Latching dynamics as a basis for short-term recall.

Kwang Il Ryom,Oleksandra Soldatkina,Alessandro Treves,Vezha Boboeva,Peter E Latham

doi:10.1371/journal.pcbi.1008809

Kwang Il Ryom, Oleksandra Soldatkina + Show 3 more

Open Access

https://doi.org/10.1371/journal.pcbi.1008809

Copy DOI

Abstract

We discuss simple models for the transient storage in short-term memory of cortical patterns of activity, all based on the notion that their recall exploits the natural tendency of the cortex to hop from state to state-latching dynamics. We show that in one such model, and in simple spatial memory tasks we have given to human subjects, short-term memory can be limited to similar low capacity by interference effects, in tasks terminated by errors, and can exhibit similar sublinear scaling, when errors are overlooked. The same mechanism can drive serial recall if combined with weak order-encoding plasticity. Finally, even when storing randomly correlated patterns of activity the network demonstrates correlation-driven latching waves, which are reflected at the outer extremes of pattern space.

Highlights

Despite much effort directed towards understanding the neural processes underlying shortterm memory (STM), what causes its notoriously limited capacity has, to this day, remained largely mysterious [1,2,3,4,5]
If one were to take a functionalist perspective, inspired e.g. by Baddeley’s theory of working memory [6], and assume that items in short-term memory are transiently represented in a dedicated cortical module, where they have been copied from their long-term traces, two riddles would arise: how would the copying work? and why would this module have such poor capacity? Multiple lines of evidence, since the advent of functional imaging, have failed to identify an ad hoc STM module, and indicated that STM is expressed by the activity of the same neurons that participate in the representation of long-term memories (LTM) [7]
While the basic model needs more structure to be predictive about specific behaviour, e.g. in semantic priming experiments [13], or about the effects of item complexity [14] or individual differences [15], and in general to fully benchmark its validity as a model of short-term memory [4], we show that it is consistent with simple experiments, that illustrate the way STM limitations depend on task demands

Summary

Introduction

Despite much effort directed towards understanding the neural processes underlying shortterm memory (STM), what causes its notoriously limited capacity has, to this day, remained largely mysterious [1,2,3,4,5]. Since the advent of functional imaging, have failed to identify an ad hoc STM module, and indicated that STM is expressed by the activity of the same neurons that participate in the representation of long-term memories (LTM) [7] This disposes of the copy riddle, but emphasizes the capacity one. A Potts network can model the long-range interactions among patches of cortex and, without any ad hoc component, shows a tendency to hop spontaneously from activity pattern to activity pattern, recalling them in a sequence resembling a random walk We call this latching dynamics and propose here that it holds the key to understand STM limitations, once combined with some mechanism, perforce imprecise, for short-term storage. We consider a number of distinct mechanisms of this type, that by adding an extra “kick” to boost a small subset of L among p patterns in long-term memory, approximately restrict latching dynamics to the subset, which is effectively kept in short-term memory

Methods

Results

Discussion

Conclusion