Scientists have uncovered a fundamental mechanism driving changes in how the brain stores memories over time. This phenomenon, known as representational drift, is a consequence of the brain’s continuous process of learning and memory consolidation. The research suggests that these changes are not errors but an inherent feature of how our neural networks adapt and store new information, even when the external stimuli remain constant.
The Brain’s Evolving Memory Landscape
The study, published in Nature Scientific Reports, explores the synaptic hypothesis of learning, which posits that memories are encoded in the physical structure of synaptic connections between neurons. While this allows for vast memory storage, it also means that new learning can overwrite or modify existing memories. Researchers observed that even when the same sensory input is presented, the resulting neural activity patterns can change significantly over days and weeks. This gradual alteration in neural codes, termed representational drift (RD), has been observed across various brain regions, including the hippocampus, visual, and auditory cortex.
Key Takeaways:
- Representational Drift is Natural: Changes in neural activity over time, even with stable inputs, are a normal consequence of ongoing memory storage.
- Synaptic Plasticity is Key: The brain’s ability to modify synaptic connections is the underlying mechanism driving this drift.
- Two Types of Drift: Drift can affect overall neuronal firing rates (rate effects) and the specificity of neuronal tuning (tuning effects), with different time scales.
- Experience Matters: The amount of time spent exploring an environment influences the rate of drift in spatial memory representations.
- Repetition Can Stabilize: Repeated exposure to familiar stimuli can reduce representational drift, as seen in olfactory processing.
Unraveling the Mechanism: Synaptic Turnover
Using computational models and fitting them to experimental data from mice, the researchers identified synaptic turnover – the continuous addition and removal of synaptic connections – as a primary driver of RD. This turnover is not random but is linked to the brain’s ongoing process of encoding new memories. The model demonstrated that this synaptic rewiring, consistent with the storage of new information, could quantitatively explain the observed drift patterns in neural activity.
Time vs. Experience in Memory
The research also sheds light on why different types of memories might exhibit different drift patterns. For instance, changes in spatial memory (like navigating a familiar route) are influenced by both the passage of time and the amount of time spent actively exploring the environment. In contrast, repeated exposure to familiar odors can actually reduce drift in olfactory cortex, suggesting that the brain actively reinforces stable representations when stimuli are consistently re-encountered.
Implications for Understanding the Brain
This work provides a unified framework for understanding representational drift as an unavoidable consequence of the brain’s dynamic learning processes. It suggests that these changes are not indicative of memory degradation but rather reflect the brain’s continuous adaptation and storage of new experiences. Further research aims to explore how the brain maintains stable behavior despite this underlying neural instability.