Abstract:Neural network models of memory are notorious for catastrophic interference: Old items are forgotten as new items are memorized (French, 1999; McCloskey & Cohen, 1989). While working memory (WM) in human adults shows severe capacity limitations, these capacity limitations do not reflect neural network style catastrophic interference. However, our ability to quickly apprehend the numerosity of small sets of objects (i.e., subitizing) does show catastrophic capacity limitations, and this subitizing capacity and WM might reflect a common capacity. Accordingly, computational investigations (Knops, Piazza, Sengupta, Eger & Melcher, 2014; Sengupta, Surampudi & Melcher, 2014) suggest that mutual inhibition among neurons can explain both kinds of capacity limitations as well as why our ability to estimate the numerosity of larger sets is limited according to a Weber ratio signature. Based on simulations with a saliency map-like network and mathematical proofs, we provide three results. First, mutual inhibition among neurons leads to catastrophic interference when items are presented simultaneously. The network can remember a limited number of items, but when more items are presented, the network forgets all of them. Second, if memory items are presented sequentially rather than simultaneously, the network remembers the most recent items rather than forgetting all of them. Hence, the tendency in WM tasks to sequentially attend even to simultaneously presented items might not only reflect attentional limitations, but also an adaptive strategy to avoid catastrophic interference. Third, the mean activation level in the network can be used to estimate the number of items in small sets, but it does not accurately reflect the number of items in larger sets. Rather, we suggest that the Weber ratio signature of large number discrimination emerges naturally from the interaction between the limited precision of a numeric estimation system and a multiplicative gain control mechanism.

A Precise Measure of Working Memory Reveals Subjects Difficulties Managing Limited Capacity

Precursors to Chunking Vanish as Working Memory Capacity Is Exceeded

First direct evidence of two stages in free recall and three corresponding estimates of working memory capacity

Individual Differences in Working Memory Capacity Are Reflected in Different ERP and EEG Patterns to Task Difficulty

Modelling Working Memory Capacity: Is the Magical Number Four, Seven, or Does it Depend on What You Are Counting?

Preliminary Evidence -- Diagnosed Alzheimer's Disease But Not MCI Affects Working Memory Capacity - 0.7 of 2.7 Memory Slots is Lost

Working memory limitations constrain visual episodic long-term memory at both specific and gist levels of representation

Estimating Working Memory Capacity for Lists of Nonverbal Sounds

Large capacity temporary visual memory.

Synaptic Correlates of Working Memory Capacity

The proportion of working memory items recoverable from long-term memory remains fixed despite adult aging.

Neural Measures Reveal Human Short-Term Working Memory–Varying Both Load and Cue Size

Short-term memory and working memory as indices of children's cognitive skills

Remembering Sets: Capacity Limit and Time Limit of Ensemble Representations in Working Memory.

Partial recall: Implications for the discrete slot limit of working memory capacity

An item response theory approach to the measurement of working memory capacity

Behavioral signatures of the rapid recruitment of long-term memory to overcome working memory capacity limits

When remembering less is more: Unfiltered items are associated with reduced memory fidelity in visual short-term memory

Working memory load affects intelligence test performance by reducing the strength of relational item bindings and impairing the filtering of irrelevant information

Retrieval-practice task affects relationship between working memory capacity and retrieval-induced forgetting

Sequential Presentation Protects Working Memory From Catastrophic Interference