by cytoarchitecture. The laminar organisation of inhibitory circuits, spanning agranular to granular cortex, determines how faithfully regional dynamics reflect E/I balance. This neurochemical signature is state-dependent: its relationship with cortical dynamics reverses during task engagement, consistent with traversal of dynamical regimes around a critical point. It extends beyond fMRI to arousal-linked dynamics, indexed by pupil diameter and microsaccades, indicating a global mode of cortical state rather than a regionally isolated property. The imprint of E/I balance on long-range dynamics follows the sensory-association axis at rest, and is reconfigured during task engagement. These findings establish E/I balance as a neurochemical quantity with structured, scalable signatures across the temporal and spatial organization of the human cortex.

neuroscience↗