PT  - JOURNAL ARTICLE
AU  - Zachary P Kilpatrick
TI  - Synaptic mechanisms of interference in working memory
AID  - 10.1101/149435
DP  - 2017 Jan 01
TA  - bioRxiv
PG  - 149435
4099  - http://biorxiv.org/content/early/2017/06/13/149435.short
4100  - http://biorxiv.org/content/early/2017/06/13/149435.full
AB  - In serial cognitive tasks, information from preceding trials can bias performance in the current trial, a phenomenon referred to as interference. Recent experiments by Papadimitriou et al. (2015) demonstrated such biases in spatial working memory tasks, wherein subjects recalled the location of a target presented in continuous space. Analyzing response correlations in serial trials, they found the recalled location in the current trial is biased in the direction of the target presented on the previous trial. We build on their heuristic computational model to: (a) provide a Bayesian interpretation of the history-dependent bias; and (b) derive a mechanistic model demonstrating short-term facilitation accounts for the dynamics of the observed bias. Our computational model is a bump attractor network whose architecture is reshaped dynamically during each trial, linking changes to network connectivity with a predictive distribution based on observations of prior trials. Applying timescale separation methods, we can obtain a low-dimensional description of the trial-to-trial bias based on the history of target locations. The model still has response statistics whose mean is centered at the true target location in the limit of a large number of trials. Furthermore, we demonstrate task protocols for which the plastic model performs better than a model with static connectivity. Thus, our work presents a testable hypothesis for the persistence of interference in uncorrelated spatial working memory trials.