Theory of neural coding predicts an upper bound on estimates of memory variability.

Abstract

Observers reproducing elementary visual features from memory after a short delay produce errors consistent with the encoding-decoding properties of neural populations. While inspired by electrophysiological observations of sensory neurons in cortex, the population coding account of these errors is based on a mathematical idealization of neural response functions that abstracts away most of the heterogeneity and complexity of real neuronal populations. Here we examine a more physiologically grounded model based on the tuning of a large set of neurons recorded in macaque V1, and show that key predictions of the idealized model are preserved.Both models predict long-tailed distributions of error when memory resources are taxed, as observed empirically in behavioral experiments and commonly approximated with a mixture of normal and uniform error components. Specifically, for an idealized homogeneous neural population, the width of the fitted normal distribution cannot exceed the average tuning width of the component neurons, and this also holds to a good approximation for more biologically realistic populations. Examining eight published studies of orientation recall, we find a consistent pattern of results suggestive of a median tuning width of approximately 20 degrees, which compares well with neurophysiological observations. The finding that estimates of variability obtained by the normal-plus-uniform mixture method are bounded from above leads us to reevaluate previous studies that interpreted a saturation in width of the normal component as evidence for fundamental limits on the precision of perception, working memory and long-term memory.

Stochastic sampling provides a unifying account of visual working memory limits

“…We have shown that adapting this model to achieve a higher degree of biophysical realism-by introducing heterogeneity in neural tuning curves and correlated spiking activity-improved the quality of fit to behavioral data. It has recently been shown that more neurally realistic population coding models preserve the key characteristics of the idealized model, and that signatures of neural tuning may even be visible in behavioral data (43).…”

Object-based selection in visual working memory

“…There is a limited number of trials for the two invalid conditions, which can make the modelling results less reliable. Furthermore, recent papers have questioned whether the lapse rate is representative of a lack of information when responding (Schurgin et al, 2020;Taylor & Bays, 2020).…”

Perceptual similarity judgments do not predict the distribution of errors in working memory

“…The first concerns the noise distribution, with TCC using a Gaussian and Neural Resource a Poisson process to describe the system’s stochasticity. While Poisson noise provides a more accurate approximation to neuronal population activity (e.g., Softky & Koch, 1993), it can also be closely approximated by Gaussian noise with an appropriate scaling of variability, and a variant of the Neural Resource model using this approximation has been shown previously to provide a similar account of WM recall errors (Schneegans & Bays, 2017; Taylor & Bays, 2020). Secondly, the two models use different decoding rules, with TCC using the MAX rule (i.e., a “winner-takes-all” decision rule) and Neural Resource using the maximum likelihood decoder.…”