Abstract:Information from preceding trials of cognitive tasks can bias performance in the current trial, a phenomenon referred to as interference. Subjects performing visual working memory tasks exhibit interference in their trial-to-trial response correlations: the recalled target location in the current trial is biased in the direction of the target presented on the previous trial. We present modeling work that (a) develops a probabilistic inference model of this history-dependent bias, and (b) links our probabilistic model to computations of a recurrent network wherein short-term facilitation accounts for the dynamics of the observed bias. Network connectivity is reshaped dynamically during each trial, providing a mechanism for generating predictions from prior trial observations. 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 has response statistics whose mean is centered at the true target location across many trials, typical of such visual working memory tasks. Furthermore, we demonstrate task protocols for which the plastic model performs better than a model with static connectivity: repetitively presented targets are better retained in working memory than targets drawn from uncorrelated sequences.
What problem does this paper attempt to address?
The problem that this paper attempts to solve is how the memory of the previous trial interferes with the performance of the current trial in a working - memory task, namely the so - called "interference" phenomenon. Specifically, the research focuses on the fact that in a visual working - memory task, the recalled target position of subjects in the current trial will be affected by the target position of the previous trial, and this effect is manifested as a recall bias. Through the modeling method, the paper aims to explain the mechanism by which this bias is generated and explore how neurophysiological processes lead to this cross - trial bias.
### Main research questions:
1. **Evidence - accumulation strategy for bias generation**: Researchers attempt to understand which evidence - accumulation strategy can explain the bias introduced by the target of the previous trial. They show how these biases emerge when observers use sequential Bayesian updates to predict the location of the next target.
2. **Neurophysiological mechanisms**: Researchers explore which neurophysiological processes may lead to cross - trial biases. In particular, how short - term facilitation (STF) dynamically remodels network connections on a time scale of several seconds, thereby producing the dynamic characteristics of the bias.
### Research methods:
- **Probabilistic inference model**: By developing a probabilistic inference model, researchers predict the possible distribution of target angles in the current trial based on observations of past trials. When observers assume that the environment changes rapidly, the predicted distribution is mainly shaped by the target of the previous trial.
- **Recurrent network model**: Researchers analyze a recurrent network model with STF. In this model, a local activity peak represents the observer's belief in the current trial, and the spatial distribution of STF represents the observer's predicted distribution of subsequent targets.
### Key findings:
- **Dependence of bias**: The response error depends on the position of the target in the previous trial, and this bias decreases as the interval between the current trial and the previous trial increases, and increases as the delay period increases.
- **Role of STF**: STF can explain the dynamic characteristics of the bias because it enhances the synaptic strength of neurons related to the target of the previous trial on a time scale of several seconds, thus attracting the activity peak in the current trial to move towards the previous target position.
### Model validation:
- **Matching with experimental data**: The researchers' model can reproduce the main trends observed by Papadimitriou et al. (2015), including the variation rules of bias with the angle between targets, trial interval time, and delay period.
Through these models and analyses, the paper not only explains the generation mechanism of the interference phenomenon in visual working - memory tasks, but also provides a theoretical basis and experimental predictions for further research.