# Dendritic heterosynaptic plasticity arises from calcium-based input learning

## TL;DR

This paper shows how calcium diffusion in dendrites can cause learning at non-stimulated synapses, adding a new mechanism for brain plasticity.

## Contribution

The study introduces a computational model linking calcium dynamics to heterosynaptic plasticity, explaining how spine timing and location affect learning.

## Key findings

- Calcium from a stimulated spine diffuses to nearby spines, causing heterosynaptic plasticity.
- Input timing and spine distance modulate synaptic changes via calcium diffusion.
- The model resolves experimental ambiguities and extends the Ca2+-hypothesis to non-stimulated synapses.

## Abstract

Stimulus-triggered synaptic plasticity is the foundation of learning and crucial cognitive abilities. Although numerous computational models have investigated plasticity within networks of point neurons, dendritic integration confers superior computational capacity compared to these simplistic models, highlighting the significance of dendrites and their spines–small, specialized protrusions that serve as loci for synaptic plasticity. Synaptic plasticity can be categorized into two forms: homosynaptic plasticity, involving changes at directly stimulated synapses, and heterosynaptic plasticity, involving changes at non-stimulated synapses. For homosynaptic plasticity, the Ca2+-hypothesis identifies the calcium concentration within a stimulated dendritic spine as the key mediator. In contrast, although theoretical studies attribute important roles such as synaptic competition and cooperation to heterosynaptic plasticity, experimental evidence remains ambiguous. By integrating insights from Ca2+-dependent homosynaptic plasticity with data on dendritic Ca2+-dynamics, we demonstrate that calcium influx into a stimulated spine can diffuse to neighboring spines, triggering heterosynaptic effects. To investigate this, we develop a mathematical model characterizing the temporal and spatial dynamics of calcium in dendrites in response to different inputs. Our model explains experimental ambiguities and extends the Ca2+-hypothesis to heterosynaptic plasticity. Notably, it predicts that input-timing, distance between spines, and local diffusion properties modulate synaptic changes, revealing a mechanism for dendritic computation.

Our computational model shows that calcium diffusion between spines via dendrites drives heterosynaptic plasticity, providing new learning mechanisms based on timing and spine location.