Memorization of novel patterns in working memory in a model based on dendritic bistability
Working memory can hold many types of information and is crucial for cognition. Commonly, models of working memory maintain information such as hues or words by forming memory attractors through structured connectivities. However, real-world information can be novel, making it infeasible to use pre-trained attractors. In addition, most models-with or without attractors-have focused on maintaining binary categories instead of continuous information in each neuron. In the present study, we investigate how the brain might maintain working memory representations of arbitrary novel patterns with graded values. We propose an unstructured network model in which each neuron has multiple bistable dendrites. Each dendrite effectively implements fast Hebbian plasticity due to dendritic dynamics and dendrite-soma interactions. This network can maintain novel graded patterns under various perturbations without fine tuning of parameters. Through analytical characterization of network dynamics during the encoding and memory periods, we identify different conditions that yield either perfect memories or several types of memory errors. We also demonstrate memory robustness under various conditions and resilience to temporal inhibitory perturbations. Thus, this architecture provides robust and analytically tractable storage of novel graded patterns in working memory.