The effects of latent processes on memory storage lifetimes were investigated using the same two-state model that Fusi et al. (2005) develop in their paper. [1] See Methods for more details regarding simulation construction. An increased probability of potentiation events can be representative of several different underlying processes, such as a teaching input that coordinates activity across a population of neurons or a seizure that leads to synchronized spiking across brain regions. [2] In this simulation, the probability of potentiation events occurring was increased from 0.5 to 0.66, 0.75, and 0.95 over a short span of time steps (t = 20 to t = 50) in order to investigate its effect on memory storage lifetimes (Figure 1). Note the steeper decrease in memory signal with a potentiation probability of 0.95 compared to the baseline memory signal loss with a potentiation probability of 0.5. Also, note the graded effect of potentiation probability: a potentiation probability of 0.66 is practically indistinguishable from baseline, a potentiation probability of 0.75 starts to deviate from baseline, and a potentiation probability of 0.95 deviates noticeably from baseline. These results suggest that when there are latent processes that modulate the probability of potentiation events, there is a corresponding drop in memory signal lifetime due to the increased plasticity of the neural population.

The effects of asymmetric cascades on memory storage lifetimes were investigated using the same two-state model that Fusi et al. (2005) develop in their paper. [1] See Methods for more details regarding simulation construction. An asymmetry in the number of cascade states between the weak and the strong states may signal a difference in the possible metaplasticity processes that can enable a strong synapse to get stronger and a weak synapse to get weaker. For example, if a synapse can incrementally change its morphology to maximize synaptic surface area, but there's a minimum surface area that must be maintained in order for the neuron to survive, then there will be greater metaplasticity possible in the strong state than in the weak state. [2] In this simulation, the number of cascade states in the weak synaptic states was set to 5, 10, and 20 cascade states while the strong synaptic states were held constant at 10 cascade states in order to investigate the effect of metaplastic asymmetry on memory storage lifetimes (Figure 2). Note the steeper decrease in memory signal with a smaller number of weak cascade states (green) compared to the baseline memory signal loss with symmetric cascade states (blue). Although noisy, the memory signal does also appear to decrease slower when there are more weak cascade states (yellow) compared to the baseline memory loss with symmetric cascade states. These results suggest that metaplastic asymmetry may be an important condition that limits or enhances the memory storage capabilities of biological neurons compared to networks with symmetric cascade structures.