The Reuse-and-Append Memory Principle: Application to Latent Cause Inference
The latent cause theory of memory modification provides a computational account of how the brain decides whether to update existing memories or form new ones, but leaves unspecified the neural mechanisms implementing this inference. We propose Reuse-and-Append Memory (RAM), a set of mechanistic principles that achieve the same goal through sparse neural coding, Hebbian learning, and…
The Reuse-and-Append Memory (RAM) principle offers a computational framework to explain how the brain decides between updating existing memories or forming new ones. While the theory outlines the process, it doesn't delve into the specific neural mechanisms that enable this decision-making.
RAM leverages sparse neural coding, Hebbian learning, and pattern-matching dynamics to achieve the same goal. Essentially, neurons that have encoded previous experiences are automatically reactivated by similar stimuli. Meanwhile, uncommitted neurons are recruited to encode genuinely novel aspects of each experience, including the passage of time.
To validate this principle, a computational model was developed. This model was able to replicate key phenomena in fear conditioning, such as acquisition, extinction, renewal, and spontaneous recovery. The model suggests that the updating or formation of memories isn't achieved by modifying existing memories, but rather by appending new information.
The RAM principle separates what is typically viewed as a single prediction error driving memory updating or formation into two distinct signals. An immediate novelty signal emerges from the allocation of new neurons, based on coverage. A second signal, an outcome mismatch signal, activates safety circuits later when an expected outcome fails to materialize.
Interestingly, RAM also explains the dependence of recovery on the timing of stimulus reminders, a phenomenon known as the Monfils-Schiller effect. However, this effect is predicted to be inherently fragile, aligning with the mixed empirical findings on the topic. Essentially, the effect acts like a bridging event, carrying safety information to new contexts through offline co-retrieval.
Importantly, no phenomena explained by RAM require the modification of existing memories. Instead, the most stable configuration of the principle involves always appending, never overwriting. This suggests that the brain may inherently favor an approach that preserves original memories while integrating new information.
Written by urgent.news from bioRxiv's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
