Attractor models of working memory and their modulation by reward

Abstract

This work reports an empirical examination of two key issues in theoretical neuroscience: distractibility in the context of working memory (WM) and its reward dependence. While these issues have been examined fruitfully in isolation (e.g. Macoveanu et al. in Biol Cybern 96(4): 407-19, 2007), we address them here in tandem, with a focus on how distractibility and reward interact. In particular, we parameterise an observation model that embodies the nonlinear form of such interactions, as described in a recent neuronal network model (Gruber et al. in J Comput Neurosci 20:153-166, 2006). We observe that memory for a target stimulus can be corrupted by distracters in the delay period. Interestingly, in contrast to our theoretical predictions, this corruption was only partial. Distracters do not simply overwrite target; rather, a compromise is reached between target and distracter. Finally, we observed a trend towards a reduced distractibility under conditions of high reward. We discuss the implications of these findings for theoretical formulations of basal and dopamine (DA)-modulated neural bump- attractor networks of working memory.

Fig. 1

Model architecture assumed by Gruber et al. (2006)

Fig. 2

The DD-curve under basal conditions

Fig. 3

Modulation of the DD-curve. The inner curve has contracted under high DA

Fig. 4

Dependence of the DD-curve on parameters

Fig. 5

Conditional model fit, as specified by the marginal MAP estimates of the model parameters, together with data. The inner curve describes the high-reward condition

Fig. 6

The marginal posterior over the slope parameter c. We see very strong evidence that the slope is greater than zero (i.e. there is a deviance-distracter effect). Interestingly we can be 89.2% certain that this slope is less than one; this is the value predicted by a winner-takes-all mechanism

Fig. 7

The marginal posterior over the reward modulation parameter b. We see moderate evidence (85.5% posterior probability) that behavioural responses to distraction were contracted into a smaller range in trials with high potential reward (1 pound as opposed to 0.01 pounds)