Comparing Theories for the Maintenance of Late LTP and Long-Term Memory: Computational Analysis of the Roles of Kinase Feedback Pathways and Synaptic Reactivation

Enhancing Associative Learning in Rats With a Computationally Designed Training Protocol

Background

Learning requires the activation of protein kinases with distinct temporal dynamics. In Aplysia, nonassociative learning can be enhanced by a computationally designed learning protocol with intertrial intervals (ITIs) that maximize the interaction between fast-activated PKA (protein kinase A) and slow-activated ERK (extracellular signal-regulated kinase). Whether a similar strategy can enhance associative learning in mammals is unknown.

Methods

We simulated 1000 training protocols with varying ITIs to predict an optimal protocol based on empirical data for PKA and ERK dynamics in rat hippocampus. Adult male rats received the optimal protocol or control protocols in auditory fear conditioning and fear extinction experiments. Immunohistochemistry was performed to evaluate pCREB (phosphorylated cAMP response element binding)\protein levels in brain regions that have been implicated in fear acquisition.

Results

Rats exposed to the optimal conditioning protocol with irregular ITIs exhibited impaired extinction memory acquisition within the session using a standard footshock intensity, and stronger fear memory retrieval and spontaneous recovery with a weaker footshock intensity, compared with rats that received massed or spaced conditioning protocols with fixed ITIs. Rats exposed to the optimal extinction protocol displayed improved extinction of contextual fear memory and reduced spontaneous recovery compared with rats that received standard extinction protocols. Moreover, the optimal conditioning protocol increased pCREB levels in the dentate gyrus of the dorsal hippocampus, suggesting enhanced induction of long-term potentiation.

Conclusions

These findings demonstrate that a computational model-driven behavioral intervention can enhance associative learning in mammals and may provide insight into strategies to improve cognition in humans.

Distinctive properties of biological neural networks and recent advances in bottom-up approaches toward a better biologically plausible neural network

Although it may appear infeasible and impractical, building artificial intelligence (AI) using a bottom-up approach based on the understanding of neuroscience is straightforward. The lack of a generalized governing principle for biological neural networks (BNNs) forces us to address this problem by converting piecemeal information on the diverse features of neurons, synapses, and neural circuits into AI. In this review, we described recent attempts to build a biologically plausible neural network by following neuroscientifically similar strategies of neural network optimization or by implanting the outcome of the optimization, such as the properties of single computational units and the characteristics of the network architecture. In addition, we proposed a formalism of the relationship between the set of objectives that neural networks attempt to achieve, and neural network classes categorized by how closely their architectural features resemble those of BNN. This formalism is expected to define the potential roles of top-down and bottom-up approaches for building a biologically plausible neural network and offer a map helping the navigation of the gap between neuroscience and AI engineering.

MMM - The molecular model of memory

Identifying mechanisms underlying neurons ability to process information including acquisition, storage, and retrieval plays an important role in the understanding of the different types of memory, pathogenesis of many neurological diseases affecting memory and therapeutic target discovery. However, the traditional understanding of the mechanisms of memory associated with the electrical signals having a unique combination of frequency and amplitude does not answer the question how the memories can survive for life-long periods of time, while exposed to synaptic noise. Recent evidence suggests that, apart from neuronal circuits, a diversity of the molecular memory (MM) carriers, are essential for memory performance. The molecular model of memory (MMM) is proposed, according to which each item of incoming information (the elementary memory item - eMI) is encoded by both circuitries, with the unique for a given MI electrical parameters, and also the MM carriers, unique by its molecular composition. While operating as the carriers of incoming information, the MMs, are functioning within the neuron plasma membrane. Inactive (latent) initially, during acquisition each of the eMIs is activated to become a virtual copy of some real fact or events bygone. This activation is accompanied by the considerable remodeling of the MM molecule associated with the resonance effect.